Transcripts
1. Welcome to the Midjourney Masterclass: Every powerful image
starts with an idea, but turning that idea into a compelling visual
traditionally took some expensive software, some skills, and a lot
of trial and error. Midjourney changes
that completely. With the right prompt technique, you can create various
characters, environments, different types of images, edit them, modify them, and turn that idea into a great visual in just
a matter of minutes. Welcome to the complete
Midjourney creative course. I'm Hose kachii a
digital designer and creative instructor
at Skillademia. I have worked with
various designs, visual storytelling, and
creative tools for many years, and I've had the privilege
of teaching over 200,000 students worldwide
through my online courses. In this course, I'll guide
you through Midjourney, starting from the basics, moving on to the
advanced prompting and some creative projects. First going to start
by understanding how Midjourney works and what
makes a good prompt. Once we know how prompts work, we can apply them into generating various
types of images. We're going to be learning
how to do consistent designs, work with various references, and try different structures. Once we have our image
generated by Midjourney, we're going to learn
how to refine them. That's going to be perhaps
changing the style. Adding more elements on there, removing some
elements, and just get closer to that final
intentional design. But most importantly, this is a hands on project based course. So we're going to be
learning together how to make a dream
interior space, a fantasy character and poster, and some additional things. So by the end of this, you'll
not only know the basics, but you've applied it
to actual projects that you get to showcase
on your portfolio. For this class,
you're only going to need a Midjourney account, Internet connection,
some curiosity, and a will to experiment. Do not need any previous
AI or design experience because I'll be walking you
through that step by step. If you're ready to turn
your ideas into images, then grab your computer
and let's get started.
2. What is Midjourney : So let's start at the very beginning by
answering the question, what is Midjourney and
what can it even do? Midjourney is an AIR generator, which means that it's
a tool that takes your text description and
turns them into images. You can think of it as
your creative partner that helps you visualize
your ideas instantly. Instead of sketching
your ideas for hours, maybe you're thinking of merging a jellyfish into space
with an astronaut. Instead of trying to figure
out how to sketch it, you can use a tool
like Midjourney to get that idea and turn
it into a reality. All you do is you
type a sentence and then in under a minute or so it gives you an artwork that you can look
at and then further modify. As you can see already
on the screen, the artworks are pretty good and they're all
generated with AI. Already you can see what
capabilities Midjourney has, but let's dive deeper into what it can actually do for you. First of all, it can create
concept art for movies, games, design projects,
and many more. If you're someone who has to visualize a lot of things
in different fields, you can use a tool like this to make your workflow a lot easier. It can also generate
product mockups, illustrations, posters,
animations and other things. You can see on the top left, we have a lipstick swatch and that's great if
you want to showcase a certain brand onto
a model in a really quick and then it can help you explore different
aesthetic styles. Let's say you have
to come up with the animation style
for a short film, but you're not sure if you
want to do it in three D, two D, sketch art, or some other art style. You put your prompt
in Midjourney and you ask it to give you
those different styles, and then from there, you
can decide which one works best with the
aesthetic you're going for. Then lastly, it can speed up your brain
storming sessions. Let's say if you're
trying to design a logo, but you don't know what
to put as a mascot, you can ask me journey
to compact a bunch of elements and give you
inspirations for your logo. Let's say you want to combine
a cat and an octopus. Trying to figure out how to best combine them is
going to be hard. So just let AI do
the work for you. Then since it's all done by AI, you can do different variations from that one sketch
that you like, and that could be ten
different variations, 20. Just by changing the
prompt a little bit, you can get a whole
different style, it's just a great tool to explore and sometimes
just have fun with it. Now, one of the most
powerful aspects of Midjourney is the ability for it to mix and match your ideas. That means that rather than combining concepts that
go really well together, for example, makeup and fashion, you can combine things that have no relation to one another. Like over here, we're seeing this sandwich with salmon
and then construction. To be able to come up with an idea on where the
people are going, what is the sandwich
even look like? What should be on top? That's going to take a while. But we can see how this has
been executed so great. Another example is if you wanted to take ancient
Roman architecture, but have it made of neon class. That's a really bizarre concept, but you can use Midjourney
to bring it to life. These unique mash
ups are often where all the magic within
Midjourney happens. Now that we know what
Midjourney is, of course, I will dive deeper into how you get to have images
and videos like this. But I want you guys
to start thinking of some things that are
impossible to come up with something really unusual and two unusual things and
then combine them into one. Now, we know what Midjourney is and what you can do with it. In the further further chapters, I'm going to go over
exactly how you can prompt and get amazing
works like these. I will walk you through
all the steps and we also have projects
within this course, so you get to apply your
knowledge into actually building different
generations with Midjourney. But now I got a little mini
exercise for you guys. I want you to think of impossible or unusual
concepts, jot them down, and then in the further
lessons when we're experimenting with generating
things in Midjourney, you can try those out yourself. I'll give you a
couple of examples. We can do maybe a jellyfish
floating through a desert, maybe a castle made of ice cream with
sprinkles raining down. These are just some really
cool ideas that you can make within seconds
using this tool. Now let's move on to
better understanding how AI creates artworks. There's a lot to it
and it's a lot more complicated than most people
think because on our end, we just type in something
and we get a result. But a lot goes on in the background that
we are not aware of. But it's important for us to
know a little bit about how it all works. I
3. Pricing and Account: To get started with Midjourney, just head over to
midjourney.com slash HOME. You should see this page where
you get to either sign up, login, or explore Midjourney
before you make an account. So if you go over here, it's going to take you
to that same home page, but you can see we're not
signed in at the moment. This lets you understand what kind of artworks
you get to create with Midjourney and see if
this is the right fit for you before actually
committing with an account. You get to explore, but
we're not able to create. Going back, we can sign up
with either Google or DISC now Midjourney started as a bot within Discord and now they
have their own platform. The cool thing is that they
still have that Discord chat feature within their website and I'll show you that
in further lessons. But depending on whichever account you're comfortable with, just click on either, and you're going to get
redirected to the homepage. Can also log in using
the same options. Once you've logged in,
you will be brought to the same Explore page except now you get to see my account, and I'm also able to
create with M tourney. You go down here
next to my account, there's three little dots. The first is account settings
where it will tell you your Midjourney ID and
personal information, which you can either
look at or edit. There's also managed
subscription, manager uploads, which are the things you
upload from your local device. You can go to DiSCord that
same bot that I mentioned, and there's also a
Midjourney magazine. And then of course,
we can log out. Let's go to the
different subscriptions. So I'm just scrolling down. If you have an account,
a subscription already, that's going to be
a little bit above, so I scrolled down since that will include your
personal information. There are four plans that you
get to get with Midjourney, starting with the basic plan
and then the makeup plan. So the difference here is the
amount of generations you get to make and some of the tools are not available
for the basic plan. Previously, Midjourney
allowed you to edit or just create for free. I think around 200
credits were given to you and then you get to
upgrade into one of these. But now they don't have
that anymore and you need a basic plan minimum to be able to create
within this platform. You can see the
different generations. We get 200 images a month. These are all unlimited. We have video, but in SD, and then it goes
to HD over here. We have general
commercial terms. These are the things
we talked about in Chapter one in the
ethics lesson. You can read them
on a separate page. You can just do a Google search. There's optional credit top ups, same thing for all
the other plants, three concurrent
fast image jobs. We have that and then get
tied with the last two plans. Basically, it means three things using their fastest
model at the same time. With video, there's
only one here, but then it goes up to 12. There's use editor or just use
Editor on uploaded images. If you upload a selfie
of yourself and you want to add in a space
machine in the background, you get to do that
with the basic plan. But here we get a few additions, we get relaxed
image generations. I will talk about their
differences later. We have the same
options as this. We have stealth mode. This is another
feature that I'll talk about in the makeup plan, which is their biggest plan, you are able to create faster generations and
more concurrent jobs. They have a monthly billing
plan yearly billing plan, which saves you 20%. These are the prices
or you can just do monthly to experiment
before moving on. Now, choosing a plan, you have to think
about your goals. If you're just experimenting
or learning this platform, then the basic plan
might be enough. But if you want to
create the content to sell or distribute on a
more professional scale, you're going to probably
need the standard or the pro plan because those give you the extra
generation credits. With everything you
create within Midjourney, it's going to take up a
certain number of credits, and if you want to
do a lot of that, then you're going to
need enough credits. For this course, I'm using the
P plan because it includes the stealth mode and more
generations for me to show you. You can either go
with a pro plan, but the standard plan
should be fine as well. There's also some
frequently asked questions which describe the
things that we just saw. And here's some answers
regarding copyright. So you can just click on this to learn about the full details. You can also cancel
your subscription plans if you change your mind. It does tell you when
the next cycle is, so just make sure to mark
that in your calendar. You can also delete your
account completely, and then we have
options where you get to upgrade or
downgrade from your plan. If you want to use
Midjourney on Discord, you just make your
Discord account and open the bot
within that platform. This is the chat that I
was telling you about, so you're going to find the same structure
within Discord. We also have the
voice chat here. Now let's move on
to our next lesson where we explore the
interface a little bit. So what are all
these buttons for? How do you navigate
through the dashboard? And what's the best
practice for doing that?
4. Exploring the Interface: Let us get comfortable with the interface of Midjourney
and their dashboard. So when you first
join the server, it may feel like a lot. There's all these moving images and there's all these
tabs on the side, but you only really need to
focus on a few key areas. And that is the bar up
here where you get to put in your prompt and
get those generations. The create bar where you get to look at the variations
made for you, the edit bar where you get
to edit things further. Organize, which keeps
all of your work, and you explore for
some brainstorming, which is what we
are in right now. Starting from here,
we're able to add images if you
want a reference. For example, we have a character that we want in outer space. We can upload an image of
that character and then have Midjourney blend that character into an outer space image. When you click on that,
we have starting frame, image prompt, style references,
and omni references. The difference here
is that each one focuses on a certain aspect
of the image you upload. For example, the image prompt looks at the element
of the image. Is how many characters
are in there? Is there any clouds? Is
there any mountains? There's also the style, which looks at the style. Is it black and white, sketched, three D, two D. There's
also Omni reference, which is the characters form, what they look like,
what are the colors, their features, et cetera. And then starting frame is when you get to
animate an image. And that's what
we're on right now, all these moving things are basically images
animated by Midjourney. If you made something and
you want to animate it, you just upload that image up here and it will
generate it for you. Here you get to click
and drag your work. This is where you
type in your prompt. Then we have all these
parameters which define how strongly Midjourney
looks at your prompt. I will get into this in
another lesson because this is the second most important
part of Midjourney. The first is your prompt. A search bar over
here if you're doing any brainstorming and you
want to look at other works. We have personalize,
which lets you filter through all of
these and only keep the things that appeal
to you the most. For example, you are a product
designer and you would only like to see products and not so much
these random things. Midjourney lets you do that. We can filter through things. There is for you, which takes
us to this personalization, which I'll get into later. There is random,
random artworks. There is hot artworks, top day. Top week, even top month. We can also go through
our liked images. When you see an image you like, you can actually heard it and it will go in your like page. And also unlike it to remove it. We can look at the
prompt of an image. Let's say this is a
really good thing and we want to see
how they made it, simply click on it once and we can see that
it's really short. It tells us who the person's
name is when they made it, and then the prompt, which we can use it if
we click on it once, which model they use version. Version seven is
the latest version. We can search similar things
with that search icon and then we're able to copy either
the prompt job ID image, the URL to the image,
and then download it. You can also report things on the journey if there's
some sort of problem. We can look at the prompt, which when I click it, it goes straight into our box. I can have it input it as a style or input it as an image. This is how we get to use what we see into our own search bar. Delete that, you can also clear things using
that trash icon. In the Explore tab, we're also able to switch
between images and videos. We can randomize
the selection if you're brainstorming or you just want to come
up with a concept. This will be helpful for you. The next thing is to create tab. Whenever you type in a prompt, you should come here to
see what the results are. It shows you the
prompt on the right. Then you have the
variations on the left. We have some options to give us another set of
subtle variations, strong variations
or even animate. Here we have the same bar. We have draft mode. We have the conversational
mode where you get to talk to Midjourney regarding maybe the thing that
you're looking for. We can start the voice mode so I could voice prompt Midjourney
to create the stuff. There's a search bar and
then there's folders. I could make a new folder, put all my cats in one. Then edit is regarding an image that you made that
you want to further modify. So this is a very detailed
dashboard within Midjourney. There's all these tools
that you get to use, and basically you get to edit a small part of
the image, half of it, you can restore some of the generations,
combine, merge, resize, re texture, even, and it's a really cool interface that we're going to take
a look at later. We have personalize,
which lets you, as we said, but you only see the things
that you want to see. Right now it's locked, but you can unlock it by
creating a global profile. We'll take a look
at that later too. I do suggest turning this on if you want to brainstorm a lot. That way you will
only see things related to that concept
you're trying to make. We also have moodboard
which are related to the previous panel, Let's say you were
brainstorming and you found a bunch of
cool product images, you can combine them
all in a moodboard to help you better organize
those appealing images. Organize is where all of your works go in unless
you delete them. Anything you generate,
whether you like it or not, is going to end up here. It's selected
separated with dates. We have today a while ago. You can select them
all to download. You can them on hide, publish them to the
community publish or even add to a folder to make
things a little easier. You can click awing. If
you go on any of them, it's going to show you the same tools we looked at
the prompt, the version, and then you have some
creation actions, which allows you to
create more variations, upscale it into a
larger version, rerun the same prompt, enhance it for full quality, and then use the editor
tab that we looked at. You can also animate the image with low or high
motion or loop it, which means it's going
to keep repeating. On the side, we can
see saved searches. If you search for
something often, you can add a search here. If I do a lot of products, can put product over
here and then we can do minimal products. For example, then
you hit Create. We can have different profiles. We can have some filters, only see the things
that we like, only the images, only
the upscaled works, versions, and other details. There's view options
for full or square, small, large or medium. This right here is
the chat feature. You can see people are
already generating stuff and we get to see it with that diffusion model
that we talked about. There's threads, so people are commenting on
this person's work. You can do the same thing too or just start writing stuff here. So I get to reply
to this person, re run their prompt,
use their prompt, and do other stuff with it. You can see a lot of things are being created
at the same time. There's also the voice chat. This is all of the images, and you can also
look at your images. You get to filter through things that were completed only, things that show up immediately or the ones that
are in progress. Or just said to Auto. We have a task compartment, which basically is an option for you to help me journey to
better their services. So you can rank their
videos and earn fast hours. You can be a curator, rank their images, do
a bunch of surveys, and then join their community. You can get some help over here. There's also a AI chat bot
to assist you with anything, updates regarding their tools, and then you can switch
between dark mode. Or light mode. And then down
here, that's your account. You can return to the homepage by clicking on the logo up here, and it's going to bring
you back to this page. Once you understand the
Midjourney interface, the anxiety feels a little less and you'll feel more comfortable using all of these tools. Now we're going to talk
about prompting basics. So the text that we put in here, what should we put in? How do we make it more detailed? And what are some of the
parameters that we need to understand in order
to get better results?
5. Prompting Basics: Prompting is the
heart of Midjourney. You can think of it as a new creative language that allows you as a person to
communicate with Midjourney. In order to communicate
efficiently, we need to know the right words, the right style, the right
grammar, and so on forth. That's exactly what we're going to learn about
in this lesson. I'm going to walk you through the anatomy of a good prompt. In the further chapters, you're going to see that at work where we actually
create things with it. With a good prompt, as we said, it's like a set of
instructions where we're asking Midjourney to
create us something. It always starts with a subject. Let's take a look at
some of these images. Just by looking at
these pictures, we know exactly what
the subject is. This is a wallet, a city, a deer, we have a
butterfly, the sea, a logo event can
be your subject, a helmet, spray bottle, and as you can see, that's where it all begins. That's your outline
that you have first and then later on
you build up on that. In the last lesson, we made
a a photo using one word, cat, and it did give us
something like this. But just by looking
at these two, you can just see the difference. We have so much more detail, so much more variation regarding the colors,
the contrast. Was with this, it's still
good pieces of work, but it's just not what
we are trying to get to. With Midjourney, we're trying to get really realistic images and we want everything to be
conveyed from our prompt. So taking a look at this prompt, a brown cat is our subject. Once we have our subject, we have to give details.
That's part two. Details are the brown part, looking to the side with
golden eyes and detailed fur. These are just some
additional information about the subject. When you're trying to
write your prompt, try to answer the questions
of what is that subject? How does it look like? What
are the colors involved? What are the details involved? Anything that can help Midjourney
visualize your prompt. Third thing is style. You have that subject
and all of its detail, but how does it
actually look like? Is it in black and white? Is it really covered
in vibrant colors? Is it in vintage mode? Does it have a three
D style, two d style? Is it a photorealistic portrait? That's where you put in
information like this. That is pretty much
it. The anatomy. But I will have a document
in the resource pack that has the format
available to you, so you can look back to
it whenever you want to. But to summarize, I'm going
to use these brackets. It always starts
with the subject, and then we add in
details about subject. An additional thing,
if it's a subject that moves or talks,
it's just the action. This doesn't really
apply to landscapes. But if we have a cat, you can see I said that
it's looking to the side. That's where that
section goes in. Next, we have style, and then lastly is
the parameters. These are technical add ons like the size of the image,
the resolution, the amount of variations,
the chaos amount, and the stuff that's over here, which I'll walk you
through later on. I will also add in
parameters here. That's basically the anatomy that you should
always keep in mind. Always start with the subject, give more details
about that subject, mention what the
subject is doing. If you're doing
something that doesn't involve a human or an animal, you don't need to really
put in an action. Mention the style and
then the parameters. I will do a dog example. Let's type in dog it enter and it's going to
start building our dog. Then this is a weak prompt
in Midjourney terms. Whenever it's one or two words, that's considered weak and it usually doesn't spend that much time
building your images. Now let's take the same dog and turn it into a
stronger prompt. A fluffy white dog sitting
in a sunlit window, soft Shadows, photo
realistic and some technical terms like 85 milliliter lens
and cozy atmosphere. So there's our subject, details about the subject, the action, style, and
the technical terms. This right here is
considered a strong prompt. Now, of course, I'm going
to show you a few of the other examples where
it gets really crazy. It's a full paragraph of
a prompt and just think that you need at
least a sentence in order to get a good result. Here's our white dog. We can even see consistency
among our variations, which we did not
get with one word. Because when you
give it one word, Midjourney assumes that you don't exactly know
what you want. It's going to give
you different styles, different dogs in this case, so that you can
make your decision. But here with a detailed prompt, I'm getting pretty
much the same dog, maybe different breed, but
with similar features, and then I can choose. If we go to explore and
look at some of these, let's go to the top of the day. You can see how this person has basically a paragraph
for their prompt. We have the subject, details, action, style, and
technical terms. Now, when you look at
other people's work, the technical terms are not
in the prompt, but down here. AR 59572, raw style, seven, and then this profile. That's why we have these
tags at the bottom. Let's do a little exercise. I want you guys to
create a prompt, write down a prompt
for a landscape scene. Whichever landscape you want, it could be a sunset, the beach, mountains,
whatever you want. But see how you can apply this basic anatomy
that we talked about and compare it with a very simple word
regarding that landscape.
6. Ethics, Copyright and AI: We can talk about AI art
and Midjourney without actually addressing
the ethical sides of using tools like this. First of all, we have to
understand copyright. Midjourney images
are new creations and they come about when you put in your detailed prompts, but they're actually trained on data from billions of
images on the Internet. That means other people's
artwork, their images, something they took
with an actual camera, and maybe their sketches even. What that means is that the
AI is actually learning from existing artists and is not taking a pen itself and
drawing things for you. You should never claim
an AI generated piece of art as entirely your
own manual creation. What are you supposed
to do when you want to share your AI artwork
on the Internet? Nowadays, everyone
in every platform is aware that you can create really realistic works like this on the Internet and
whether you're on TikTok, Instagram, Bhands Adobe Stock, there is an option for
you to check that tells that platform that you make
this video or image with AI. What you really have to
do is be transparent. If you want to sell
or share your images, you have to let people
know that AI was either partly of the process or
entirely part of the process. Next is to avoid directly copying the living
artists names. For example, make this
cat in the style of studio Jible because that can raise legal and
ethical concerns. You are probably familiar
with the scandal that open AI went through
with Studio Chipl. Avoid doing that with
living artist names. You can do maybe Picasso or Da Vinci or anything that
isn't living right now. Again, it's going to be
very clear that this was generated with AI and you made it in the style
of that artist. Next, you want to understand
Midjourney licensing. If you're on the free plan, you're not actually allowed to use your images commercially, which means to sell
and distribute. But if you are on a paid plan, you can use the
images commercially. But I do recommend that
whichever plan you choose, be sure to read the terms
on image distribution so that you don't run into
any legal or other issues. Now, if you just want
to use Midjourney for fun and you don't actually
plan on posting it anywhere, you don't really need to
worry about the stuff that I mentioned because no one's going to really
see your work. When you make an image
on this platform, it does go to the community, but anything in the
Midjourney community is made by AI and
everyone knows that. Apart from those legal concerns, there are also some
social considerations. Some people do worry that AI
will replace human artists, while others see it as a
tool to speed up workflows. If you think about
it, it's both. The best approach is
to treat Midjourney as your collaborator and
not a replacement for actual artist or a reason for you to claim another
person's work as your. You can think about
your own use cases. Are you going to use Midjourney
for fun for business? Do you want to sell the work, which you can and we'll
talk about it later, or do you just want to use
it for some brainstorming? Now, when I say you
can sell AI artworks, we are going to be doing it. It's a common practice, but we are going to be
doing it with all of the considerations that I
mentioned in this lesson. It's important for you guys to understand all
the points that I made so that when we
reach that chapter, you can apply them when you
post these works for sale. By the end of this course, you'll know how to
use Midjourney not just efficiently but ethically. Now let's move on to
our first chapter where we learn about the
basics of prompting, which is the text we put into the box that we saw
the text like CATs. We're going to see what
actually makes a good prompt, what are the parameters
and how we can master that skill so that
in the next lessons, we just use it smoothly and get exactly what
we're looking for.
7. Organizing your prompts: Once you have made a few
images in Midjourney, you'll notice quickly that
your promise can get messing. You'll try variation
style tweaks, and suddenly you have dozens of generations without
knowing what worked best. So right here in the create, we can see on the side
the prompts that we used. But it's hard to track for later which ones were the best ones and which one gave
us bad results. Here are some tips on how to
organize your prompts and a different version
so that you can always refer back
when you want to, let's say, make a dolphin
in this very style. Just like anything else, you need a system
to stay productive. The first thing you should do is save your prompts
in a document or a spreadsheet so that
every time you either discover a different keyword
to get you a good result, you can just jot it down in
that document or spreadsheet. Also helps to put down some notes regarding why
that was a good prompt. As I mentioned in the
previous lessons, we do have a document
for you that has some good prompts for you to try and also the basic
structure of a prompt. You can either
duplicate that and then edit it or just
create one from scratch. The second practice
you need to have is to name your outputs. So when you download something, you want to make sure that
instead of it being just this, it's maybe Dog Version one or dog with Golden
Lighting Version two, something that will
let you know how your prompting
experiments progressed. The last thing you
want to do is to go to organize and create
different folders. If you're trying to
get a specific design that includes, let's say, outer space and a desert, that's a folder that you need to create where you put in
all your different images. If I'm really keen on
getting the perfect cat, I should have a folder
that's called that. You can make one over here, name it cat and then you
could just put in cat. You drag the images here and that way everything
is organized. Of course, you're thinking, why should you even do this thing? As you can see, this is just all my creations
over the years. If I wanted to recall how I got the most
realistic human feature, I have to scroll all
the way down through my organized tab to find
the one that I liked, and then I have to
go through each one, see which one I favorited, and then use the
prompts that are here. But that does take time and it's just not
the best practice. That's why we want
to make sure we have the folder structure and the most important
one is the document. Where you get to try
different styles and also try styles that
you see in the Explore tab. For example, here we can
see that this person used this term for
a product shot. We can try that with our product and then take a
look at the style. The style is the thing that you really need
to take notes on. This is giving a very
dark professional mood to it and using the words
that this person used, we can see that they have
used high key lighting, dark red gradient background. Highly detailed composition. You could just jot those
down and then try it on your space photogeneration. Now, if you were experimenting
with Midjourney so far, making little fun images like
we did here with the cat, I want you guys to just open
your Midjourney history, pick one image you liked and
copy down the exact prompt. Then you can write a short note about what was good in that prompt and
what could be better. Then later on as
we move forward, you can work on expressing that.
8. Parameters and Settings: Midjourney has hidden
knobs and dials called parameters that lets you control how your prompt
is being interpreted. We saw them over here
when we went to settings, all these familiar sliders, determine how strictly or how loosely Midjourney
takes in your prompt. Then you can add the technical terms
that we talked about, such as the frame of your
image, the aesthetic, the size, even the model, and all the things that
you see right over here. These at the end
of your prompt can dramatically change the
results of the outputs. I'm going to walk you through some examples where
we don't have any of these terms in there versus
when we do have them. Now, there are a bunch
of sliders over here, but I'm going to walk you
through all of them and then tell you which ones
are the most useful. First step, we have image size, which is pretty
self explanatory. Here we have one, two, one, which means the length and
the height is equal a square. If you grab the slider, you are changing that
ratio and on the left, you can see a preview of
what that looks like. Portrait is going to be
your vertical shots and then landscape is going to
be your horizontal shots. If you wanted to reset,
just hit this button. You can also type in a number here or like we
did use a slider. Next is the model, just like any other
AI tools out there, they run on different models. The latest one is Version seven, but there are older
versions, as you can see. The reason why you may
even consider using an older version is for
very specific styles, such as if you want something
to not look as realistic, you can go back to an
older model so that it doesn't provide you something really
detailed and crisp. If you're following
a tutorial or are trying to get
a specific style, pay attention to
the version that they use in that tutorial so
that you get the same result because each of these are
good for a certain type of images and you'll discover that when you play around
with each of the versions. Now, draft mode allows
you to just play around with your generations before actually
publishing them anywhere. Standard is your usual
generation with Image journey. Now we also have mode
where if you go over here, it tells you that
raw mode replaces the default aesthetics of
the Midjourney versions. Also if you are into photography or have worked
with a camera before, raw versions are
usually the bigger, more complex outputs, where you have more
details to work with, more pixels to work
with, in that case. And then you in this case, would use it for really
hyperalistic footage. If you're doing something like an anime like this
one down here, you don't really need raw. But if you're doing
something like this, where you have all
these crispy details, then you do want the help of raw because it's adding a lot
of details for your images, it may take a little longer to produce that output for you. The next thing is
the aesthetics box. This is regarding the
style of your photo. First up, we have
the stylization, which is by default set to 100, but this basically
controls how strongly Midjourney applies its
own aesthetic flair. Low stylization gives you
a plain literal result. If you say black and
white with golden hues, it's exactly going to
give you that thing. But the higher you go, it gives you more artistic
and dramatic results. It's going to play around with that black and white
golden hues giving you different things each
time you produce it. Now, this is the
stylization bar. I'm going to show you how you can shortcut these
into your prompt later on because coming
in here and using the sliders is going to
be a little bit tedious. The second thing is weirdness. Now, this one is regarding the unpredictable qualities
that Midjourney can provide. A higher value gives you wilder, more dramatic outputs where the lower value gives you
more consistent things. For example, if I do a clown and I produce
it with zero weirdness, it's going to give me your
standard plain clown. If I increase this all the way, I'm going to get
really crazy colors, crazy features, maybe
scary at times. But this is a pretty fun
slider to play around with and you can see the shortcut
is dash dash weird. The next thing is the
variety which talks about how different each
of your productions are. Whenever you put in a
prompt and you hit Enter, it gives you four
different images. If we increase the variety, those four images are going
to look really different from one another versus when
we have it at zero, they're going to
have that subject consistent with
every production. Then down here, we have
additional options. The first one is the speed, which is how fast Midjourney
is going to produce your image and how slow
it's going to do that. The slower it does, the more time it journey takes to create that image for you, Wess if you rush it, it's going to just give
you a quick result. If you're going for a
really hyperrealistic image with let's say, a prompt that is
a paragraph long, you do want to put
in the time for a relax production versus a turbo which will
rush over everything. It will still give
you what you ask for, but it just won't be as
rich as it could be. The next thing is stealth. Now when you produce
in stealth mode, you're going to have your
images hidden from the public. The people in the community, like how we are in Explore, are not going to see
your new productions unless you get off of this
or you publish that image. By default, it's turned off, but you can turn it on
depending on whether or not you're just playing around or if you're
just experimenting. Resolution is
regarding the quality. We have SD, standard definition, which is 480 pixels
versus high definition, which is 720 pixels. I'm on the P plan, so you can see that
I have access to HD, but if you're on another plan, you may only be able to use SD. Just look at if this is
grade out for you or not. I could just switch
between the two. Last thing is video batch size. When you generate videos, this is regarding
how many videos you get from one prompt. Remember that videos
take longer to create and they will
take more credits. If you are running
short on credit, you don't want to
create four videos from one prompt and therefore, you can just come over
here, switch to one video, two or four if you have
tons to work with. These are the parameters and settings that are
within these sliders, but there are a bunch of other ones that aren't
really listed here. The first one is chaos, which you can also access
by typing down chaos or C, and that makes your outputs
really unpredictable. The higher you put
in the chaos value, it's going to be really crazy over there and the lower it is, the more controlled the
outputs are going to be. The next is seed or
you can do seed. Now, this is going to be the random starting
point for your images. Every image you see
here has its own seed, which is a few numbers. If you want to continue
that randomness, but with the same style, you can copy the seed and
put it in your prompt. Another thing is image weight. If you have a reference
image and you want to just talk about how much of a weight and effect that input image has on
your output image. You do IW. We also have Q or quality, where one is your
default quality and two is your highest quality. Now let's see how adding these parameters can
actually change our prompt. I'm just going to type
in a very simple prompt, say a fluffy Cat
floating above DC, glowing with purple neon colors. Let's do. Period
dramatic lighting. Shallow depth of field. Ultra defined. Here's my prompt without any of the parameters. Again, recognize the subject, what they're doing,
the environment, description, and
technical terms. Let's create landscape. Here are my cats. Very defined, fluffy for sure, purple light, ultra
high definition. And basically everything
we ask for is here. We also have that
shallow depth of field, which is the fact that it's
blurred in the background. So that's without any
of the parameters. And right away, you can see that my cats don't
exactly look the same. We have a fluffier one here. We have one with
different colors. This guy is super fluffy. We got a different breed,
maybe. I'm not sure. But the point is that
they don't look the same. The color purple, you
can see that here it's the lightest shade
and then we get a really dark one
here, the environment. We have a stormy
situation here, sunset. This one looks like the
sun is still out there, and here it looks like
the cat is underwater. Let's take a look and
see how we can use the parameters to get
a different result. First, I'm going
to click on this to get the same prompt up here, and we're just going to put in the most useful parameters
such as chaos, stylize seed. Now let's start putting
in our other parameters, starting with the chaos because already we had the aspect
ratio set to four to three. I'm just going to do AR. Let's do 16 by nine, chaos. I will do let's say 100. Let's add some stylize in there. I will also add some weirdness. These are all some random
parameter numbers. You can see here what
they run between, we have zero to
3,000, zero to 1,000. This variety, which
is the chaos. I did mine 100, which I think is a little
too high. Let's do 80. Then aspect ratio was
the image size, chaos, stylization and weirdness.
Let's see what we get. You can see when I
use the 16 by nine, it had to expand, so it's two rows instead of one. So already this is pretty weird. Great details, definitely a cat, but it's not really
making sense, and that's precisely how the weird works, the
weird parameter. G to try this again
and just remove weird. Let's put it in the chaos, the 16 by nine version
seven, stylized, let's take it down to 100 so it pays more
attention to our prompt. Let's generate that.
So this was before. We have all these parameters. Now I just cut it down
to only these three. Initially we had no parameters. It was just the image size. And you can see how different
these generations are. It's still a cat
floating over water, but you can see it's nowhere close to what we
were getting before. The only difference here
is that this is not weird and it doesn't have
any sort of stylization, so it's not really depending
that much on our prompt. Now let's do it again. Oh, we had variety set
to 100 in the last one. Yeah. Stylize is
if I go over here, it was set to 100, so we're getting completely
different styles. I'll just set this back to zero, and whoops meant to
add the chaos as well. But let's just compare
stylize 100 versus zero. And there we have
it. So now we have at least a decent looking cat. Similar settings versus when this is with the stylize
being set to zero. So it's being very literal
about what it's giving out. But in the previous set, we had stylized set
to the maximum. So we had completely
different variations of a cat that's
glowing over the sea. Let's use the same thing again. And I'm going to just work
with weirdness right now, which is the the weird
outcomes, 1,500. But as you can see, each of these sliders have a
completely different effect. In all of these, I did not
change my prompt one bit. I'm only working with those
parameters set at the end. You can see how 1,500 weirdness gives us
some weird results. You can do the same thing, remove the weirdness and simply work with
the stylization. High stylization
is going to give us artistic and less
connected to the prompt. So these are the results with 1,000 or the maximum
amount of stylization, where Midjourney
uses more creativity to generate your images. We have a cat, but it looks like it's being electrocuted
by these purple things. Here, it has a more
magical galaxy like look. We have these jellyfish
things around our cat and this just looks
like a dragon I guess. It also combined different
colors with our purple, even though I only had
purple neon in my prompt. Now let's try to look at seeds. I'm going to grab the
exact same prompt, make sure none of
these are enabled. Let's do a square this time. And I'm just going to do dash, seed, a random number. So you can see seed
is right over here. And the reason why we
did a random number is because seeds are
random by default. But the cool part comes after generating
this first round. Here's my random seed. You can see it used a completely different style for my cat, even though we had the
exact same prompt. I will do the same thing,
but without the seed, just to show you the difference to stylize zero version seven. The only difference is the seed. Now let's do a different prompt. We're going to put in
a fluffy dragon maybe. With pink, neon colors, pink and blue and all that stuff and just add in this
particular seed. I will do another one
without the seed. And you can just see the
difference between the two when we use the seed
and when we didn't. Now, for a quick exercise, I want you guys to grab a
prompt and keep it consistent, but simply change the parameters as we did in this
lesson and then find out the differences
between each of the quantities.
9. Hallucinations and Consistency Tricks: As we all know,
AI isn't perfect. Sometimes when you
put in a prompt, it produces something
strange, broken, or odd. These are called
hallucinations and they happen a lot with AIR
tools because as we said, they get trained on millions
of images and they try to combine their understanding
into one image for you. So Let's take a look
at some examples. You probably have seen them on the Internet too on Instagram
when you're scrolling. There is, for example, a human subject that
has 20 fingers. There is a house with windows
in just impossible places, text that looks like
gibberish and so many more. There is a way to actually fix these hallucinations because they're not
really permanent. All you really have to do is work on your prompt
a little bit more. Times putting in a single word can help you solve that
hallucination issue. Let's go over three ways that you can reduce
these hallucinations. Number one is to
be very specific. Instead of saying a
man holding a cup, you should try something
like a man holding a ceramic coffee
mug with one hand. Secondly, use reference words. You have to mention
something like realistic photography or architectural
blueprint to anchor the AI to a certain
style and for it to not get confused over those images
that it gets trained on. Lastly, you have to simplify your prompt because
overloading it with too many conflicting ideas can of course confuse the model. Let's try that example with
the man holding a cup. Midjourney doesn't have
that many hallucinations, but you could still encounter it when you're playing
around with it. That's our first one.
Let's see what we get. Then the second one, I'll
do a man holding a ceramic, let's say, white coffee mug. With one hand.
Let's take a look. There is an example
of a hallucination. I'm not sure what
they're holding here. They're also missing I
guess one other fingers, or is this combined? I'm not sure what's
happening here. The cop is above his fist, but his face looks fine. Let's take a look at this one. Again, another hallucination. They have one, two,
three, four, five, six fingers, then this
parts in the shadows, we're not sure what's
happening there. But again, the face is okay. Let's take a look at this.
It's a bit hard to see, but I think this one is a lot
better than the other ones. I think there's a
hole in the middle. It's a little dark, and then this one so this
one's actually fine. You can see out of the four, we were able to get one
that makes more sense. Now let's take a look at
the new ones that we made. It completely removed
the man's face because we put in the word hand. You can see what a big
difference that made. Here we got the right
amount of fingers. The ring is a little
weird, but that's fine. There's also some weird. I guess his finger
is elongated here. Number two, wide
amount of fingers. Weird position to
be holding the cup, but it does seem natural. Third one, wide
amount of fingers. The handle of the mug
is not really there, but the mug itself is indeed
a ceramic coffee mug. Then lastly, we have this one shorter finger and
we got a bunch of others, even have some details
on his fingernail, which we didn't really ask for. But you can see the
difference between adding a simple prompt
like this versus a prompt that asks the journey
to focus more on the hand and give us
a white ceramic big. Now, let's add those
reference modes. Because I'm going for something realistic, I should mention it. Let's click on use text. After our prompt, let's
say realistic photography. And then maybe ultra high. Oops, ultra high definition. Now it went for a product
photography mood. Immediately, there's a very
obvious hallucination. But here's the thing about
these hallucinations where some part
of it looks fine, but the rest looks weird. We can actually use the
edit tab to fix this, which I'll show you in
a different chapter. Because the only problem
here is this part. The cup looks fine,
the sleeve looks fine, and then the thumb
is placed right. We can also change this
with that edited feature. This is the second one. I
actually looks pretty good. We even have some
depth of field. Next one, really good
detail on the arm. And the right amount of fingers. This is the last one. Again,
a small weird position to hold the mug handle, but we can change this
with some editing. But now you can see
the hallucinations were a lot less when we added these two words regarding
the style of this image. We didn't have any weird
rings, weird fingers, no handle, and
then short finger. Then our last tip was
to simplify our prompt. Mine is pretty simplified
because it's one action, one man. Holding one object. But say you had an entire
paragraph where you're just repeatedly
talking about the mug. You're mentioning soft edges, but also rough texture, maybe the color white, but also a little bit dark gray. When you put too many
descriptions for one object, that can of course,
confuse Midjourney. Even if you're reading
that in a book, you're just confused as to what the mug
actually looks like. While we are on the topic, let's talk about some
consistency tricks. One challenge that Midjourney provides you with
is consistency. Keeping the same
character, product, or style across
multiple generations. Right here, we technically have a very simple
subject, a white mug. But if you're dealing
with something a little bit more complex, that's not going to be consistent
within the generations. Even here, we're getting
different colors and it's not that identical, different handles,
and so on forth. But there are some
tricks that you can do to maintain a certain style. The first one is using seeds that we talked about in
the previous lessons. Using the same seed with the same prompt produces
nearly identical results. This is useful if you want to create variations
that still match. Going to show you three
commonly used seed numbers that you can find on the
Internet in the communities, but these are just three
really different styles that I want you guys to try. Let's go for a very
simple subject. Let's say a mushroom or
red mushroom, maybe. I red mushroom castle on
top of the. Very simple. We have our mushroom
and because I didn't specify if I want
this to be a sketch, a minecraft style image or a realistic one,
it gave me all four. But I could just have
it give me a photo realistic one by doing
ama photorealistic. Immediately, you can see
the before and after. We have some pretty
decent stuff. Now, what I'm going to do is choose one of these.
Let's go with this one. It looks pretty cool.
I'll just go to UE. Let's use the image, and it brings it
in image prompt. It's going to use
the elements of our image and keep it
consistent throughout. I'll also use the same prompt. Just click on it.
It's over here. Then we're just going to do da SD and put in our
first SIT number, which is a trippy one and
it's just 420-42-0420. At Enter. At the same time, I'm going to do the other one. Image, use text. The only reason
we're doing image is because I want
the same castle. Otherwise, you could
just remove it for different castles in
those seed styles. Das dash set then let's do
a futuristic 12025, 2025. At the time that they
made this set style, I guess it wasn't 2025 because that's where
we are right now. It's not that realistic. Then we can take Dada seed and go for
a fractal like style. It's basically the Pi number, but without the decimal. So here's our result with
the image reference and the 420 s. The glow is very potent, as you can see, and the
colors are cien looking. You can see it
kept that style of one hill in the middle
and the castle on top, pretty consistent because
of our image input. Next up is the 2025 with
the image reference. This is supposed
to look realistic. I don't really see that, but it's definitely
a different style from what we had before. This is one, two, three, four. More definition for sure. If we upscale this, I'm sure we'll get a really
high quality one. But you can see how it's
really different from our 420. Then lastly, we have
our fractal one. You can see that they went for these dome shapes for
our mushroom castle. And their heads look different. I would say this gave
us a lot more detail, smaller details than
the previous one. I'm seeing that sharp dome
consistent across each one, maybe not this one,
but this one has it. I also did another set
without the image input. This is the fractal one. Again, we're having that
sharp look on the top, different castles,
but you can see the consistency with
that pointed top. This is our 2025 version
without an image input. It's going for another floor
on top of the roof look. Then we have the 42420
without the image input. It's looking the most
interesting, I would say. I could see it added these
depth of field as well. You can see how
different the style of the castle became when
we put in these seeds. I'm just going to do this without any seed
just to show you what it would look like when we generate that same prompt. Can see stylized zero, seven, that's the same
thing as we had before. So nowhere close to what we
were dealing with before, and each of them look
pretty different. Those were some
examples of seeds. You can get tons
of other seeds on the Internet and
even via ChatGPT, which I'll show you
in a further lesson. But you can use seeds as a way
to keep things consistent. Next thing you can do
is playing around with how much the image
that you put in as reference weighs in
on the final result. When you upload a
reference image, you can assign it more
influence using the IW command, which we looked at,
I think last lesson. Let's say I put in this image as my reference and I'll do something different. Let's put in the same prompt, but instead of a mushroom, let's do a glass castle, same thing on a hill
and I'll do one more with this
prompt, replace that. Let's do W zero
and compare it to. Here, it's using the default of, I believe, one, it's actually combining the mushroom
with the glass. I still has that hill structure, and it's actually
looking pretty good. We have a mushroom glass castle. In the next one,
we set it to zero, the input image
has no influence. As you can see, there is
no mushrooms to be seen. Even the hill structure
is pretty different. The last thing you can
do is multi prompting, which is using the collins. When you have two elements
that you're trying to combine, in our case, mushroom and glass, we can decide which element has more of an influence
in the final result. Let's say we have this prompt, but we're going to add a
mushroom glass castle realistic, but we'll do let's actually do that castle
on top of a hill. We'll do mushroom,
Colin Colin, two, then glass, Colin Colin, one, and I think that's it. Here I got an error because
I use version seven. I just did dash dash version
six, and now it's working. Here's our version six. You obviously see the
difference between the versions which we were
talking about previously, but we asked for more
mushroom over glass, and that's certainly
what we got. Not so much in the
two middle ones. But I tried it again with
more glass over mushroom, so two to one and then
Castle photorealistic one. I'm definitely seeing more glass this time compared to before. So now we have a mushroom
made out of glass. There's no castle in there. It's just a ball. Here,
we do have a castle. So that's looking pretty good. I'm going to show you
what these look in further lessons as well
because right now we are clearly dealing with
some hallucinations. As we move forward with
more detailed prompts, I think that would be better
part for you guys to see. But this is just an introduction
as how you can control and keep things consistent throughout your
image generations. Consistency, you may be wondering why is
it even important. It's important for
projects like storybooks, branding or even product photography because you
want that same subject. Same character, same product to be consistent in
all of your shots. If you were, for example, trying to advertise
a perfume bottle and there's mushrooms
all around it, you can reduce the amount of mushrooms by doing
this multi prompting, doing perfume bottle
two, mushrooms one, and that's going to take out and mellow out the mushroom
aspect of the image. Now, for a quick
exercise, as always, I want you guys to generate a fantasy character with
one of the seeds and then repeat the prompt
with the same seed but tweak the background
using the multi prompting.
10. Adding Styles: One of the most
exciting parts about prompting is when you get
to apply different styles. Midjourney has
absorbed patterns from countless and millions
of artistic references. You can actually
guide it to take your prompt in a different
visual direction. For example, we can do
watercolors, all painting. Realism, minimalism, brutalist, and so many other
styles that are out there. One style that we
have been using a lot is the
photorealistic style, which we put in with the
coma photorealistic. You can see it
definitely does look realistic in terms of details, highlights, shadows,
depth of field. It actually looks like
someone took out a camera and took a picture of this
magic mushroom castle. Going to go over
some styles today, but as I mentioned, there's tons more out there, and you can experiment just
by putting coma that style. If that style is very niche, maybe it's your
own drawing style, you can use it as
a image reference and then continue making
images using that. Let's use our same prompt. I'll stick to the
glass mushroom castle. I'll just remove the
multi prompting since we want to use Version
seven or the latest one. Let's try this with the
first style, photorealism. It's Version seven,
style is zero. And I want to set the
stylized to default, which is 100. Let's
run that again. Then see how realistic that can look like because that's
what we're going for here. Here's our first
set, great detail. That's some really cool
reflections going on. This is with the stylized set to 100, which
is the default. Again, photorealism.
Look at that. It looks really sharp, really detailed, and it
looks pretty cool actually. Not a lot of glass on there. You can see maybe the windows
could be considered glass. But other than that,
it's not made of glass, but it does combine castle
and mushroom pretty well. Now let's try a different style. I will go with our same prompt, simply replace the photorealistic
with a different style. The first one we're going to
look at is a painted style, which you can activate
by putting in painterly. This is supposed to
add brush strokes, textures and soft lighting. I will do, let's
say, watercolors. Then we can do another
one with, let's say, oiltlls and we can just compare. Since we're using
the word painterly, anything in that
category should work. But I'm just going to work with watercolors and oil pastels. You can see how great it
executes the watercolor. It even had that
textured paper which you would use for such painting. The details are there. We have great colors, and it's once again
a mushroom castle. This is the oil pastels. Those are some great textures that mimic real oil pastels. We can go for thick
acrylic paint strokes and see what we get instead. I'm not sure what
the style is called, but I want to see those thick
lines used to make castle. Let's see if it listens. There we go. It does have that
thick paint stroke effect. I don't paint with acrylics, but if you do, you can
see how accurate this is. Since you know what real
acrylic paint should look like, you can add more details. Let's say if this is too thick, you can put in soft edges in your prompt and it should
give you a softer edge. So that's our painting style. The second style that
we're going to look at is cinematic, delete these two. But in cinematic,
I'm going to do one with just one word and
then we can add to that. Cinematic with, let's
say, dramatic lighting. With cinematic, you can
think of a Hollywood film. You're going for depth of field, different moods, and of
course, good lighting. Cinematic is
supposed to give you a very high quality output, whether it's animated or
three D, in real life. You can see how great
these guys look. We certainly have good
lighting, depth of field, great detail, and that is consistent between
all of our images. Here I asked in the next set, I asked for dramatic lighting, so it's going for a
darker tone, more mood, and great separation of
shadows and highlights with, of course, high contrast. We can do the same prompt with our next style,
which is surrealist. Just type that in. This
is supposed to give you strange and dreamy outputs. I'll do one which just
surrealist and then we'll do glowing highlights maybe. Glowing highlights. Come let's add
another one dream. So this is going for things
that are just not possible. You can see we have a
champagne glass over here, we have a marble ball. Here we have a champagne glass holding that castle in
the middle of a desert. It's inside a glass ball. Here it's made of clouds and here it's like a
tree but a big mushroom. Things that are just impossible, you can activate by
putting in this keyword. Then in our next set, we asked for glowing
highlights and a dreamy scenario which it
definitely did deliver. Maybe not the first one, but the rest are pretty cool. The last one is going to be a minimalist look, put that in. Minimalist minimalism
is regarding the clean, simple look of images with
a lot of negative space. There isn't that much detail
for you to see and that's precisely the reason why
this style is unique. Again, a lot of negative space, not a lot of detail over there, one element in the
middle of the field. That's also minimalism and this one is the most simple one. We can do the same thing, but we can try, let's say, futuristic and then ray since that's usually the color
that's used in this style. In a similar fashion, we can do the opposite, which is brutalism then we
can do maybe rough textures, harsh lighting,
something completely opposite to what
we have right now. Here we definitely have gray minimalism and
it's futuristic for sure. Pretty cool. Let's see what the
opposite looks like. Definitely, we have
rough textures, harsh lighting, and
just the exact opposite of what we had initially. You can see how by
adding certain words, we're able to completely
change the style of our image. Now, for a mini exercise, I want you guys to write
one subject you like, say it's your card, it's your teddy bear, it's your pet dog and
try different styles. You can write some of the
ones that we mentioned here, but you can also go for
maybe animee style. Sketch style, charcoal,
three D style, two D style, whichever you want. But the thing that you need to do is have that
subject consistent, just like we did
with the mushroom castle and just see how these styles apply themselves
to that output file. In conclusion, style is the
secret sauce that turns your generic prompts like a mushroom castle into
something pretty unique.
11. Advanced Prompting Techniques: Something. Once
you're comfortable with the basics of prompting, you can explore some advanced
prompting techniques that give you even more control. So far, we have looked at the basic formula of a
good and strong prompt, how we can add styles, how we can avoid hallucinations, keep consistent
results with seeds, and now we're going to look at something rather different, which is the use
of curly braces. Curly braces allows
you to create multiple outputs and
all using one prompt. Essentially if I want to try say bird wearing
a wizard hat, but I want to see what
that looks like if the bird was green,
yellow or purple. Instead of doing three
separate prompts, I could do one prompt and use curly braces to get
those three sets. Just to show you what
that looks like, we're going to type in
a curly braces, red, blue, green, and curly braces, bird wearing a wizard hat. When I hit Enter,
I get three sets. One is red, one is
blue, one is green. Instead of doing this
three separate times, I could have this going
on all at the same time. As you can see, I'm getting
a red bird, a bluebird, and a green bird, all using version seven, nothing else, and of
course, our curly braces. This is going to be great if you want to experiment with
different subjects, but want to keep that
concept consistent. The wizard hat is consistent
across all of these sets. But what I'm experimenting
with is the color of the bird. In a similar fashion, I could experiment
with the animal. Let's do a purple. Curly braces, bird, dog, comma fish, really
different creatures. Hit Enter. We're getting
three more sets. Let's see how Midjourney puts the wizard hat on
a purple animal. And there we have
it. We have a fish, a dog, and a bird. All of them are purple and they're all wearing
a wizard hat. That is our first technique, the use of curly braces. The second technique is
called prompt injection, which we have been doing so far, but now I'm going to
give it more attention. These are ways of structuring prompts to force
certain behaviors. Essentially, we're injecting
it into that prompt. For example, using words
like ultra defined or trending on art station
or trending on B hands, eight K render for
three D works. This basically forces Midjourney
to give you that style. We have a three D sandwich
floating in outer space. Let's generate that. Then I'm going to start injecting
things into here. We can do maybe eight K render. Since it's a three D thing, we can do ultra definition, and I think that's and then
let's compare the difference. Once we ran those prompts, we are left with
three sets of images. The first one was the
most basic one where I just mentioned the sandwich
floating in outer space. It is indeed a sandwich, but it's very basic and I
didn't really say what sort of sandwich and so it
gave me a bunch of different things and
then it gave me this, which is a slice of toast. But then we took that a
step further where we added the word three
D. When we did that, we first get this
three D illustration. Again, I didn't say
what image I want. But the results are a lot more detailed than
our first set, just by using the
word three D. Now, we even took this a
step further when we added eight K render and ultra
definition in our prompt, right over here, we can
see the glow of light. We got some halo effects, some particles
around the sandwich. This one has some
depth of field, crispy lettuce over there, some shiny elements around it, and look at those bread crumbs. You can see how we started from a very basic thing
and then we slowly injected words into the prompt to get closer to our end goal. Now the last thing is
something called chaining, which essentially
means that when you start with a prompt, you get a result you want. Rerolling it or remixing it can give you more control
over your final image. For example, I really
like this image, but maybe I wanted to
have more breadcrumbs or maybe a rock in the middle or in front,
something like that. We could just go down
to more and then hit rerun. That got submitted. We can also vary
this very thing with either subtlety or
really strong presence. I want to do strong. I will go into these
options later. We can even use this as an image input and then put in the same
prompt, of course, perhaps add a rock floating or maybe with
rocks floating around. Then the same eight K render
and ultra definition. Now let's take a
look at our results. This is our re ran prompt. You can see it's
the same sandwich. However, it's not exactly
what I asked for. Here it's really close up,
here it's really long, and then we got two
random generations or the hallucinations
that we talked about. Then over here, this
is the variation. It's the exact same image input. We can see the sandwich. It has those golden
particles around it, but now it's adding some
additional maybe angles, colors, and other details. Then this is with our image
input it as reference. We can see again, it's
the same sandwich. Plus we got some rocks
floating around. That just gives me more control over that exact same sandwich. Of course, if you were to take the sandwich that you
like, like this one, and then basically use
it as image and style, you're going to
maintain the colors and all the other details
and maybe add some additional elements
within the prompt. We could do maybe
the same sandwich, same style colors, but
astronaut floating around. That's something we
would have to add in here just as we did
with the rocks. These tools let
you work smarter. You can generate
multiple variations without having to write
everything from scratch. We maintain the structure of
that subject we generated, and we just keep
on building up on it until we have
that perfect result. Now it's time for
your mini exercise. I want you guys to find a prompt where you get to use the
curly brace prompting. Maybe you can try one subject in three different colors or three different animals
in one setting. One animal in three settings, whatever you want
to do, run that prompt and then just
compare the differences.
12. Using ChatGPT for Prompting: We have this idea in our head, it's actually going to take a while for us to come up with a perfect prompt because great prompt writing
takes a lot of practice, but in today's age, we don't have to
do it all alone. There are other AI tools like
ChatGPT, Gemini, Deep Seek, and many more out there that
can help you brainstorm, refine if you already
have a prompt and further structure your
prompt for better results. An example would be me
taking this prompt, putting it into ChatGPT and
asking for more details. I could simply just
copy this guy put it in ChatGPT and have it expand upon a certain
part that I want to fix. Maybe I want the
lettuce to be bigger. That's something I could
have a tool like ChatGPT, refine it for me, send it out, I copy paste it and get
my better sandwich. Here's how these tools. I'm going to speak about ChatGPT in this
lesson, but again, there's tons of others out there can help you in four
different ways. The first one is idea. When you want to you have
this very simple subject, but you want to turn it
into a abstract idea, can ask ChatGPT to suggest five, unique ways to
describe that subject. For example, we could
ask it to give us five creative descriptions of the beach at sunset and it's going to
give us exactly that. The next thing is
prompt structuring. I can help us organize
the prompt into subject, style, mood and
technical parameters. If I just have a
regular sentence that doesn't exactly fit
in our base formula, we can have ChatGPT reconstruct our original sentence and
turn it into a strong prompt. Third is refinement. If your image didn't
come out right, say I put this prompt and
I got a burger instead, I could ask ChatGPT
to fix that for me. Maybe it will decide on
changing a few words, emphasizing on the
fact that we ask for maybe a sandwich that
has toast on it. We can also have it expand
upon an output like this. We can ask it to make
this more cinematic. Help me give a prompt to Midjourney that will give me
a more dramatic sandwich. You could do that.
The last thing you can do is batch prompting, which means you put in
your prompt and you ask JAGBT to generate dozens
of variation at once, which we can then test in Midjourney using
the curly braces. We're going to try all of
these four out in this lesson and we're going to start right over here with our chat bot. Let's start with the
idea generation. I'm going to choose
a different subject. Let's say I'm going to go
for a halonia very random and we're just
simply going to ask for first, let's set the mood. I'm trying to make
images with my journey help me find two
creative ways to describe a jalapeno Army
writing bell peppers. It gave me two different
ways and you can see that it starts with
the style as well. We have epic fantasy war scene
or playful cartoon style. I did not even ask for that yet. Let's focus on the
prompt itself. We can see that it
has the subject, the action, the description,
the colors even. And it's giving the technical terms
like dramatic lighting, smoky atmosphere,
fantasy illustration. Just to show you
the starting point, we're going to first of all, let's make the same idea but
with a very basic language. I'm going to paste it right here and just remove the
descriptive things. Let's say fighting against bell peppers in a battlefield, then no mention of
the other things. Then we can see how
we can elevate this. I'm also going to
paste the prom that ChatGPT gave us,
paste it right here. These are both with version
seven. Let's take a look. So here's our basic prompt. We have character
illustration here. We got some three D
pieces, close up. These aren't really jalopinios. It's people fighting
amongst bell peppers. This is our second set
and you can see it understood that the fighters
are actually jalopio. In the first one,
we did not really get that apart from this guy. Chachi Pitt was able to
communicate to my journey that we wanted actual Jalapenos fighting with bell peppers. Now that I have this, I'm actually going to do it
without fantasy illustration, just because I'm going for I didn't decide
on the style yet. We're going to leave
that for later. But now we can take
this to step two. I'm going to just copy it. And then we're going
to have ChatGPT fix up a few of the things
that we notice in this set. First of all, these
look more like pickles. This one looks fine,
and the rest is okay. So I'm going to go over
here, put quotation, paste my prompt quote and have it dig deeper into a certain style and more
details on the character. We're combining two
and three right now, which was prompt
structuring and refinement. Let's go make these changes
to the prompt above. One, make sure four things that I asked ChatGPT to do. First of all, I don't
want to see anymore of the maybe pickle face
or human soldiers. I want this to be more epic with cinematic and
harsh lighting, vibrant colors because
we're dealing with peppers. And I'm just mentioning for it to look like
a Sparta movie. I'm really going
for that epic look. Then lastly, another element
I want in there is add smoke in the background from
all the intense fighting. And let's see how Chachi Bit takes our points and
combines it into one. Copy that and let's
see what we get. Then the last thing, we're
going to do the batch prompt. Let's say how opinions
aren't really my thing. Give me different variations of peppers or
vegetables, let's say. Let's say of tomato. Bananas, same setting, lighting, et cetera, different subjects. Give me three prompts for each. Sorry, one prompt for each. Okay. This is our tomatoes
versus bell peppers. Paste that in. Number two. And of course number three. Now let's take a look
at our first one. We've got a ego one here. For some reason, they're all ego characters, that's not good. Here is our tomatoes. They actually look like
tomatoes. That's good. We got cucumbers, more
legos for some reason, and then bananas.
I like this one. This is a great example of when you're supposed to use another technique
that I'll show you later, but don't worry too much
if Midjourney just adds an element consistently in your images such as
ligos in our case. I did not mention any
ligos in my prompt, but you are able to disclude something within the results. That's something I'm
going to show you later. But ignoring the first round, we can see that the
tomatoes look pretty good. Here it even considered
these Sparta outfits. We got more legos
for some reason. Here we got some goopy
looking tomatoes, cucumbers, more legos, and it's just funny how it brought
the legos out of nowhere, but nonetheless it
looks pretty cool. Going back to our first set, this is where we started, and we ended up here. Again, as we saw in
the previous lesson, if you see something, something that you
like, for example, this one, I could
just rerun this to get more like this
and there we have it. This pairing of Midjourney for visuals and ChatGPT for words is like having two
creative assistants working together to give
you these great results. Now, you can type into ChatGPT for yourself
a different subject. Maybe you can try a futuristic city in
three different moods. Hopeful, dystopian,
whimsical, the 1980s, whichever mood you
want to go for, and see what prompt ChatGPT gives you for you to
try out in Midjourney.
13. Lab: Generate Image in Different Styles: Now, let's put everything
we learned together in this chapter in
a hands on lab. The task for this lab
is to pick one subject. It could be an animal, a human, a vegetable, even, and then generated
in three different styles. Here's an example. My subject
is a lighthouse on a cliff. My first style
would be cinematic. Second style is watercolor, third style is cyberpunk. Run all of these prompts and
then compare their results. You can also use ChatGPT like we learned about in the
previous lesson and see which one you like the
most and which one was able to be communicated from
your prompt to Midjourney. Point of this exercise is to show you how much power you have by simply changing the style of the words or even
the words themselves. The subject is constant
in all of your prompts, but you're simply
changing the mood and that completely
transforms your image. Once you're done
with that, you're now ready to move on
to Chapter three.
14. Image Types: Versions Overview : It. For this chapter, we're going to focus more on the different types of images
that you can make with Midjourney and what are some of the creative
features that you can utilize when trying
to make those type of images? For this lesson, we're going to focus more on humans, faces, limbs, and anything that involves a lot of
complicated features. Unlike a bell pepper or
a Jalapino with humans, we're dealing with
hair, skin, eyes, eyebrows, ethnicity,
race, age, even gender. All those factors play into how Midjourney interprets
your prompt. Now, this is one of the most
used features of Midjourney. It's also one of the most
requested uses of Midjourney because people are just trying to create realistic humans, like you see on the
screen right now. But the problem is,
as we all know, AI has a history
of struggling with faces and especially
when it comes to limbs. In the previous lesson
when we were trying to do that ceramic cup in a hand, we saw that it gave us
additional fingers, it gave deformed fingernails, shorter fingers than
the others and there were just a bunch of weird
outputs that we saw. That was with a very
straightforward prompt. But ever since version six, those areas have improved a lot. However, you still need some strategy for
that ceramic cup, which I'll just bring to remind you guys what
it looked like. So here we use Version seven, which is the latest version, but we still got
some weird results. It doesn't really matter
which version you use, of course, the later the better. You still have to be
very specific with your prompt and use
certain keywords to trigger my journey to
pay more attention to the fingers and the human
structures in general. Before I dive into how we can get better faces
and hands and all, I'm just going to
walk you through the different versions because
pretty sure you may have wondered why we use Version seven for some things and then Version six for the others. Versions are basically
the new models that Midjourney gets to use and every time they work on something new, they release a new version. Each of these versions are known for one thing
or multiple things. Usually you would use the
latest one by default. However, going back to
some of the versions give you more opportunity to
experiment with certain things. Going back to version one, which was the very
first public version that Midjourney gave out, that was very experimental and people just don't
use that anymore. However, some like to use it because of its rough outputs. I'm going to just show you
the different version with a very simple subject and you
can just see for yourself. Let's do a sponge character. Tanning on the beach. Then we'll do dash
dash version one. You can see one over here. This is also referred to as the Alpha version
and you can see already it's taking a lot of time to develop
this simple prompt. Here's our Version one prompt. Already, the size is very small, and it's not looking exactly like a sponge
character tanning at all. You can see the options
down here are also a lot limited compared to
one of the version seven. But if you're into this
sort of art style, which is very abstract, it's very messy, you can just go back to Version one by
typing in Version one. Now, Version two improved the realism and
composition a bit more. It also handles
the colors better. Right now, we have
this very vibrant yellow and it just doesn't
look that good on the eye. It also gave us the opportunity for more consistent images. A lot of people refer back to this version for fantasy scenes, but not human faces because it's still
struggled with that. Let's do dash version two. Better yet, let me just show
you what humans look like. We will go for a man with a top hat and fire mustache. Let me just show
you Version one and then we'll do the same
thing with version two. I think humans are a
better way to show you the improvements
within the versions. Already you can see the speed between Version one and two. This one's already at 56. This guy is still
going on from 13. So here's Version two. It has an outline of
a man but not really a man and it's very artsy, I would say, even though
we didn't ask for illustration or colored pencils. Here again, we have
very random shapes that don't really resemble
what we asked for. Then we go into Version three, which was a big jump
in photorealism and it also was supposed to improve
human figures and anatomy. Let's try the same prompt
with Version three. The different
communities, people use this version for
fantasy, sci fi, and just general
purpose realistic arts because of its more
predictable outputs. Whereas here, it was just
completely unpredictable. So here we can see the jump
from this guy to this guy. It looks more like a person, but not exactly that human
figure we were looking for and it's definitely a lot
better from version one. The fourth version was a
complete model redesign because it was
supposed to give us a stronger realism and
more creative freedom. I also works better
with complex prompts. This is a very simple prompt, but I'm still going to show
you what that looks like. On version four. This was known for
stylized concept arts, portraits, dramatic
lighting, and it was also supposed to handle
intricate prompts better. Immediately, we are
seeing actual humans, and it was pretty fast when
giving us the subjects. It's more concept
art, I would say, but at least we're
getting the features. Jumping onto Version five and 5.1 because they
did release both. I was meant to give us near
photorealistic outputs, sharp detail, improved depth, and better handling of humans. It was also supposed to be
faster and higher resolution. Let's see that for ourselves
by putting in Version five. Then we'll do Version
5.1. And just compare. People also use this apart
from realistic portraits, they use it for landscapes, product mockups and
cinematic visuals. This is version five. We're seeing that
depth of field. Contrast, we even got some smoke and here his
mustache is actually on fire. Here we're getting a lot more
detail with our subject. Now their top hat is on fire. However, that mustache
is pretty consistent. I didn't really
specify the outfits, but it goes well
with their top hat. Next, we have Version six. This was supposed to be
the cutting edge realism with better consistency. It was also supposed to have a improved understanding of composition and the styles
that you mentioned. We worked with Version
six previously, and to point out here, this is the version you have
to use for multi prompting, which was the thing we did with, I believe the mushroom castles. These were Version seven. These were version six.
When we multi prompted, we had to use B six. Let's go back up and see what our prompt looks like
with Version six. Which I would say it's the version that a lot
of people go back to. The earlier versions, not so much unless you want to
do a specialized project. But in my opinion, it gets better from
Version five and up. You can go to
Version four if you want these concept arts, but they're not really for me. You can just see the
speed of this guy, and here we're actually getting the better understanding
part because I asked for the mustache
to be on fire, and here we can see
that it actually is. The last version is the latest version
as of April 3, 2025. So that is Version seven, and this guy was known for enhanced realism,
even more realism. It also gives us the
great improvements regarding textures,
lighting, and anatomy. It also gives us draft mode, which is the feature that allows for us to just experiment and prototype the images ten times the speed and have the cost
of standard rendering. This is for you to
just experiment, see which prompt and which
image looks best before you use the full credit requirements and publish it on the community. Is also the
personalization feature, which we talked about
earlier in the chapters, where you get to rate
a series of images, and it's going to enable Version seven to work according
to your needs. If you're really
into mockups and you keep telling Midjourney
through the personalization, that that's the type of
thing you want to create, Version seven will keep that in mind when you're asking
for a product mockup. It has even better improved
prompt interpretation. It works better with
complex prompts with the realism factor. Another thing that you can
do with this is that you get to use the reference
code from 6.1, which is that CID reference. Let's see what this looks like. I will go with one of
those seed references or dash dash seed. Let's do the 42420 and you can see that it didn't
switch to another version. Here's our version seven thing. This one looks
amazing. Great detail. We got the fire reflection, and this is with
a seed reference, completely different, but
it's looking pretty good. You can do the same thing
with the other ones. 2025, the ones that we
went over previously, and a tip, I think
using SRF rather than C works better with
the earlier version. Here's another reference
that we used previously. It's giving us
this illustration, which looks pretty cute and you can just see the
jump from this to this. Those were just a
quick overview of the different versions
in case you were wondering why we're switching
to six and then seven. Then of course, how did Midjourney evolve
from version one, which is this guy
two version seven. It definitely works a lot
better, especially with humans, and that's going to
be the main point of our next lesson
where I tell you how to work better with humans, their faces, and their limbs. We're going to go for prompting, what words you can use, what tricks you can try
to get certain effects, and then you would be a lot more comfortable and ready for
our project chapters.
15. Human Faces and Limbs: Alright, welcome back. Now
we're going to go over how we can generate better human
faces using prompting. So let's go to
Explore and see what people use to generate
great works like this one. So we have an attractive woman, head turned to look out from behind and towards
another man walking by, shot through glass reflection, natural lighting,
natural colors. This is Midjourney
understanding of an attractive woman looking out the window. Pretty simple. I will say that Midjourney doesn't really
struggle with faces. I mostly struggles with limbs like the ceramic
hand that we saw. Let me see some more. Let's get some human faces. For example, this person's
finger, there's four. Here we're getting three. And the rest is fine.
Then of course, we have an extra leg
for some reason. But you can see that their faces look perfectly fine.
They are all smiling. The position of the arms
is also pretty natural, but it just struggled with
some of the basic features. We have another example. Face looks great. Her arms look good too. It's not always, but that's the area you would
have to work on later. Here we're getting
this cut in her pinky. I don't know what's
going on there, but her face looks pretty good. This is an area that AI
still struggles with. It could be because it
doesn't get trained on that many finger images, legs, images, and all that
stuff, but mostly faces. And unless you're doing
a super close up, it will not really generate
the right amount of fingers, natural looking
fingers, give you extra foot and all that stuff. Even though the versions
are getting a lot better, as we saw in the last lesson, you still need strategy
when using these tools. First of all, you have to mention more details
about that subject. Apart from seeing a man
or a woman or a child, you have to be more specific
regarding their ethnicity, their age, what mood
is that photo in? What's the lighting
in that photo? Instead of saying a woman, you should say something like a 30-year-old woman
with curly hair, brown curly hair, smiling
with warm natural light. We're going to try
that right now. A 30-year-old woman
with brown hoops. Brown curly hair. Then we can have her
smiling, warm natural light. Here's our first woman smiling,
which is not that bad. But if I'm looking for a
certain type of subject, I need to be straightforward
with me journey. Giving more details in your prompts allows
me journey to ground itself when generating
your prompts. It's going to be adding
certain types of details that give you a
more natural human figure. You can see how much more
detail we have here, the friz in her hair, the harsh warm light
coming in, the shadows. Even we got dimples, freckles, brown eyes,
and all of that. The other ones are
not bad looking, however, they do
look very generic. And don't really match that
subject that I was going for. We're also getting
different styles here. This has a retro look to it. This is more cool toned, so is this one and
this one's warm toned. Whereas here with this prompt, I was able to get all warm
tones for my subject. If you know the specifications regarding your subject,
be sure to say it. We can also add in
different elements here. A 30-year-old, let's
say Albanian woman. With brown curly hair, smiling, we have age, race, gender, some features,
and the lighting. Now it's going to
be incorporating some known Albanian features. If you're going to be working
with different ethnicities, you do want to specify
that over here. You can do the same
thing with race as well. Just put in the word
like I did here. You can see now it's
not that different. However, it's more
focused on the face. The eyes are a bit more open, and we're getting more details
regarding clothing too. And these look really good. Now, when you're
dealing with faces, because there are close ups, we do want to add some
camera terms as well. These would be your
technical terms from our basic formula, and when I just use the same prompt and
fix the natural light, there are some things
I should put after. Because I just want to deal with faces and I want Midjourney to really focus on giving me
the most amount of details, I need to put in words
like portrait photography. Which is the type of
photography that is done of the face
and not the body. We can go for golden hour, which is when the
sun is setting and the light reflects on the face, but in a very warm tone,
similar to this one, we can put in the location even. But I'm going to skip
to the camera lens. We're going to act as if this
was taken with a camera. An aperture of 1.8 and
let's do ultra Definition, the injection, trick we
learned about earlier, and let's do four k quality. Here's our new set of images, a lot more definition that
aperture lens is really coming through and the
detail is just amazing. As I mentioned, with faces, it doesn't really
hallucinate much. Now let's move on to hands, which is the bigger
struggle that any AI tool has with
this particular subject. We're going to make
a very simple image. Let's say woman
with hands hoops. One sentence, I'm not saying what that hand is
doing, where it is, what's the lighting,
and immediately it assumes that I want the woman
with her hands visible. It's not bad, but I wasn't intending for human
faces to be in there. We got to just take
this very simple prompt and give it less
room to hallucinate. I putting the human
face in there, even though I
didn't say portrait photography or
great face details, it just puts the face in there. Because I only said woman. Let's put in more words. Woman holding, let's say, a te monk with both hands
resting on a table. So we're going to
give it less room for hallucinations by describing
what the hand is doing. In my case, it's holding a coffee mug and both of the hands are
resting on a table. Then we could add
in the other terms. I'll do superior, actually. We can say golden hour,
street photography, which is a different type of
photography than portrait, fork quality, ultra definition. I'm just removing the
lens specification. Instead, let's put in maybe
85 milliliter cannon lens. You can also do
without the lens. I'll do one more. If you're not sure what lens you're
supposed to use, there's no worries
about that at all. Here's with our lens. We got the right
amount of fingers, a slight fish eye
effect, which is fine. Here we're getting
a weird thumb. However, the number
of fingers is fine, getting weird fingers again. And here the fingers
are merging. It worked better here. We got the right
amount of fingers, no human face though,
which is fine. Here it looks pretty good. Just a little bit of up there, but we even got rings. We got more rings here,
right amount of fingers. And here just getting
this issue here. But you can see
how just rerunning the prompt gives
us better results. Again, I could just re run something with
this as the input, and it could just work
more on the fingers. If I'm doing a ring commercial, this is a type of
prompt I need to give it rather than this. Now let's try this in
even more definitions. Both hands resting on a table, we can say comma
fingers coping in monk. Then we can even switch
to a different style. So let's go for dramatic
or just cinematic maybe. And then we'll do just
for fun, watercolor. Here it's looking a lot better. We even have that
dramatic cinematic look. Then for fun, I just did
a watercolor version. If you saw that only one of them out of the four
looks the best, remember that you can
always re run that image to get different variations
of that correct version. Let's do variation, strong. This image has strong
influence over these. So to summarize, start
with a strong prompt, a lot of definition
regarding your subject. Once you find a correct
version in one of the outputs, re run that with variations. Now, try this yourself with a different subject and maybe
put it in different styles.
16. References: This is where Midjourney
allows you to have more personal control
over your images where you get to rely on
uploaded footage as references rather than just
purely on text prompts. So what you can do is either use the images that
you've made that are listed here in your Create tab or just upload something
from your computer. We previously looked
at the references. When you have this image, you have the option
to use this as a structural guide,
style guide, prompt, then we also have a omni
reference in the back here, which is regarding
the person's likeness and the form of an object if you're not
using a human subject. Essentially, what that does
is that when you put it here, it's going to look at the
elements of your image. I'm just going to type
in a man, very simple. But it's going to use
this image as reference. You can see how similar the smile is and then we also
have a similar face angle. He has curly hair. This guy looks pretty similar like
a brother or something. Then this guy has the
exact same lighting. All of them have that lighting. It's just the addition of
the beard and the mustache, and of course, different
facial features. Here you used the
structure of the image. But we can also have this
woman put in to image prompts. And then ask Midjourney to give it one of these glowing
sparkly effects. I'm going to combine this
subject with this look and then put in another really
simple prompt, a woman. While that's happening, let's try to get different styles like anime sketch art
without any reference, so I could just add it
onto my subject here. Now, here's our subject, into an environment
like this one. We have those sparkly
glowing elements, curly hair, same
lighting, same colors. Here's our animee sketch art. I could add this in my style reference and then
put her in image prompts. Once again, I'll
just do a woman. We can also take this a step further by putting this
in style reference, omni reference,
and image prompts. Omni reference is going to make sure that it looks exactly like her and there's no altercations when generating your new image. For example, here we
go a different face. The hair is a lot straighter, a lot shorter, less puffy. Here we had shorter hair again, longer hair, and it makes
these variations on its own. But now we have our
subject in animee format. We even have a pencil down here. Here she has really long hair, but with our next set, which is still generating, it's going to follow it in
a more stricter manner. Go ahead and run
the same prompt, but I think it's too
much for the reference. I want it to be less intense. Let's play around with
this and I'll do one where we just have
omni reference. Here's our subject, does have
that colored pencil look, not exactly anime, but it's
certainly not a actual photo. Once again, we have that
same lighting hair, sweater almost. Then here I asked
it to only look at the human subject at
75, it's 0-1 thousand. This is a very low reference. You can see it combined
the hair from this image. And the subject from this image. She has wavy hair now. We have some sketches in
the back too, which is fun. We can do the exact opposite
by just taking in one of these subjects as reference
and put in a woman. I will also try
bringing her to life. You can see using this as an omnreference
doesn't really do much because it's meant to
influence another image. When I put it in, I just
got the same thing, but I guess with a body. But here I added the sketch as an image prompt and I asked for a photo realistic portrait
photography look. This is her sketch, and I guess this is
her brought to life. Then from here, if I like
one of these things, I could use this image
as our image input. Here I switch to
Version six just because it's known for
better realistic shots, but I also do one with Version seven so we can see
which one works better. Here's our
photorealistic subject. This is what she
looked like before. Now she's brought to life. Here's with Version seven, a lot more close up, but Version six gave
us a full face. So we started from, I believe, which sketched do we
used. From this image? This guy and we ended up here, all done using references. How else can we use references? As we saw, we turned a
sketch into an actual photo. That's one of the main
examples that you can try. You can also do the other
way around where you take this realistic shot and you
turn that into a sketch. We can also take a specific
type of photography, like a Fuji film
photography style and paste that effect
onto a image like this. If you have a black and
white film, camera. You can take a photo with
that, upload it here, and then use it to turn
her into a vintage model. Of course, remember that you can play around with
the image weight or the omni weight
with IW and OW. The image weight works better with the older
versions. We have 0-3. With the omni weight, we have 0-1 thousand, there's a bigger range there. As we learned before, you can combine different styles from your uploaded images. If I have this image, let's say, I took
this from my front, I could turn her into an anime character or
maybe a three D character, a ego character,
whichever I want to do. Now, you can try this yourself since we've reached
the end of the lesson, where you combine
different references, use either a human subject,
different environments, different styles,
and see what you can make to blend
different concepts. So
17. Edit Tab Pt. 1: Midjourney isn't just about creating from scratch or
from your favorite prompts. It's also about being able to edit and iterate
your original ideas. There are some tools
here in the Edit tab, which is right below
Create that lets you take an image that we have into something completely next level. This tab is what we're going to explore in the next couple of lessons because it's pretty intense and it's pretty
cool at the same time. First of all, to quickly
access that part, all you have to do is grab
your photo, then go to Edit. This will bring you to a
version of the Edit tab. You can see we're
still in Create and it's showing me the prom
that I use for this image. On the edges, we have the ability to scale
down or scale up. With this middle button, we can move our image around. We also have these smart guides. You can extend an image by
grabbing the edges here, and then you are able to
undo with the tools above. You can also reset the
entire thing and go back to the original
or reset the prompt. Here are some templates
for your aspect ratio. This is the square which
we currently have. This is scale in and out. Then down here, it's the tools that you
get within the tab. The first one is erase, which as the name says, it erases the image. You can use a brush
size to make your brush smaller and simply use it to
erase parts of the image. The reason why you would do
this is because it helps you blend in an image or
images into each other, which we'll take
a look at later. But as you can see,
I'm just easily erasing the sides of
my original footage. When I let go, it
stays that way. If I made an accident like this, I could use the restore brush to bring back the original work. We also have a smart
select feature which separates the components of
your image all by itself. Right now, I only have a close
up shot of a human face, but the way it would work
is that you click once on that part and it detects
the edges for you. You can erase that selection
or erase the background, which means getting rid of everything except
your selection. Can clear the selection like this by clicking that button and that concludes the in painting tools which
are right over here. Then below it, we have some quick adjustments
that we could do. First one being vary. You can vary in a subtle
way using this as your reference or have it
done in a stronger way, which means this
original image will have more influence
over the variations. Below, we have upscale, which means it increases
the resolution of your image and provides
you with more details. Once you're done with the
prompt and your changes here, you can also change
things up here, so edit your prompt. I'm going to turn this
into a raw image, which as we said, makes it less creative and
focuses on realism. Since this was a
portrait photography, I do want to add raw there. All you have to do is put in rock and then we're going
to hit submit Edit. Before you do that,
you're also able to change the previous sliders, which we already went over
and then see the new results. Here we go. I asked for a raw footage and we extended
the edges of the image. AI was used to give
the subject some hair, or details, maybe a sweater over here and it's
looking pretty good. Once you see an
image that you like, you can upscale it for
even higher quality. Upscale, go for
subtle if you want the original details
or create it for Midjourney to put in a
few additional elements. I'm going to try
both for you guys. So you see the difference. We only get one upscale. Here's our subtle version
and the creative version. You can see there's clearly a
difference between the two. Now, if you want to dive
deeper into editing, you just have to go down here, which will open in Edit tab. Now we're actually moved
to the second tab. This was Create, which is
where we were all this time, and now we're in
this second tab. You have the two option
between edit and re texture. Edit is in regards to the
components of the image, whereas retexture
is about the style, the mood, the texture
of the overall image. The first thing
you're able to do is get a prompt suggestion. So if I click this,
it's going to do some thinking and give me a
prompt based on this image. This is pretty accurate. It even gave us some cool technical things
that we can look at. Close up freckles 6.1, it decided that 6.1 is
the best thing here. The next thing you're able
to do is move or resize. This is the exact same thing as the previous page where we were using the
quick editing tab. Then we have paint, which is just the in painting
features, erase to remove, restore to add, paint size, smart select, and then down here is where
things get interesting. If you've used Photoshop or any other platform that
deals with image editing, you've probably heard of layers. This is a new feature for Midjourney and it
basically allows you to add different images
on the same canvas, combine them, and
have them interact. Right now, we have layer one, which is this layer, but I could add another image. You can add from File
or add from URL. So I'm just going to go through some of the images
that we've made so far and see what I could combine
with my human subject. There is my second image and now you can see I have
layer one layer two. When you have the
checkmark selected, that means you're
dealing with that image. If I choose layer one, I'm able to move it around. Select Layer two, I'm able
to do the same thing. Then I could use something like this to combine
the two images, maybe put both characters
in one situation. Et's zoom in. You
can add things from a URL as well if you don't want to upload
from your local device. Then using the painting, you remove the extra
edges from the top image. This gives Midjourney
more room to play around and fill it in the
empty areas with AI. Grab layer two and
do the same thing. All right. Here is
my base footage. Now since I have two subjects, I could do another prompt. We can go for raw and
then try high resolution. Standard, 100 is good here and just simply let's go for
landscape, submit our edit. When you submit an edit, your original is
over here and you get all the variations
on the side. You can add a new tab
to start editing, a new project, essentially. But I'm just going to go
over here and show you how Midjourney attempted to
combine these two things. Before, after, you
can see it extended some parts and tried to
give a realistic look. Of course, it struggled
because these are two really closed in faces
placed next to each other, but that's okay because
we're going to look at a better example later on. Now you can upscale
this to the gallery, which means that it's
going to give you a high definition
full quality version and put it in the gallery
with the rest of your work. You can also download the
image to your local device. Now going back to that Edit tab, when you want to
start something, you can also create a new image by clicking
on that button. You can either start from a URL or upload an image from
your local device. Then this way, you have the exact same tools to work with, but you get to
decide your frame, maybe you start with
a background image and then add on your
different elements. The next thing is
the re texture tab, which I will just go back to that first example to show you. When you go over here, things
are relatively the same except for the fact that
you're submitting re texture, which as I mentioned,
is the overall mood, overall color, and it will preserve the
original structure. What I will do is put
cinematic dramatic. Raw version seven.
Let's see what we get. The structure is
pretty much the same. There is some hallucinations, but I think this one
looks the most normal. You can see how it
didn't even change the divide between
our two footage, but it was able to add that cinematic and
dramatic effect over the original components. That's how the we
texture tab works. Same settings, same panels, it's just regarding a
different part of the image. Now let's move on to our next lesson where
I'll walk you through a decent example
where we start by creating the different
components of our remixed image. Use the edit tab, use the prompt structure, change it up, and
then try to make a full composition using
the create and edit tab.
18. Edit Tab Pt. 2: We're going to start with
the different components of our composition and then put it all together with the edit tab. First, let's come up
with a nice idea. I'm in the Explore tab and
just scrolling through, you can get yourself some
inspiration and then either use that image as an input or try to recreate it with
the create tab. I'm going to go on random and see what thing we're
dealing with here. I was thinking of
putting a model in a gown standing on the moon surface with the
Earth in the background. But you guys can come up
with your own concepts and scrolling through
these amazing works can get you inspired. Let's head over to our
creative and first, we have to decide the
structure of the composition, meaning that it's certainly
going to have to be a landscape shot so we can actually fit in all
the other components. I'm going to go for 16
by nine landscape and everything else is going
to be the default choices, except I'll choose raw for the mode because I'm going to go for a hyperrealistic shot. If you want to do
illustration or other things, be sure to switch to standard and work back
through the versions. Maybe version five or six to get better
aesthetic results. Now let's use our prompt box. You can say background
or maybe hyperrealistic, background photo maybe
landscape photo. Then we're going to do period, ultra definition, and all the classic terms that
we've been using. Here I'm using landscape
photography to force it to go for a Zoomed
out situation. And we're getting some
pretty decent things. You can see that
it even worked out the moon's surface and the
Earth is pretty far back. We have some options
to work with. I think this one
looks the coolest. I definitely love the
little hill situation, and then we can put our
model right in the center. What I want to do now is redo the Earth because
it looks like a moon. What I'm going to
do is put this in here and remove earth. My put space with
space in the back. We're going to do some
strong variations as well. Just compare things
and see what works. Put that in there,
our original prompt. Then what we're going to do to not include the element Earth, we can do no Earth. These are pretty
good. I'm trying to decide if I want the Earth to be here or if I want to do another image and merge
it into our background. But I'm liking these so far. Let's see the one
where we say no Earth. As you can see,
Midjourney Listen. We do have the sun in the back and there's just
no signs of Earth. The no command is one of the ways that you get
to exclude things within your images if you were not to move
to the Edit tab. I think my favorite would
have to be the guy. I'm going to heard it for now so that I could
come back to it and simply upscale it for the
perfect starting point. See before and after. I'm going to filter
through my footage. This is before, this
is after you can see the texture and
quality coming in. I will also do a creative
version so we can compare and decide which
one works best for us. I'm just going to
hide this since I decided to keep the
Earth in the background. I think this looks better, let's heard it and work
on the other components. Back to the creative tab, we're going to
input this image as a style and then put
in the next element. We can do a spaceship
or maybe we can do a model wearing a red
gown, a long red gown. With her hair in an upto loading up slightly and her
hand reaching out. To this side.
Hyperrealistic, full body four k quality, and then we'll do dash for all. We will try it this way and
then another way that you can try this is removing
that style reference, finding the way your
model looks like, and then merging it with
another photo of a dress. If I can't really
get her position, that's something I could
recreate without this input. So this looks pretty good. Her face not so much and these guys I'm not sure why she's
facing the other way. But so far, this is my favorite. Let's now create some variations and we run the same prompt. This is our strong variation. This is our subtle. Actually
liking this one now. I'm not sure what those are, but we can get rid of that
with the exclude brush. You think this outfit
is more space looking, not so sure about that area. So now, once we have
the position of our model decided on, I think I'll go with this one. We can use it as a image brant, not using the moon background anymore and putting in the
same text. Okay? Let's see. After sight we can do her
skirt is very long and is floating in the air as
well, showing its velvet. It's smooth fabric.
Then of course, dash dash a. I'll do a
hyperrealistic model here. Now, while this is happening, I'm going to ask a chat bot to enhance this just so that we're using all the
things we learned about, enhance this prompt to make
my image more surreal, focus on the flow of the dress and hair floating in
the air due to gravity. Put a quotation mark
and paste that prompt. Then this is for Midjourney. Midjourney version seven. Let's paste that and
compare the two. Here's our first prompt, which I think looks great. The fingers are okay, and I'm loving the outfit too. Then this one is
completely magical. This is our second
set. It's almost done. Once again, it
looks pretty good. I'm gonna run this again, but with a white
background so that it's easier to separate when
we put it in the edi tab. Now, this is even better because we see the
entire subject, and it's not cut
off at the bottom. All right, so let's take
a look at the fingers. Got some hallucinations there. This one looks fine. You can't really see
her fingers here. And the last one,
it's a little weird. But focus more on
the composition of this image for now
because eventually, my journey would have to
regenerate her anyway. For now just focus on the dress and the
position of the model. I think this one looks
the best for now. I'm going to like it so that
I could find it easier. Then going back to our backdrop, I'm going to think of other
things to put in here. If she's standing here, I think we do need
to remove the Earth. Yeah. I think I'll replace the
Earth with our subject and then work on some
other elements, maybe. Let's try perhaps
a shooting star. Then put this as our
style reference. Just make sure that
the aspect ratio and the style is consistent, or else they will
not really fit well together when we push
them all in one image. There is our shooting store. I think this one
looks pretty cool. I'll just upscale this and then we forgot to do the same
thing with our other model. With the images that you've
already upscaled and finalized, just download them. Okay, so now we
have three images to combine into one and then later on we can re
texture it to put in an overall mood or
an overall element, maybe some sparkle dust
or a fade film scenario, which we will do in
the next lesson. Now that we've reached
the end of this lesson, make sure you have
your background image, your main subjects, and
that additional element. Upscale them, edit your prompt,
whatever you have to do, and just download them
so that you're ready to move on to the next lesson
where we move to the edit tab.
19. Edit Tab Pt. 3: In the Edit tab, we're going to start by making a new image and then upload
our background image in here. Once we have that, we can add our other components by using the ad button and just doing the same thing
with that additional element. Once I have all three,
I can start positioning them and resizing them to fit nicely on my
background image. Move resize, scale it down, move it to other places. For the model,
we're going to have her be somewhere in
the front for sure. Then the star, I
think I will increase the size and have
it go from the end. Now you have to clean up
the edges and once again, give Midjourney more
room to work with. Starting with layer two, which is our subject,
go to paint. Actually, we can do Smart
select because she's very easy to separate.
Click once on her. If you missed the spot, just do another click and it should give you a
nice separation, then I'm going to
erase the background. Now with the top image
because it's a little harder, I'm just going to
switch to paint and start cleaning
out those areas. Now that I can see the head, I'm going to just
position her in front of the earth and then
go to layer three, which is our base layer, use the paint tool to remove the areas
around the other images. Because we want
Midjourney to remix the two and think of ways to connect the different colors
and different elements. Go to go to layer two
and clean up this area. Let's try to make the
brush a little bit tinier, roughly, remove
this middle part. Now I have my three elements. Going to clean up the
edge here as well. Okay. Then maybe this
little moon or sun, whatever it is, and leave
little room for hallucinations. Layer three, remove
that part two, and I'm just going
to go behind it to make sure I don't
miss a spot in the back. Same thing here. Now we're
just going to put in a prompt. Let's start with this and then edit in our other elements. You can see I input those
two elements as you saw. Everything else
should be the same. I'm going to do maybe a
little bit of variety, a tiny bit of weirdness. I'm just curious what
results we're going to get. Hence why I did that. Now we have journey's attempt
at merging everything. This is our weird and variety
sliders coming into play. I would say this
is the best one. I worked around the
selection we made, which is fine, since we're going to re
texture this anyway. But overall, you can see how it generated and filled in
the empty spots that we had. This is where we started. Then with this prompt
Midjourney filled in these pixelated areas with
what it thought works best. Now, in terms of which
one I want to use, think I want the one where the model's hair is in the air. Then I'm going to extend
the frame a little bit. Something like that.
Let's actually go to the ratios 16 by nine
and zoom out a bit. I will submit this again so
it can work on expanding the background just to make things a little
bit more dramatic. Right now, you don't
have to worry about the colors not matching
and all of that, even though we did use our background image as one of these style
reference inputs. But overall, we're going to do the final coloring and blending
with the re texture tab. I think this one looks better. I like the shadows. Once you have that selected, simply go over here and put
in the different values. Cinematic, cool tone, color grading,
bright highlights. And then we can do maybe some
stars in the background. Dash dash, I think we
have the settings here. Let's try one with these
and then remove the weird and the chaos for
another set of results. For my purposes, I wasn't able
to get the re texture tab to give me at least the
original structure of my image. What I will do is run that same color grading
effect into my edit tab. I think that will be safer. Hopefully, by the time you
guys watch this course, the capabilities of
the re texture is a lot better and it listens
more to the prompt. Let me just input
our things here Glue tone colors,
cinematic, dramatic. Then for the sky, I want to put stars
in there too, blue sky, starry sky. We got some stuff in the back. And this is looking
a lot better. I'm not too sure about
that shooting star aspect, what I will do is go
to paint and erase it completely and then
ask Midjourney to replace it with a
longer shooting star. Right now, Midjourney is only working with this
little section. Oops. Here's our
different shooting stars. I forgot to change the prompt, but nonetheless it
understood the assignment. When I zoom in, even though the quality is not
that good right now, we can see that it did a really good job with
the subject's hair. The only thing missing really is the shadows from the subject. I'm going to attempt to
do something with that. Let's erase the bottom part
or any part that needs the shadows and then
mention the shadow casting. Shadow casts onto the moon
surface round the gown. Now it gave us a
little bit of shadow, which works pretty well,
especially this part. Overall, a good look. Now that I have my completely
edited and generated image, I could upscale it
to the gallery. That's going to push it back
to the Crea tab where it's working on giving us the highest quality
possible for this image. There we have it. This is our image made
completely from scratch, meaning that we worked on
each component separately, brought them into the edit tab, combined them, blended them in, and we got ourselves this really cool
model environment where this person is
advertising that dress. Can also try this yourself
with a simpler concept, since this took a long time, you can generate something
like a bowl of salad and then blend in your
favorite vegetables. Now that we are comfortable with the edit tab and the create tab, it's now time to move to a different image type,
which is patterns.
20. Patterns and Tiles: System Midjourney
isn't just for photos. It's also great for patterns, textures, and other
design experimentation. So if you ever wanted
a pattern to put onto your mockup or put into the
background of your poster, you can also come to
Midjourney to make one. So there are a couple of ways to perfect the pattern prompts, and I'm just going to show
you a before and after. First of all, let's decide
on what is the pattern. I will go with a very
simple flower pattern and then we can do a
geometric one so we can see how Midjourney
handles the repetition that either has to be symmetric
or has to be abstract. Let's start with a 16 by nine. I will reset this to normal, go back to standard, since it is mostly
an illustration. Go over here and do a
background of many flowers. Two D Illustration,
vibrant colors, and then that's it, something very simple and see how a journey
interprets many flowers. So as you can see, it did
give me a background. However, this is not a pattern. We define a pattern as a seamless repetition
of a certain element, whether it's a stripe pattern, it's a poka dot pattern, it's something that needs
to repeat consistently. Here we're getting
different elements at different sizes and it's
just simply a background. To convert this into a pattern, simply grab the same prompt, and I will change
these words with seamless floral wallpaper with various flowers and then
put in the word pattern. Let's run this again. Then we'll do another
one where we put in the keyword tile, the parameter dash tile. You can see this went
into one of the tags. So here you can
see the jump from this simple background to the high dimension
flower wallpaper. We're getting, I guess, that's a butterfly, different
angles from the flowers. We got I guess acrylic
paint going on, and some pretty cool colors. Now, here's our
seamless pattern. You can see how everything
is flat in a sense, meaning that if we were to have a glass surface in front of us, all of these flowers would be meeting that glass
at the same time. Whereas here we had some
that are a little far away, different angles, some
shadows, et cetera. We can also try this with a
different type of pattern. The whole point here is
to use the word pattern, seamless and add
the parameter tile. Let's try geometric pattern with the colors black and
gold, modern luxury. I think that's good tile. I will do another one
with the word seamless. Now we're getting some
pretty cool backgrounds. If I put the same prompt
without the tile, we're going to get something
a little different. This is with seamless added in. It's doing a better job
at the repetitions. Of course, it's
not that perfect. And this is without the tile. Already, we can see
it struggling with the connections, which in a way, it is still pretty good, but it's not exactly a tile design where every
part of it somehow connects. We have the dark colors
with the shiny gold, but that disappears here. The first one, we're not getting there's no pattern
to begin with. Here the lines are merging and here we have a bunch
of random shapes. But when we use the word tile, we can see the consistency
at every edge of the photo. Squares everywhere,
these shapes everywhere. That's just Midjourney way of understanding consistency
for a pattern. We can also put in a
whole different design. Let's try seamless B wallpaper, pastel tones, dash dash tile. And there's our B background. This one is pretty good. What we can also do
is use your pattern, just like we did with
the other images and experiment with different
colors and aesthetics. Let's try one that is maybe retro 90s web graphics, tile. However, we uploaded
our favorite pattern as a image prompt, so it should keep the
structure of this photo. I will try another
thing, once again, put that in as a image prompt
and try a different design. Let's try this.
Then just for fun, we'll do one more that
is completely different. Here's our 90s
retro web graphics. We can see the green
coming through. The colors are definitely retro and we even have
some glow over here. This is our baghouse, I guess, that's how
you pronounce it. It's a lot more modern,
a lot more flat. But again, we have
our first piece. This is our serials dreamscape. A lot more detail on our Bs and we even have some
flowers thrown in. Now, because this
is a tile file, when we edit it using
the same tab that we looked at in the previous
lesson, let's open it here. You can extend this to
whatever size works for you. I could extend this to
get the rest of my B. To trigger it, I could just erase a tad bit from the edges here so that it could work on connecting this new section
with the older one. With the B, the flour, and this other B. I think
this is another B too, then we got a half flour. Let's do the same prompting open another tab here
and see what I wrote. Seamless B wallpaper,
paste that here, da tile. But now it's only
going to work on this side. There we go. Got some variations, same style, same bees, but it's extended. That's how you can create easy
patterns with Midjourney. If something is not
to your liking, for example, this
area, remember, you can always use
the erase tool to have Midjourney try that
section once again. All I have to do erase it, put the exact same prompt and see a different
version of that B. Now you try this yourself
with different aesthetics. You can pick your
favorite subject, turn it into a pattern
like we did with the B and then put it into the
aesthetics that we looked at. We went over Surreal dreamscape, BahusRtro 90s web graphics, pastel tones, and
you can even try a luxury version and just see for yourself how this
pattern technique works.
21. Motion: You can now create short looping animation using your images that you
made in your journey. When we went to the Explore
tab, there's a video tap, where you get to see these
amazing works animated. All of these started
with a image. And then people animated it. However, Midjourney does have the option for you to start
animation from scratch. What we're going to do is go to our crea tab and start by bringing some of
these guys to life. Right over here, we have
Auto and then Loop. Auto is going to create a video starting from this
image and then Loop is just going to create a looping animation of some
movements within your image, but then start back on
this original scene. So let's go with high motion on both with this very
simple flower look. While that's happening,
I'm going to zoom into a human subject. Try that in. Now
we are still 50%. I'm doing four animations
at the same time, so I don't blame it at all. But the eye version
that we're doing, I'm going to try
the frame animation so not the loop and show you
how this is going to work. So we have this
really high quality, highly detailed people
with green and purple. I'm just going to upscale it and try some
different variations. That's basically going to
be our start frame which we looked at, I here. Animate an image, but
with a starting frame. Here's my first
looping animation of this really nice
flower background. You can see how it starts from our first image and
then continues on. That's a loop. There's
a starting point, it does around and then comes back to that
starting point. The symbol for a looping
animation is this, but we're able to
rerun the same prompt or use it as a starting
frame and extend the video. This is really cute. The others are still running. Let's go up here and look at
these really defined images. I think I will go with this one. Now I'm going to drag this
into my image reference, put this guy in and do a super Zoom parto
realistic image showing only the
people's details. And then run that prompt. Hopefully, our
animations are done. This is our video.
It does not loop. You can see that doesn't go
back to the starting frame. You have different
types of videos. This is a pan motion. We have each of the elements
moving left and right. Then this one where it separates the butterfly and
pops in one of the flowers. Midjourney is clearly very
diverse with these videos. We even have a three D animation where this flower is
moving in three D space. There's a flower
blooming and here is the flowers just
acting differently. And it looks pretty cool. This is our other subject. She's attempting to walk. She even turns around. You can see the way it moves through three D space, zooms in. I didn't really ask
for any movement yet. It's pretty interesting how
it understands that this is a human subject and that's a dress that she gets to move around with her fingers. This is the loop, so it just
goes in, comes back out. When you have a video, I'm not going to
do the loop ones because there's not
much to do with them, but when you have
a video that you like, you can extend it. I'm going to take this
footage because it zooms in on her face and
then it stops. You can go to extend manually. Starting frame will
be this video, and then we can say
what the model does. Model stars into the camera with an enchanted
look on her face. Or maybe with a
smile on her face. Her hair is floating in the air. You can decide the motion
here and generate. I'm going to let my video
extend as it is marked here. I'm doing high 16 by nine. Now it's nine second. Previously was 5.2, so that's
definitely an extension. Then we will come back
and see what we have. This is 9 seconds long and
you can see the model's phase is a lot more defined since the camera does
eventually zoom in. These are different
variations and you can see how much detail goes
into each of these seconds. Down here you can see where
the extension happened. It started from 5 seconds
and then extended to nine. You can also make
this full screen. If you wanted to check
for any imperfections, you have the option
to then re run this, use this as a start frame, so you can extend it
further, use the prompt, and then go for even more
extension over here. Let's take that out and
that's how you get to turn a static image into motion
and then extend it. Now let's go back
to our eyeball. You can prompt this
subject as we did and then animate it for social
media purposes, album covers, or
even storyboards. It's still in Beta,
but as you can see, did a pretty good job with the details and the consistency within our starting frame. One thing that I want to
try while this is happening is zoom out from
our first image. Starting frame, but
I'm going to move it to I'm not sure which
eball we ended up using. Let me try. I think we use this. Track this into image
prompt and then to a woman's face with both her
eyes being green and purple. Hyperrealistic,
portrait photography. My goal is to zoom out of this eye and then
show her full face. Here are the eyeballs. Great shadows over there. We even have pupil movements. This is the second one.
There's some dilation. Because this is zooming
into her eye by default, I'm just going to extend this starting frame and then I'm going to do
my zoom out situation. Because the other images, it didn't really
zoom out that much. So but let's go to
that same video. We'll do let's see. Portrait of a woman with
purple and green eyeballs, gorgeous eyes, great
detail in her pupils. Then let's see, soft features, red lipstick, hyperrealistic. Same terms that we've
been using all this time. Portrait photography. And just submit that. Let's
see if it can understand the fact that we want
to zoom out and show the full face. That
can be hard to do. Usually, the other way
around works better. But my main goal was to get this galaxy looking pupil
moving and of course, have the lashes move
in a natural way too. So with motion, there are
some options down here. The first one is a resolution
that we already went over. SD is your standard definition, HD is high definition. So you may want to
switch if you're doing a specific aesthetic. Then we have video batch
size, which is 1-4. If you want something
more concentrated, you would want to do
one per generation. But if you're just trying
to deal and finalize the concept, you
can go with four. Then of course, the rest
is your usual things. When you have a video
like the one here, you can go to more and
just re run it, copy it. Make other things with it. When you go inside, extend this with one of
the auto features, low motion means that there's
not a lot of movement. High motion means that there
is a lot of movements. Now, I chose high
motion for all of my works and that's why there's a lot of movement in her dress. She's moving around a lot. But if I were to extend
this with low motion, her movements should be
a lot less dramatic. I'm going to let these two run. Video does take longer
to generate naturally. That's why we're not
going to see the results until about a minute max,
so I will be right back. So here's our extended
one to 12 seconds now. Here's when it turns
to low motion. Here was high, high low. You can see how around
here she stops swinging around so much and the camera
just zooms in on her face. We got a bunch more,
I'm going to skip. Once again, we zoomed in
on the subject space. That's just Midjourney
determining the next move for us. Now here's the next
one to go in here, it's supposed to be zooming out. However, it did not really
understand the prompt and it instead just zoomed
in even more very slightly. Got some eye movement, got a blinking as well. And it didn't really zoom out. But I will say as a general tip, if you see one of
your base videos, zoom in by default, just keep on extending upon that because Midjourney already understood that that's
the main subject and I need to zoom in or
zoom out in another case. Instead of me ever typing zoom in on the subject,
it did it for me. Extending that is no problem, and as we continue on extending, it's going to work more
on the subject face. Because as you recall, the image that we made initially because it
was in the editor, didn't really have a good face. It's deformed it's
not fully there, and then the second one was just completely out
of the picture. You can see how the
motion fixed all of that and gave us a pretty
cool result in the end. That's the motion feature. They're still working on it, but as you can see, it's really easy to use and all you have to do is
press a bunch of buttons. Now, try this with
your own creations from the last couple of lessons with the editor tab and see what motions you can
create with Minturing.
22. Post Production: Midjourney outputs are amazing, as you've seen so
far in this chapter, but you can also take
this another step further using the post production tools that are available to you. So these tools that I'm going to mention are not
part of Midjourney. However, if you are proficient in the tools that I'm
going to mention, you can really generate
some great content for, as we said, your social media, your portfolio, and
other platforms. The first tool is going to be an obvious one
that is Photoshop. Whole thing about
Photoshop is that the editor that we use
in the previous lessons, that's something that you could have done in
Photoshop as well. However, Midjourney brought
it into the editor tab, so you don't have to
export your work, import it into Photoshop, do the editing, then bring
it back into Midjourney. You can completely finalize
your work within Midjourney, use the same AI model to do those little fixes and then
bring it into Photoshop. With Photoshop, you get
to retouch the images. A great example would be the
composition that we made. It looked great, but
when you zoomed in, our subject doesn't
really have a face. That's something you
could fix with Photoshop. It's a really tiny space. And if we were to
regenerate this, we would get rid of
the rest of our work. Another thing that
you could do with Photoshop is putting stuff in the background while maintaining
this entire section. If you try to do that
with Midjourney, it may alter the
texture on the moon. I may change the height of these hills and replace
it with something else. To avoid that, you can bring it onto another platform
like Photoshop. You also have the layer option, which means you get to separate each of your edits
in a separate layer, and then that way you
can work your way back and forward in a
more efficient way. It's not one permanent image so that if you mess it up a little, you
have to start over. But you can go to that one
layer that has that problem, fix it, and then go about
the rest of the editing. It's the same concept as
the layers in the edit tab. Just to show you
right over here. You have this exact same
capability within Photoshop, but with a lot more options. If you're not familiar
with Photoshop and that seems too intimidating, don't worry because
now we have Canva. Canva is a free online
photo editor tool where you get to
make quick layouts, social media posts, marketing
visuals, and so many more. They even have animations
and presentation slides. If you make your perfect
image or your perfect mockup, you can just export it from Midjourney and bring
it onto your Canva. That way, you're relying on perfectly customized
work for your, let's say, sales pitch, for your social media, just remember that
you do have to label it as AI if you did generate
the photo with Midjourney. The last thing is something
called Topaz Labs. This is a tool
that also uses AI, but it's used to
upscale images for print and digital use
without losing quality. So here in Midjourney, we do have an upscale feature, but that only upscales
to Midjourney maximum, which is not anywhere close to what we need for
an actual print, like a giant print screen, a billboard, and
that sort of thing. So when you upscale it, you can see that this is
our aspect ratio now. It used to be 16 by nine, so you can see how
much it expanded. But with a print or a
large digital screen, you want even more than that. Go ahead and export
things from your journey, import it into Topaz, and then try to
experiment with posters, merge, large prints, et cetera. The only free tool from the ones that I
mentioned was Canva. However, if you do want to use Photoshop and you
don't want to pay for it or want to see if it's even something
you'd like to use, there is something
called Photopea. This is pretty much like
Photoshop except it's free. So you can try things here. You can see all the
exports you get to do, and there's a lot
that you can create. Now, before I end the chapter, I'm going to show you guys the personalized tab because
it's really straightforward, but you just have
to spend time and rate the images in order
to unlock a profile. So you can see, when
you go on this one, there's this weird number
with P, this code. Same thing for these guys, where basically,
when you do this, it's going to use what you
like within your profile, so your aesthetic and apply it to your new
image generations. These three pictures
that you see, which you can refresh like this are from someone's profile. Notice how similar these styles are even though the concept, the subject is different. When you use the
term dash dash P, it's going to put in your
profile and you will get these really nice
consistent styles across your generations. Right now, by default, your personalization is locked. All you have to do is create a new profile and then
start collecting. So in B seven profiles, you can also do the same thing
with the other versions. You just go in there and
you have to rate 200 images for majoring to understand your aesthetic and preferences. I really like realistic things, and I'm going to keep that
consistent with my ratings. You can also skip an image if none of them
is to your style. And it will give
you another set. So once you do this, you're
able to unlock the profiles. So take some time to do this yourself, and
at the same time, you're able to see what sort
of works you get to create using Midjourney and all of the tools that we've
been over so far. Now I want you guys to take an image that you've
made with Midjourney, export it to either Canva, Photoshop or Photopea, and see what sort of post production
works you can make. You can try a poster, social media post,
a presentation, just a photo, and just see
what you can come up with. I
23. Project 1: Your Dream Room - Moodboard and Inspiration: Welcome to our
first project where we're going to design
your dream room. Think of this as building
your personal oasis, the kind of space that
perfectly reflects you. But before we touch on building
the images in Midjourney, we need to gather
some inspiration for that personal space that
we all want to have. Can draw inspiration
from your current room, a room you've seen
on the Internet. But this lesson
is just for us to create our first moodboard within Midjourney and get some inspiration from
Google Images or Pinterest. I'm going to use Pints
for this lesson, but you can use any
other platform, and we're going to start
searching for rooms, different design styles, or even color palettes that
we find interesting. The reason for that is because Midjourney is going to work best when you have clarity
about your own vision. We're not going to be
using the Explore page anymore, but external platform. When you come on Pinters, if you haven't used it before, it's a platform where
people put in their work. You can draw a lot
of inspiration, make moodboards on PinterS and basically organize
your different ideas based on the colors, the styles, the project. People like to comment and like, you can save stuff. It's really cool platform if
you haven't used it before. Before we start searching, you have to ask yourself, what does that dream
space look for you? Is it a modern minimalist area, cozy cottage core,
high tech cyberpunk? Is it earthly
filled with plants? Is it full of fun
shapes, fun colors? What does that room
look like to you? Spend about five to 10 minutes thinking about the
room you want. If you're just not sure what
to even start searching for, you can just put in dream space. Or better yet dream room. Plenty of options. You can see they're all
rather different. We have one where there's a lot of black and
white pictures, we got plants, we
got green on green. We have floral stuff, lights, and a lot of
really beautiful rooms. Now, the point of
this project is not to create something
that's realistic because you can actually make this if you had the
budget and you were able to put in a bunch of plants and lights
in your current room. But we're going to
be doing some rooms that are just
impossible to make. If you're really into robots, your room is going to be filled with different
electrical components. Maybe there's a mini
robot flying around. That's the thing that
we're going for. Let's start making a
decision regarding our room. Let's look for
different aesthetics and see which one looks best. I'm going to delete that and put in modern minimalist, the
ones that I mentioned. This is one aesthetic
that you can explore. Another one is something cosy. Lots of plants, lots of colors. Next one is a completely
different vibe. Got high tech, neon colors, not neon, but RGB colors. We got nice interesting shapes. This is one of those examples that I was talking to you about. Notice how it is labeled as AI. This goes back to
our ethics lesson, and this is not
something that you can just go on Amazon and get. Try to think of really unrealistic rooms that you could make with Midjourney.
That was high tech. There's also cyberpunk. It's a lot more crowded
than high tech. Got some examples here out
the windows, the city. We can do earthly. Again, a lot of
plants can also put in AI in your search term, you only get those
AI generated images. Here you can see the tree
trunks inside the room. Got a whole Jacuzzi, I guess. Amazing colors. This one is in a cave and they all
look pretty good. Now, what we're going to do
is make a board on Pinterest, or you could just download
the pictures directly onto your local computer so that you can upload them
in Midjourney. I want to do a high tech room. I will go to I'm going to try to combine cyberpunk and
the high tech stuff. This time with the
term AI and try to think of what my room
is going to look like. I definitely like the
window facing the city. Just download the images. Then in terms of bed,
I'm liking this one. It's very simple framing. There's a light bottom, the sky screen looks
pretty cool. Save that. I'm just filling in the
different parts of a room. We have the window,
the ceiling, the bed. Let's try to find some working desk that we can
put into our dream space. Maybe we can put in bedroom, see what other stuff comes up. I think this is cool.
We have the desk, we have some shelves, some plants even,
that's a good mix. I believe I have four
images right now. I'm going to stop here and then if we needed more
components to our room, we can come back here
and get more images. Now let's go to Midjourney and we're going to
go to moodboards. Let's create our first
moodboard and upload images. These are the four that
I found on Pinterest. I'm also going to
add from Gallery. When you go to the main
menu in the moodboards, you can see the name, use it in a prompt, rename it, view the creations you made with this moodboard,
or delete it. You can also filter through your moodboards if you
have more than one. As a search bar, you can create a new one and begin
creating things. Before I started the chapter, I also did my
personalization task, which was to rate
the 200 images. You can see now I
have one profile, which is my global
version seven. I basically had to
choose the images that I liked out of 200 like we saw. If you are not seeing this, then you haven't
completed the 200, go back to this one, Version seven profile,
and when you click on it, you get to do this
whole thing again. If you want to use
another version, just be sure to switch
swap to one of these. That way you will be consistent with the version
that you are using. Now we're going to go back to our moodboard and
essentially pick up keywords that we're going to incorporate into our
prompt in the next lesson. When I look at these, obviously, I'm seeing a lot of purple, I'm seeing a lot of pink. There is glow. We have city lights. I see abstract shapes, smooth surfaces, and those are basically things that I see. If you're struggling with coming up with keywords
from your image, we can also use the
help of ChatGPT. I'm going to use this image
because it has the city, the lights, and
maybe this one too. I think I'll do
these too actually. Pick two images that combine all the elements
that you're looking for. Open ChatGPT, upload
the two pictures. Like that and then
type in extract. Let's say ten keywords
for me to use in my Midjourney prompts
from these images. Here is my list. We can also ask Chachi B
to give us more or less. It really depends on how
things go within Midjourney. These are my keywords. I want you guys to now look at your moodboard and pick out the same keywords or in the same manner that we
picked our keywords. Be sure to notice
the color schemes, whether it's warm,
beige, emerald green, pastel pink, the materials that are involved
in your pictures, such as oak wood, marble, linen, And of course, the ambience. Is it calm, luxurious, playful? Because these words, while they may seem a little
random right now, they're actually going to
become the DNA of your prompts. It's really important that
they reflect your moodboard. If we go over my list, we have the ambience. We have color. This one, two, we
have some materials, technourban, immersive
atmosphere, that sort of thing. And I could come
back here and make a collective prompt from these things just so that I
could add it to my moodboard. That's not going to be used
as one of our base images. Let's ask HachiPT combine
these keywords into one. Let's copy this. Once again, I'm only doing this to
populate my moodboard. Paste it here version seven. And I have my
personalization turned on. You can also turn
it off like that, but I'm using my
global version seven. If you have other profiles,
it's going to show up here. You also have your moodboard. The more populated
your moodboard, the more accurate your
results are going to be, and that's why we have to
use something like this. This is using my profile. I'm going to do the same thing, but without my profile, just to see what difference
we're going to get. This is all of those
keywords combined into one. It's not really matching
the colors I was going for, but that's okay because we
can work on the colors later. This is with my profile. You can see it has a lot of that story element
and then of course, the city on the outside. This one is a lot different. It's not consistent because
it's without the profile. Whereas these ones
look pretty similar. Now, let's go ahead and
go back to our moodboard and simply add these
generations from our gallery. Let's go to the top
and just select all. Et's go back and there we go. We have a lot more
to work with now. For this lesson, be sure to write down your ten
descriptive words, whether you come up with them yourself or you use ChatGPT. Populate your moodboard
with at least five images. You can either use
Pinter, Google Images, or a journey like we did so
that in the next lesson, we can start creating
our base images.
24. Generating Base Images: Now we are ready to translate
these inspiration into Midjourney prompts
using our basic formula and all the tricks that
we have learned so far. Come to the create tab
and we're going to start by building something
called base image. Essentially, what that means is that you're going
to be generating these starting point
images and then combine and remix to
make your final result. Right now, we're not
really looking to generate the perfect
room with one prompt, but to put in the base concepts. The formula you should follow is room type, then the style, then materials, lighting,
ambience camera perspective. This is your formula
and based on what we have here with the keywords
that ChatGPT found for us, that should be pretty
easy to put in. Let's start with the room
type. Just type it in. Mine is going to be a bedroom. Then we're going to
put in the style. The style of my bedroom
will be futuristic, cyber punk, high tech. This is the style of my bedroom. Then we're going to talk
about materials and colors with let's say neon, purple and pink lighting
across the room. Then we can do coma. The bed is made of soft fabric and the ground is made of, let's say concrete. Now let's talk
about the lighting. We will do a little
light with the source. Let's actually
switch this around. Light source from window looking out to a techno
urban city scape. Light source from window, let's say with a view of
a techno urban cityscape, light and dramatic shadows. That's our lighting. Now
let's talk about ambience. We will say, I'm looking
at the keywords here, Immersive atmosphere,
highly detailed. Cinematic lighting. Then finally, the
camera perspective. We can do low angle, high angle, white shot or anything else. I will do a wide shot
with low camera angle. So this right here is my prompt. Once again, I will point
out to the base formula. The room type bedroom
with a style, which is futuristic,
cyberpunk, eye type bedroom. What's the materials
and lighting? This is material and lighting. Sorry, colors, not
lighting. This is lighting. And I will put this
guy in that section, somewhere around here,
cinematic lighting, immersive atmosphere. This is our ambience and
then our camera perspective. Try to fit your prompt and your keywords into
a similar formula. You can see from the prompt
that ChatGPT made for us, I'm matching the keywords. Futuristic Cyberpunk, I did
high tech instead of Sci Fi, I believe neon colors. I have the word neon in there, holographic interface panels. I did forget that
one. Let's try to do ground is made of concrete. We will do holographic interface
panels on the ceiling. Okay. Now I have
all of my keywords. We will generate this with
raw because I really want a high quality shot and I'm
going to turn on personalize, just click on P.
Going to settings, I want it to be a
landscape shot. Stylization. Let's keep that at default, weirdness variety. Let's bring this up at ten. I'm just experimenting here. Let's generate our first batch. You can see profile. There is my profile involved, and these are the rest of
the other stuff I have. Okay. I definitely did
the city concrete, the panels or there instead
of here. That's fine. Soft fabric for our bed and
it's just during the day. What I will do is remove that day component.
I really like this one. Let's put in the same prompt. I will put in at night. Let's try that one more
time with the same thing. Chaos, AR, raw version
seven, and our profile. I will also do one without my profile just so I can
get different variations. This one looks more
promising, not this one. I really like this. We
have the panel up there. We got some extra windows, and I didn't even ask, but there's plans here which
I actually really like. I will do a subtle and strong
variation from this image. Then we also have this one. I will try the same
thing, not upscale. Very. Then let's go to
our new sets of images. Believe should be this one. It's not completed yet,
but we're getting there. These are the variations
from that first image. Looks pretty good. Then
let's see. Lost track here. These are the ones we
looked at earlier. This is another variation list. I really like the ceiling here. This is our older variation. This is another variation, I believe this should
be the subtle one. Lots to take into account here. We even have a sunset one,
which looks pretty cool. I think we did one
without the profile. I'm trying to find
which one that is. One has profile profile. I think I didn't
make one. I didn't make one without a
profile. That's fine. There's a lot to
choose from now, but what you can also do is make separate components and then we're going to merge
it in when we remix. The bed right now,
I don't like that it's not made, to be honest, although I really
like the fabric, a silky situation, and
there's fluffy pillows. What I will do is mention a silk material bed with say holographic non
glowing ring around its frame. Soft pillows. And we'll do maybe
an elegant one. It's nicely made and everything. We can do the same thing. AR 16 by nine, all
version seven. Turn that off and
experiment a little bit. We will also mention, actually, let's try this and then
build up upon what we see. Other than the bed, I
like the atmosphere, but I do want more
tech stuff around it. Maybe a desk here, shelf there, that sort of thing. Maybe especially with this one
where we have two windows, we can do some stuff over here, like a TV, something like that. Here's our beds.
Looking pretty nice. I don't like circular beds. Let's do a rectangular. That's the only thing
I would change. Did I mention the color? I didn't mention the
color with elegant. It's too white silk and put in the same stuff run that because these
are just not fitting. Maybe this guy, but it's a
circle and I don't like that. This one is actually
pretty cool. Look at that. The rest don't
really have a frame, but I really like this, so I'm going to do some
variations of this in both ways, just to see how Midjourney
applies its own creativity. These are way out of that environment. These
are looking good. A second thought if I want
to do a holographic blanket, it would be nicer if
it's not made so you can look at all the folds and
the reflection of light. I'm going to keep that and then we have some more down here. Let's see which one
I want to keep. The shadows here
are a lot stronger, I think I'll go with one of
these guys, perhaps this. I'm going to heart this so
that I can find it easier and then go back to the
environment, the actual room. That's going to be
a bit more tricky, but I definitely like the
multi window situation. Let's choose from one of these. Here we have some plants. Don't know what that is.
I think that's a couch. Let's take that out, heard this, and then look for another room that has the elements
I'm looking for. Actually, this one looks better. Let's unhart the other one. There we go. Just because of that plant in the
corner, this guy. Then with the other ones, it's more purply, but I certainly like the
panels underside. So this is from
behind the bed frame. We have another bed over there, but this will be
interesting to remix too. I want this part of it. Then from the bed,
we have our bed. We have the windows,
we have the floor. Then finally, maybe we
can come up with a plant. Let's just copy this
hoops, not this one. We can work with this. Futuristic cyberpunk
plants, maybe. Instead of bedroom house plants. Neon purple and pink lighting across the room can
remove this part. Not sure what plants, but let's see how Mitrini interprets high tech plants.
That would be interesting. I guess what I'm going for
is the high tech element to be in the pot rather
than the plant itself. Okay. This one looks okay, but let me be more specific
regarding the pot. Let's say, copy this
house plant with neon purple and pink glowing rim in a futuristic
cyberpunk high tech pot. The leaves catch the pink
glow from a light source. Then our favorite things. Tags. This is a lot more accurate as to
what I was looking for, especially this one, not this, but it looks like a smart
plant has that rim. The pot looks good too, the flower, the plant. I don't know what plant that is. But this is looking good too. I'm going to grab both of
these for easy reference. Okay. When I go to organize
filter through my light, this is the things that we have. We have our first potted plant, second one, the bed, the window scenario, and
the high tech ceiling, and then the side
panels that we've got. So in the next lesson, we're going to use
the editor and even so the remixing
image reference capabilities to
combine these images together and go for
that perfect room. For this lesson, your exercise is to create these base images. Make sure you have the exact element that
you're looking for, because when you
remix them together, you don't want to leave that much room for hallucinations. So if I didn't really
include the plant, I would have been stuck
with that situation. Half of it is dried, the other half is alive, and that's not really
what I'm going for. Make sure you're happy with your base images because they're going to determine the
final result for you.
25. Combining All the Base Images: We have multiple generations and our base images are ready to be combined into our dream room. Midjourney is not
always about finding that one perfect shot
immediately after prompting, but it's more about layering
the iterations that you get. So let's go to our
organized tab. And just to remind you all, these are the images
that I finished. The goal here is to have the window facing
that urban city, the tech ceiling,
this particular bed, this plant, and then
this side panel wall. When you are choosing from
the different variations, make sure you choose the one that captures
the most detail. You may have to create
variations like we did here by choosing the images that have
the highest amount of details and are more close
to what you're looking for, you avoid the hallucinations that may affect
your final results. We're going to first start by combining the images
here and then take it to the Edit tab
where we can fine tune it and erase
things if need be. First, we're going to add the hearted images
to our image prompt. Just click and track them. You can do multiple and do
the same thing with the bed. The reason why we're putting
it there is because we want the structure of all of those elements and only keep the main bedroom image
as the style reference. This guy, I really like the way the bed is positioned and
of course, the side panel. I will put it once in the style reference and the other one in the image prompt. Then our last
picture was this one for the windows
that are available. I have five images for my input image prompt and
then one style reference, we're not dealing
with a character, but I'm going to
just drag the bed in omni reference because I want that holographic design
to be maintained. I have it currently at 100, the default, we can move
it up and down if need be. Now, as for the prompt, we're going to start with the original prompt and add the elements that we're
inputting as images. Mine would be a futuristic
cyber punk me hi tech. Bedroom with neon lighting, and I'm just describing
the overall effect. So here's my prompt. I had to expand it with the details from each
of these images. For the bed, the plant, the other bedroom picture. Once again, we started from
this particular prompt, which we got from JGBT. You can see it
starts the same way. The only difference
is that I've added the different elements
that I've added here. So the glowing
holographic interfaces, take panels covering the left
wall, just like this image. Holographic bed shimmers
softly. That's our bed. The room has raw concrete
and walls and floor. That's this image.
Massive window reveals a dazzling techno urban
cityscape view at night. Then this is about the city. Ceiling is an immersive
starry sky projection. So that one is one of the
images down here, that guy. The entire scene
radiates a moody, immersive cyber punk aesthetic. I'm just describing what
I see here with a balance of sleek Sci Fi design
and lived in details. That's the bed not being made, since we want the details and the glow to be
even more dramatic. My colors are purple and
blue based on this image, but I think I will do magenta and blue for
this pinky effect. All right. Let's
do dash dash raw. AR 16 by nine version seven. Then I'm thinking
ten chaos Generate. We're going to rerun this
again for some variety. You can see all the
inputs here for images. So far we haven't
done more than three, but you can see that
this is possible. Here's my first set. We have the bed, for sure, the ring, the city, the plant. There's some resemblance of it. These are some other things. There's our high tech panel, there's our plant, and
this is the other rerun. All of them look pretty good. Now the plant isn't
that visible. We can't really see it, the
glowing effect, at least. What you could do
is import this into edit and replace it
with one other plants. I'm going to choose this one and then another one to combine even further. This looks good. This looks really good,
and this one too. Once again, put it
all in image prompts. So this is our bed combined. This looks really good. I'm going to heard this
for sure. I like this one. It has an anime look to it too. All right, so we went 6-2. Let's do one final remix. Same prompt, send it in. Here's my final result. You can see it combined
everything very well. Here we have the messy bed, but it also gave us
some made up beds. So you just have to decide
which works better for you. This one gives me more of a
hotel vibe for some reason. But I really like
this one because the high tech panel is on the
wall rather than a screen. So I think what I will do is vary this and I'll do
the same with this one. It's not that bad. I just
don't like the screen aspect. With this one, we're getting
the actual hotel thing, which is a random
chair and a TV. But I want it to be
more less hotel. Here's our first variation. Let's go with that
one. Second variation. We even have the AC over there. These ones are okay. This one looks really clean, even though we have the TV, there's still this
high tech panel on the left, so
it's not all bad. We have our base workork. Now I'm just going
to upscale them. Let's do a creative one and then the same thing
for the other image. There we go. Here's the first look,
definitely a lot cleaner and it has
a lot more detail. This is our second one. We just have to decide
between the two. I think it's hard to choose. I actually went back and upscale this version
just because this looks a lot less realistic and that's the
goal we're going for here. And it's giving me
that square window. It looks like we're
in a spaceship, which is pretty cool. We have that silky holographic
bed all messed up, the left panel, the
glowing around the bed, and of course, the ceiling
that's right above. What I'm going to do
because my journey wasn't able to put in the
other ceiling that I like, I'm going to add in that part from another image and combine
it using the edit tab. Let's take this
base image which I upscaled with the
subtlety subtle version. We're going to edit it,
delete all of that. We don't need that. Clear this whole
thing, this guy. Then we're going to
add our other layer. Let me just download
the other image. Go to create the image that
has the glowing ceiling. I think I want to
do stars rather than a technical ceiling. Take a look maybe this guy. Let's download it and bring
it in onto the first layer. And now using the arrays brush, we're going to remove everything
else except the ceiling. Because we had that structure
from the first image, you can see it fits really
well with the image below. Just make sure to go on the first layer and create some space between
the two pictures. There's some room for your
journey to combine and remix. You should see this checker
thing between the two photos. Now we're going to
add in our prompt. I will do the same prompt just so it can
continue the colors and the atmosphere
onto this combination. With this image, we are getting
that smart glowing pot. You can see the light
coming from inside, and that is looking pretty cool. I will go back and
remove this further. Let's actually move layer one, which is the top layer. Somewhere here, try to
match the first image. I'll go to restore
and bring back the original trying to see where the frame from the
bottom image starts. Let's put this away for now. Use the restore
brush to bring back the square frame and then
fit this guy in like that. Maybe here and then
do some cleanup. Trying to match the corner
with the corner of the room, and then be sure to restore as much as you can from
the bottom image so you don't get weird
symbols in between. Get rid of that a little bit. Okay. Let's run this again and see if we get a cleaner result. That looks a lot better. The first one got
this random light, but the rest okay. All right. So here is
my final bedroom image. I will upscale it to gallery. And then we're ready
to do our next step, which is to either
create a motion from it. You can even add a human subject or a cat in there if
you want to do that, export it to the ideal
size based on where you want to share this and do any other final adjustments. There is my final room. Now let's do our
final adjustments. I think I will add a spaceship here and maybe
another plant on the corner. Just to further refine it. The reason why I upscaled
it first was to see if Midjourney can clean up some of the edges for me before
I add on anything else. So we're not remixing it with
any sort of imperfections. I will see you guys in
the next lesson where we finalize our dream bedroom.
26. Refinements with Photopea and Exporting: This is where we left off
in the previous lesson. I just want to add a spaceship, a smaller element over here and perhaps a
plant on the side. Let's generate the
two by going into create and putting in our last
image as style reference. Since we have the
style reference, you can see the
spaceship looks pretty good and it looks like
it's from that setting. The next thing I'm
going to do is again, put this in as a style reference and write about the plant. I think the best one is this. Since I'm trying to put it in the distance and not bring
that much attention to it. This one could
look good as well. It's just the moon may
be hard to blend in. Let's heard that and
then our house plant. This one's pretty good. Let's
heard it and go back to our edit tab where we had all of the previous things we
made in the last lesson. Add the two images as a layer, I'm going to have to
download them first. And then just upload them. Now using the same resize
and painting capabilities, we're going to fit the two new images onto
our current landscape. I think I'm going to
put it over there. They spaceship could
be somewhere there, and rotate it a
little bit and maybe have it go a little behind
the edge of the window. Now begin erasing the excess. If you click and hold, you're not going to accidentally go onto the background layer. I'm just using my cursor here so that's
pretty easy to do. Now onto the background, just delete those extra bits. Then submit the edit and
let's see what we get. If you saw that it's
not blending that good, the spaceship is looking better, but the plant, we're going to have to erase
more from the edges. This is what we're left with. I ended up making the
spaceship a little smaller and put it inside
of the frame only. Plant had to be moved
and rotated as well. I'm going to upscale
this one more time. Once this is upscaled, I'm going to bring it
into another program where I'm going to work around with the lighting a little bit. So here's my futuristic room. The problem that we're
facing right now is that the spaceship looks like it's inside of the room
rather than outside. So we're going to work with
the colors, as I mentioned, the lighting, and the plant here needs to have some shadows. These type of smaller edits you can't really do
within Midjourney. Hence why I mentioned
Photoshop, Canva, and Photopea. I'm going to be using Photopea since it's similar to Photoshop, but you don't have
to pay for anything. So photop.com. Let's import our image. We're going to first grab
the background layer. It's okay if this looks a
little bit intimidating. You can just follow
what I'm doing right now and then you'll get
the proper results. Right click on your
background layer, duplicate. We have two of the same thing, and this is just for
safety measures. We may not even do anything
on the background layer. Go to the adjustments down
here and choose the levels, lock it onto layer one. Clip it, and then
simply play around with the sliders so that the
Spaceship is a lot more matted. There's less shadows, less
attention towards a spaceship, and that way it
matches with all of the buildings. Now
go to layer one. Click on this mask icon.
It should be white here. If it's black, you can just hit Command or Control I to
flip it the other way. Now it should be
black and not white. Grab the brush, this tool right here
with the color white. If it's black or something
else, just flip it. Using the white brush, we're going to right click. And reduce the
hardness to around 35, increase the size a little bit, go over here for the flow and
reduce this to about 40%. Then on this black shape and
not on the image itself, we're going to start
painting our spaceship. Then to right click and
make it a little smaller, click away and start painting. It's going to look
a little weird, but we're going to zoom
in with this tool. Then grab this layer, take the opacity and lower it. Then going in with the
same brush, right click, lower the hardness
all the way to zero, and then switch the colors
so you have the color black. Then we're going to lower
the flow even more. We're going to start
cleaning the edges so that it looks a
little bit more natural. This was before, this is after. Now let's go in with a
Zoom tool, right click, fit the area to zoom
all the way back, and we're going to repeat a similar process for the plant. Grab layer one.
I'm going to call this spaceship table
click on layer one. Grab spaceship, hold down Command or Control.
Levels one as well. Command or Control G to
group them together. That will click spaceship. Now we're going to duplicate
background one more time. Right click duplicate layer. I will call this plant. Make another levels adjustment. Use the Zoom tool to go in and then the space bar to click and drag so you
can see the plant. Now with Level two, we're going
to make the plant darker. Then simply clip this onto
plant only using this icon. Grab plant, make a mask. Command or Control
E to turn this from white to black with
the same brush. I have 27 right now. Let's do 0% hardness, choose the right size, flip the colors so it's white, start painting the edge. Then go in with the
color black if you want to clean things up a bit. Lower the opacity.
This was before, after just a slight shadow. Fit to area, close this off, and then I'm going
to group these two like we did
with the spaceship. Now let's do some
overall color grading. You can add more adjustments to these singular components
within the image, but I think I will
leave this as it is. I do want to however, blur out the window completely so that it has
more of a futuristic look. Let's actually duplicate the
background one more time. I will call this glow. Bulter blur, Gaussian blur. Add a lot of glow and then wait, actually, let's combine all of our edits and then do that. Hold down Command
Option Shift E or Control Alt Shift E on Windows to make a merged image of
all of your other groups. This is the thing, this layer, the same thing
as the ones below it. Let's call this final. Grab your final layer. We're going to duplicate it. Go to filter blur, Gaussian blur and add quite
a lot of blur, maybe 16. Then change the blend mode, which is normal right
now to lighten. You can duplicate glow for
more intensified effect. Then grab both and we're
going to merge the layers. Change the blend
mode one more time, and then click on
this mask icon. But before you click,
hold down Alt or option and that should create
a black shape for you, which is an inverted mask. Grab the brush with the color
white, increase your size, make sure hardness is zero, and then start painting
in those lowe areas. My flow is still 27. This way, I get to build
up as I draw more. This is before the
glow, this is after. Then lastly, we're going
to do that overall design. Go to the half circle, then choose one of
the color lookups. This is the fastest way to give your image a nice coloring. Maybe not that one. It's hard to tell which one's going to work since there's
no preview here. This one's not that bad.
I'm just going to lower the fill to make
it less intense, gives a subtle blue hue
and then we're going to add another adjustment just to make the image
a little brighter. Take the contrast lighter
and increase that too. And there we have it. Now, all you have to
do is go to File and then just export this as
your favorite format. I will do a PNG for
the utmost quality, but you could also try a JPEG. Give this a name. Format. You can also change it here. This is by default the
width and the height. Keep quality at 100% and clicks. And there is my room. Now it's your turn to
export your final image. You can skip the whole photo
P editing if you'd like, but I think it's just a great
way to finalize your work and do some tiny
adjustments that you can't normally
do with Midjourney. Export it to the format and
the size that you want, have it ready for whatever
reason you created this for. You can also use other
platforms like Canva and maybe video platforms if you choose to go
the motion route, then export it as
we already learned. This was our first project. Now we are going to
apply what we learned in this project to another project that's going to be a
little bit more elevated. In this chapter, we learned
how to build a moodboard and extract descriptive prompts from one image or multiple images. Then we learned how
to generate and refine the room designs
or the base images. Next, we blended and remixed all of those
images into one, and then finally, we exported
and edit it using Photopea. This process took you from the initial idea to the
final presentation, and that's exactly how you will tackle every creative
project with Midjourney.
27. Project 2: Your Fantasy Character - Brainstorming and References : Something. For this project, we're going to be focusing more on a character
rather than a room. So we're going to be
following the same routine, which was first to
brainstorm the idea, then create the base images, edit them in Midjourney and
some external platform, then finally export it and put it on something
you want to do. The goal of this
lesson is for you guys to imagine your
fantasy character. This could be a superhero, a villain, a warrior,
whatever you want to do. We are then going to build that character and
put it onto a poster, and then that's going to be a different media that you've
made using Midjourney. First, let's think about
our character archetype. Is it a hero, a villain, mage warrior,
magician wizard? You have to think about what
does that character even do, and then you can start thinking
about the way they look? You can just go
over to Pinterest or the Explore tab in Midjourney and start
searching for them. I want to do something
like a warrior. I to search here, warrior, and I'm getting some
pretty cool designs. We're going to make
a moodboard with the same process that
we did last time. Whatever image speaks to you, just add that to your moodboard. Let's actually go over here
and create another one. Call this my fantasy character. Just as we did with
the dream room, try to think outside of the box, things that are really
hard to do in real life, because that's going to give you more creative outlook and you can work harder on combining those different
elements together. Don't just go for
a medieval night, try to go for a
medieval night in outer space using a sword made of stars,
something like that. Once I named my moodboard, I can go back to the Explore
tab and continue my search. So these are just your
regular warriors. Let's try to put in
another keyword, such as moon warrior
and see what comes up. I definitely want mine to have a glowing aspect and this blue
mood to the entire image. Now let's go to Pinterest and start looking for those
different components. First, let's look for the same
thing, see what comes up. I'm going to choose AI here. These are all made with AI, and they will work better to input for our own generation. This one looks really good.
Again, it's made with AI. Make sure that you're
seeing this tag before you use it for your own
Midjourney generations. Now let's look for
our color palette. Blue moon, color palette. Say blue moon AI
landscape, maybe. And we're gonna use these
for our style reference. And just like before,
we're going to go to moodboard and start
uploading those images. Now to populate this even more, I'm going to make some base
photos using the Crea Tab. Let's close these off. I just made this by accident, but you can see just
one word gives you some pretty decent results
in different styles as well. I will do a female
moon warrior with glowing eyes and a shiny
armor with glowing edges. Com, she has white hair. Something simple like this. I'm using ten chaos in the 16 by nine aspect
ratio, version seven. Okay, this one
looks really good. I love how realistic it is. So one looks fine, too. So what I'm going to do is
use this as a reference, put it in the same put
it as an omni reference. So that's our character,
and then type in. Et's try the raw style. And don't worry too much about
the subject's face for now because we could just make a separate image
just for the face. But here we're trying to
think about the armor mostly and then the position she's
standing. She has right now. So I wanted this half turned position and hence why I
had my prompt like this. There are two others with
her full body visible, but I think I prefer
this half body. Going to fix the
background and everything. I kept in smoke because
she would be easier to extract from the background when we try to combine
everything together. But you can see this was we used this as our omnireference,
and it's still her. She's just in a different armor and in a different position. Let's get some
variations with this. Then I'm going to
use the same image, same prompt, but I
want her hair to be open and maybe more
glow in her armor. Here's our first variation. Again, her hair is tied, so I'm gonna wait for the
other generations to come too. Here we're seeing some
sort of open hair. There is the glowing
armor, which I asked for. She has an elf ear for some reason, but we
can get rid of that. I'm liking these
armors a lot more. If you could combine maybe a hair like
this with this hair. She even has a glowing hair
clip, looks pretty good. I think this one's the best one. Let's get some more variations and then I'll do another
image input prompt. Same prompt. But describe
her hair as more smooth. I think the best one
was the first one that we made right up here. Go to favor this, and then we're going to think
of the other elements. What's the background like? What is she holding?
What's her weapon? If your character
does need a weapon? Then what edits we want
to do to her right now? I do want her eyes to glow because I'm not going
for a human character. But before I do any
changes to her, we're going to start
with the sword itself. Let's say a sword
made of the moon. We can do fantasy elements, and then dash raw. This one looks pretty good. These two not so much, but it's definitely
unique and I like that it added though the plant
wines that we asked for. Let's create more variations and see if we could get a more
realistic look to this. Let's see our selection. But I think this one
has more of that glow, but I did like the
moon hanging from the end as if it's giving
power to the sword. So we will combine this image with the sky
because I like to tip here. So let's put in
both as referenced. You can see the harded ones, makes it easier to choose, and then put in the same prompt, dash dash ra and then a
hyper realistic sword. Here we kept the moon inside the sword and then the rest
don't really have this one, the moon is in the middle again. But I think this one
looks pretty cool. I'm going to add this as
our input and do the prompt one more time with the dash
dash raw and Kos set to ten. I think this one looks
the best actually. The moon is put inside, but thinking of it if this
were to be used in battle, if it's hanging there
may be hard to grab. I think this one
makes more sense. Once again, we're
going to favorite this so we can put it
into our moodboard. Now the last thing
is the background. Right now we have smoke, but that's a little
bit too simple. Let's think about where this character is fighting
or where she's standing. Can do a mythical forest with concrete spaceships
flying around. There are wines
coming from the sky. The surface of this landscape
is made of the moon, maybe. I'm going with that moon thing. There are glowing cracks in
the concrete spaceships, as if they were
strung by her sword. But spaceships, high details,
hyperrealistic fantasy. Dash, dash a. Again, if you're
having a hard time imagining that landscape, you can ask HachiBT to
give you some keywords. But you can just start with a really simple
prompt like this and then build upon it as you move forward with
your pace images. This one looks really good. I'm going to keep that. I think we're good with our
inspirations right now. Let's go over here,
ad from Gallery, filter with the
favorites or the, grab them all and put
them into our wood board. So this is what our
character looks like. We have our sort, different versions
of it, environment, the coloring from these images, and we could even combine
this pace with this body. The possibilities are endless, but these were our
inspirational shots. Feel free to add more using
either the Explore tab, your own generations or
Pinterest AI content. Just make sure you're grabbing the AI modified ones
and not actual artwork. This concludes my moodboard. You guys should
populate it right now before we move on
to the next lesson so that we have an easier
time generating the finalized based
images so that we could put them all together
into one final shot.
28. Base Images and Blending: The first thing I want to
do is take this background shot and apply this
particular style to it. So I'm going to go into the
same photo from the creative, put in the same
prompt, structure, and then put that image that we got from Pinterest
as style reference. I want to do one with
the golden one as well. This is the first set with this particular image put
in as the style reference, and we're putting this
as our image prompt. It's following that structure, but with this blue color. This is with this image
as the style reference. We got that golden
color in there, and this is looking pretty cool. I really like this one
because I want to do some sort of depth
of field edits later on so she could
be standing somewhere here and then this stuff is just following
her in the back. This one looks
pretty good as well. Not a fan of the blue one
because it looks too normal. That was my base, and now I have every color I want within
this particular shot. I will put that in as a
style reference and then use our subject as the
image prompt. This one. Put in the same prompt, but just change the
smoke background. Let's delete the
background part, go up here and see how we
refer to this. There we go. Then I want to put in the
same image as Omni reference because I don't want Midjourney
to give me hairstyle, another character, but
just keep her consistent. So I put in the same prompt, but this one as
the Omni reference and this one as the
style reference. Maybe we can try this as image prompt and
then this is Omni. I'm just playing around and seeing which variation
works best for me. This is our first set. Not the best. Let's take
a look at the other ones. And this is looking a
lot more promising. She's standing on the side and
there's all our spaceship. The moon is a little tiny, but we can fix that in
the edit type later. I'm just going to take
this image and make some subtle and
strong variations, and then let's
upscale it as well. The next thing we have
to focus on is putting the sword in her hand because right now she's
not holding anything. While that's happening,
I'm also going to do a full moon, a
hyperrealistic one. Full moon with great detail in the night sky and then use this guy as
our style reference, and I will do raw mode
one more time so that we could merge that moon and
put it somewhere in the sky. This is our first variation. This is our second variation. She's more zoomed out.
This is our upscale, which is completely off. Changed my character entirely. We're going to disregard
that and this is our mood. Honestly, it looks great, but I want to make
sure it's isolated. Let's use this image as the style and then put
in the word isolated. We're also going to
use the no commands, no clouds, no stars. That way I can
separate it easier. Then going back to our subject, trying to see which one looks more like our original subject. I think this one
looks pretty good. I like the depth of field, which came about without me having to write
anything, to be honest. She has a little cape situation, nothing in her hand, but
we're going to fix that. The glowing armor, and
I'm just going to upscale this and then heard it so later I could just
bring it into my Edi tab. All right let's take
a look at the moon. I still not isolated. So I will try the same thing, but without any style reference. Maybe we can reduce
the reference. Let me try that S ref, which is style reference. Let's put it onto ten, and then I will do one without
the reference entirely. This is our upscale shot. When you put creative, it messes up the subjects face. I'm going to go back to
subtle and then hard this. Now we're getting an
actual isolated mood. This one, we use the
style reference command. For some reason it's red, but this one looks better. I will go with this
yellow one to fit into that glowing design we
have with our character, and then all that's left
is to put in the sort. Let's go to the Edit
tab and we're going to start by taking
that first base image. Go to your liked images. This was the normal version
and this is upscale so you can see the detail
coming through. I'm just going to
edit this likes and start importing the
different elements. We can also do
this with the URL. When you go to your liked ones, choose the sort that you wanted. I'll take this, and then we can copy the image URL and paste it here instead
of downloading it. I'm going to repeat
that with our moon, which we didn't actually save. Let's go to the Create tab. I don't know where my mood went. There we go. Let's
take this one, copy it like we did. Put it in so. First, let's start with the moon because it's the
easiest element, and I'm just going to lower
minimize the sword for now, bring this guy and put
it somewhere far back. Then we're going to use SmartSelect because it's
a very obvious circle, grab it and then
erase the background. Then on the base image, we're just going to erase a
little bit from the exterior. I will use the
mouse wheel to make a perfect circle right in
the middle. That's my moon. Let's take the sword and position it where
it needs to be. I'm going to rotate
it, put it in that. I don't know how big
we want it to be, so let's go for a size like this and then clean it up
like we've been doing so far. Now onto the prompt, we have to mention the
sword and the moon. Regarding the sword,
I will probably have to apply the coloring from our base image onto the sword
with the style reference. But let's try it like this and then we'll see
if we need to do that. So let's remove the full height with her back turned slightly. In the picture and
we'll start here with her smooth shiny hair is long
and is flowing in the wind. She has a fierce look
on her face and is holding a magical sword, I guess I'll call it, period. The background is a
mystical force with concrete spaceships
with glowing ones. The ground has
glowing components or elements all around faded
or blurred in the distance. Her armor has growing lines all across and it's made up
of a smooth element. There is a full moon in the sky. Now, before I submit this, we're going to ask ChatGPT
to further refine this. Further refine this,
prompt for Midjourney. Let's copy this and replace
it with our prompt. I will just make
this to show you and then put in a new prompt. This is our first set. As I imagine, the sort is
not being blended in at all, we may have to go back to well, we have to go back
to the Crea tab and do another
remix, this method. Paste the ChatGPT prompt, put in your parameters, and then we're going to put in the sword base Image prompt. Put this in as well, actually, let's get
the sword first. There we go. Put
that in as well. I think this guy is
calling for the moon, so I'm just going to
add the moon here too. Style reference. Going
to try one without any style reference and then
do one with the base image. Put in as well. So here it's completely
changing the image. We got a completely
new character, so that's not going to work. The sword is not even
like the one we made. Here we're getting again the same outfit,
the same location, same moon event,
but the sword is just not what we asked for. Let's see what the third one is. I think it's having a
hard time understanding the sword that's over here. This one, we have the
moon, location, subject. But again, the sort is just
not what we asked for. We're going to just
try the same thing, put in the input
images like we did. But I will add this to
Omni reference as well. At least I could get this part remixed and then if the face is all messed up, we'll
think about something. This one isn't that bad. We have this interesting weapon. It's not really a sword, but it's not that bad either. The moon is nicely blended
in the background. We have the same
glowing elements and trees, the concrete spaceships. This is not a
terrible generation. Now here we're
getting the sword, but our subject is
completely different. I think if we were able to add more than one new reference, that would have worked great, but that is not happening. I am going to attempt
blending this sword into the background
because now it has the same colors lighting,
and even backdrop. That may be easier to blend in. Let's take I think
I'll make a new one. Then get the other image
that has the sword and try to mix the two
together. This one. Copy the image URL
imported here. Now we're going to just resize, we don't have to resize
the sword, really. I will have to
rotate it so that it fits the subject's hand. Basically try to rotate the sort so that it's
right above the original. Run the same prompt. You can see that it brought
in all the elements. All I have to do is submit and hope that it can figure out the location and the position of the sort in the
subject's hand. All right. It didn't
really do that. Here we have a good
lending, this one. I will grab this
image and then erase the surrounding so that I could redo the parts
that I don't like. Use your eras tab, the eras brush, and go around
as closely as you can. So that it could
rebuild the background. Here is not that bad actually. I think the area it struggles
with is from this point, and then of course, this part. That's a lot better. We have different variations. Let's see. I added an extra spaceship
over there, which is fine. I do think that the way she's grabbing the
sword is not that good, so I will try to redo
that part as well. This one looks the best. I'm going to grab this and
just go over her hand. You can see it's the other
way around for some reason. Let's erase that bit
and then hope that it can give us a more
accurate grip. Okay, give us a watch
for some reason. That's not a good one. This
one looks pretty good. We even have the
glowing bracelet and then this one is
just your regular arm. I think I will go with
the glowing bracelet. And now we got our
sword put in along with all the other elements
that we started with. Let's upscale this to
gallery and just for fun, I'm going to do the
other one as well so we can have two options when we're doing the
final exporting. So this is our first one. It's looking pretty good. This is the one with
the glowing bracelet. The other one is the
one with a normal hand. But you can see the subject
looks the same, same sort. It's just that one arm. I think I'll go with this guy. We are now done with our final
image within Midjourney. In the next lesson,
we're going to edit things and maybe
put it onto a poster. We could put some text in there, some color grading and
just make this really mystical fantasy and sci fi. So try blending in your base images like
we did in this lesson, and then you're ready to edit it further using either
Canva or Photopea.
29. Upscaling and Refining with Topaz AI and Photopea: We now have this combined image. However, it's lacking
a lot of details. Even though we upscaled it and tried different variations, you can see that some of the
elements such as her armor, which is supposed to
have a lot of detail be shining is looking bland. Same thing for the sword, the hair is probably the worst part because it just looks like a
bunch of lines. And this is where we get to
use a tool such as Topaz AI, which I mentioned a
couple chapters ago to upscale that image
using AI as well. We're taking the output from
an AI tool like Midjourney, putting it into another
AI tool to upscale. Head over to your browser
and look for Topas labs. This is where you should end up and you're able
to try it for free. Of course, not every capability is available for you for free. And there are some limitations. However, the one
that we're trying to use is the upscale one. Upload your image.
This is the original. What you got to do is go to the left side where it tells
you the different models. We have three models available. I'm going to go with standard
two because we're trying to enhance detail and that's
in the description here. We have upscale factor times one times two, all that stuff. Preserve the pace, make
sure you turn this on. We don't want a different phase. I just had to sign
in and you can see that I have ten
left on my free plan. Let's render and see
what result we get. We're doing this before any color grading because
the more detail we have, the better the outcome
is going to be. When we apply a brightness
or contrast setting, it's going to be applied to every detail that's
visible within the image. Whereas if we just exported from the journey and
started editing, we were going to put that
effect on a blurred surface. Here's the original side and
this is the result size. I'm going to upscale
this to full screen. You can just see the
before and after. There are tons more
details on our subject. Going to go back.
Exit full screen. Let's try to do
another version where we increase the
sharpness even more. Let's try 88 and reduce the
denoise by a little bit. We should get another version. This is where all your
versions are going to end up. Then you decide
which one looks the best. This is Version two. This is before, this is after. A lot more sharpened effects. I'm going to try zooming
in for you guys so you can see Choose my cursor. You can see how the smaller
details are brought in and it's not too
sharp, which is good. Except the full screen, and now I'm going to
just download this. When I open it, you
can see I don't get any watermark and I'm just
left with a better image. There are also other
tools you can explore. Let me go to the homepage. We have Bloom, which is basically
the AI tool themselves, the AI tool for this
company, we have blur. We have faces, lighting,
Sharpen and upscale. We were in the upscale tab. Now I'm going to
go to the Sharpen and basically
further refine this. Let's go to 50 and I'll do
selection for everything and see if we're going to get guess more detail on the
little lights. This is before, after. This is a lot better. I think I will
download this version. Let's download it with this icon in the corner
and now I got Image five. That's how you get to
upscale with another tool. Let's open it up in Photo P. I really want that
glow to be visible as I keep saying it just
because I'm going for that moonlit mystical
fantasy environment. You guys can also change
the entire coloring here. I might do that with a more cyan color rather than this deep blue and we'll
just see what happens. First of all, always duplicate
your background layer in case you make a mistake and you have to go
back to the original. On layer one, I'm going to grab the hue and saturation
from the adjustments. This will be applied
to the entire image. If you grab the slider here, you can see how you're able to change the hue of
the entire thing. I wanted to go towards the side. We got that red
color and the cyan. Then this is regarding how
deep those colors are. If it's too intense, just
lower the saturation. Then this is
regarding how bright or how dark the color looks. I will make mine a little
darker, negative three, just to intensify the
contrast between the glow and the rest of the
image before, after. If this is way too intense, you can always grab the fill and lower it until
it looks better. That's regarding the color. Let's make another one for
curves and play around with the highlights and the shadows will increase my brightness too. By making points on this line, this side is the highlights, shadows, middle
is the mid tones. It's going to be a mix of both. If you want to
increase the shadows, you just grab this guy and the
same thing for highlights. Highlights is the brightest
part of the image. You can see the moon
started changing and the shadows is the majority of the image because
everything is dark. I'm going to increase the midtone brightness
by dragging this above towards the
highlights just so we're able to see some
of the smaller details. This was before or after. Let's make another one
just for the moon. Go to levels and increase the far right slider to
intensify the moon effect. If you want, you could
also limit it to the moon only but I think I
like it this way. But if you wanted
to make a mask, remember you can do it with
this button and then use your black and white brush to introduce or remove the
effect from the image. Now that I have my coloring
and my lighting done, I'm going to merge
everything together into one layer so that I could
add that glow effect. Hold down on Mac Command Option
Shift E on Windows that's Control Alt Shift E. All at the same time until
you see a new layer, I will call this
the combined layer. So on the combined layer, we're going to make a duplicate with Command or Control J, and then we're
going to grab this, go to Filter, Blur,
Gaussian blur. Increase this quite a bit, okay. Better yet, we should right
click and convert this to a smart object so we can adjust
the Gaussian blur amount. Now go to Filter Gaussian blur and you can see we're
getting this sub menu, and when we dab a click on it, we can come back here
and adjust the radius. 24 changed the blend mode to ten and now we got pretty
cool glow effect. If this is too much, you
can grab the gaussian blur, double click and we
have our slider back. I could lower this now that it's on this blend
mode so that we don't introduce the
blur onto her face or any other detail that
is supposed to be sharp. I'm going to make a mask, hold down alter option and
click once on this icon, grab your white brush. Make sure it's soft around zero hardness adjust the
size of your brush, click away, Zoom
into the areas that need the glow and
start painting. I will reduce the flow to 50 or 40 and start painting away. The more you go over that area, the more you're applying from
that gaussian bleur effect. That's why we lower the flow. I'm also going to
zoom in on her armor and try to only get the lines
and not so much her armor. Zoom in and adjust the size. It's too small,
something like that. And then start painting away. Use your space bar
to move the canvas. This is completely optional. You don't really
have to do this. It does take a little time, but the more you spend on your image,
the better the outcome. We're going to do
that for our image. Okay. Now we got this
really cool glowing effect. If you want to intensify it, you can hold down Command or Control J to make another copy. Now, what I want to do is further play around
with the colors. I do want the glow
lines from her outfit, her armor to be a
different color, she doesn't blend in
that much with the trees and the mystical creatures
that are glowing. Let's make a gradient fill, and I'll change this to color, which is all the
way at the bottom, and then change the gradient, the white part into the
color that you want. Let's try maybe red. We can always come
back and change this. Forgot to change my
save my adjustment, it okay, and we can switch it the other
way, so it's more on her. Click away and then flip this mask so it's
black and not white. Command or Control I, get your brush with B, color white, low flow. Zoom in with your
scroll wheel or just the ZoomT and
go over the glowing lines.Th I will lower the fill here, and then the flow as well. Just so I could get a
natural color change and not a really strict one. Now we're getting specs of pink. I do want to experiment changing
the color of her armor. Maybe that's better
a better decision. Yeah, I think that one's better. Go to quick selection subject, and now we have the
subject selected. If it's missing a few parts, just hit Q on your
keyboard and basically make sure that your subject
is visible and not red. If it is red, you just
have to use your brush, black to remove, white to add, and just clean up the area. For example, for me, did
a good job in my case, but there's this little part where I believe her pants
is not in the picture. If you hit X on your keyboard, you're going to switch
between black and white in a more efficient way. Hit Q again and we're going to while the marching ends
around the subject, make sure you're
seeing this line. If you are not, that means
you cancel your selection, you would have to choose
the subject again. Now, turn this on your
gradients map, go on it, make a mask, and now you have that pinkish color
on your subject only. Problem is that we
got the hair as well. Using the black brush, we're going to just
go over it like so. To zoom in and make sure we're not adding it
in any other place. I use the backslash, the left bracket key
and right bracket key to increase the
size of my brush. If you know Photoshop and
have used the Adobe version, then this should be pretty
simple to navigate. If you haven't, you just have to follow the shortcuts
that I'm telling you, and then you can also use it a lot quicker and
more efficiently. There is my subject. We change the color
of the armor. If you change your mind
regarding the color, you can always dub a click
on the gradient shape, go to here and just switch
into something different. To lower the intensity, just go to the fill
and lower that until it looks good to you.
There we have it. The last thing I want to do
is at the glowing eye effect, which Midjourney, we actually forgot to do it in Midjourney, but the good thing about Photopea is that it's
basically Photoshop, so we can just do it all here. Her eyes are weird
to begin with, so this will be a good cleanup. Go to the adjustments,
color fill, and choose a bright color. Let's go with cyan
and then go to Color. Go to the mask,
Command or Control I, hit B on your keyboard
with the color white and just introduce the
color inside the eye. While I'm doing this, I might as well clean
things up a bit, move gradient below color, grab everything,
including layer one, Command or Control
G. Table click, call this color and and body. Anything outside of
it is for the eye. Now we're going to make
a exposure adjustment and increase that a lot. Commander control
eye on the mask, brush, lower the fill, and bring that into the eyeball. To hide this weird eye shape, I'm going to hit
X on my keyboard and reduce some of the
lighting from the corner. Same thing with the blue. Her eye socket looks normal. Do the same thing like that. Okay. Now she has blue eyes. We do want it to glow. Let's make another adjustment. I will do levels this
time and grab the edges, reduce the contrast by
grabbing this middle slide, closing this
Commander Control I, and B X to get white, increase the size and make sure your hardness
is set to zero. Lower the flow around 24 and
just start brushing around. If the image is too intense,
just lower the film. One last thing is that
I'm going to go down to adjustments,
human saturation, but we're going to
clip it to levels two, which is that brightness
area we just made, and then grab this
so that it's blue. Just look at the exterior, we're going to fix
the eye later. I think this is nicely done, increase the saturation and
keep the lightness to zero. Now, use your black brush
and just exclude the Well, we can just invert the mask. Command or Control
I on this guy and then change the
blend mode to color. With the white brush, we're going to just introduce
that color to the edges. It's not just a white glow. And there we go. Now she
has glowing robot eyes. If you want, you can do the
same technique we did before. Command on Mac Command Alt
Shift E, all at the same time. Let's Control Alt Shift E on Windows and do the same
gaussian blur trick. 7.2 pixel lighten. But we're just going to make a mask with alter
eruption held down, so the mask is black and just introduce that glowing
part to the eyeball. To make it a little bit more
dramatic, lower the film. This is what we started with and this is what
we have right now. Now you get to put
this onto your poster. You can put it into a mockup and that's something
we're going to leave for the next lesson just to keep this one
a little shorter. Make sure you are happy with your final edits with Photopea
and then we're going to just export it and
put it into Canva for some poster design or maybe
just put in a few texts.
30. Making a Poster with Canva: First, let's export the image. Go to File, Export
as JPEG or PNG. Quality 100, I will
give this a name. All right. Then exported, I went over to Canva and
looked for posters. Search and find
something that you like. Let's remove this
middle section because that's where we're going
to put our subject. Double click, get
rid of the text, upload your image, and then just bring it onto the Canvas. Then you're going to
click once and use these sidebars to crop
the image into the grid. I'm using a template here
that was made by this person, but you guys can use the blank Canvas over here
and make your own designs. Let's go back in and make a
duplicate of this picture. Click once, Command
and Control D, and then just put that
right above the one below. With the top layer selected, click on background remover. Now you should get the subject separated from the background. If you don't have Canva
Pro, which is this guy, you can do the same thing in Photopea just by
merging everything. Let's right click,
merge the layers, go to Quick Selection,
select subject. You have to watch
a five second ad, but then you're able
to use it for free. Processing my image
and there's a subject. Then you can just
click on mask and hide the background so
that you're able to export this as a PNG, which means it's going to preserve that
transparent background. Hit safe and then
export that into Canva. But if you have Canva Pro,
you can do it this way. Both options are available. Now, we're going to
grab our text tool, get a heading and write
something like a big bold text, say a new down, maybe. Command or Control A and
just increase the size, can grab the edges here. I'm going to choose a font
that is more futuristic. Let's go here, go to headings and experiment
with these guys. That one looks fine,
but I'm going to go for something like this. Maybe something like this.
It looks futuristic. Then go to this option and reduce the space
between the lines. Put it in like that
right in the middle, then go to effect. And let's do a hollow, just the outline and then
I'll do commander control. Let's actually reduce the
thickness to maybe ten, Commander control D on the text to make a duplicate
like we did with the image. Make sure this is positioned
right above the old one. Then remove the effect
from this top one. One has the hollowness,
and the other one doesn't. Now push the top one all the way behind your
separated subject, hold down command
left bracket key. Until you see
something like this. Now we have our text. I'm going to grab
the background layer and the second one that
holds the subject, right click and just
lock them in place. I'm able to edit the text alone. We forgot to do that
to the background. Go to position and
then I'm going to grab this and lock
the background as well. Now I should be able to
move both of the text, this guy and this guy. Something like that. This is
a cool effect you can do. The style, I'm
going to switch out the colored paper background for something completely white and then change the
font for these guys into something more sand serif. It looks a little normal. We could even do Canva Pro. Honestly, anything that isn't that vintage style,
it's going to be fine. We can keep the logo. I guess that's the logo. Maybe we can try something. I did see this font.
That looks pretty cool. Let's lower the size
so they both fit in our screen and switch
out the text and colors. Nothing that has the sera font, which is the handles
that you see. We want these bold funky text. Play around with
the fonts until you find something like Okay. Then this guy is out of nowhere, so I'm
going to remove it. What we could try is having a
accent color so we can grab the shapes or maybe the outline from the
bolded text, this guy. And then go on to maybe
even the fill color. I'll go with this
yellow color rather than the previous
white one and then change any shape that's on the poster into the
same yellow color. I think you have to
do them one by one. Then paste that
onto the other guy. The fonts will now
have to be black. You can copy the style and
then paste it like that. Everything is related. Same thing to this guy down here and I'll leave
the stars as they are, but I do want to copy
one with Command or Control C and paste
it down here on top. So click drag alter
uption shift to scale it from the center and make sure it's actually
in the center. Now we got ourselves
a fun little poster. I'm going to just
move her a little higher so that it's not
cutting off as it did before. You miss this guy. Let's switch that to the same font
as the ones down here. And it's looking pretty good. You can play around
with the fonts and all Canva has a lot of
presets that you can use, so you don't have
to start anything from scratch unless
you really wanted to. And there is my fantasy poster. We made this image
from scratch in Midjourney and then
brought it onto Canva. Well, first to Photopea
and Topaz to do some refinements and then onto Canva for our
finished poster. That concludes our
second project. I hope you guys
were able to make your fantasy characters using the same methods that
I've been showing you. In the next chapter,
we're going to be doing some applied projects, things that you may be doing more often if you were
to use Midjourney, and I'm talking about real life scenarios
such as product shots, children book illustrations,
and then some more posters. So that's going to be
even more of a hands on experience to make sure
that you're ready for that. After that chapter, we will
look at how to actually monetize our creations
using Midjourney.
31. Mockups: Brainstorming Your Brand : In this first project, we will be creating product
style mockups that you could sell on Etsy or
use it in your own shop. Mockups are basically
placeholders for logos, designs, and different
artworks to be pasted on. Right here, we have some
examples on Midjourney. You can see they're
basically blank shapes that are meant to hold
those additional arts. It could be a t shirt, a card, a book, a board, and there's so much options out there when it comes
to building mockups. We even have a coffee mug. Now on EDC, when
you search mockups, there's a lot of competition. So it's important for
you guys to figure out your niche and what
area you are best at. If you look around your room, try to find one object
that is very common. For example, if you're
really into candles, you can start by
making candle mockups. If you're really into books, you can do book cover mockups, shirts and just see what is something you
know a lot about. Another thing to point out here is that everyone
is selling things in bundles because that's the
best way to sell mockups. Either 100 frames, 150, even smaller numbers would
work like 600, maybe 50 only. It really depends
on how much work you put in when it comes
to making that mockup. If you're doing it
with Midjourney, then you can easily export 100, 500 with just the same prompt. So we're going to
basically tie in AI as a form of replacement
for photography studios. So let's say that there is an artist out there
that has this product. They have this great logo,
this great wallpaper, but they don't have the studio to actually take
those good pictures. So naturally, they would come on platforms to look for mockups. That way, they just have
to go on Photoshop or some other tool to paste their artwork directly
onto that mockup. The important thing
about mockups is the client being
able to blend their logo into the
different lighting, textures and lines. Here, it looks
perfectly natural. We have different skewed angles. We have something
blurred out since it's in the distant, see, we have different colors, working well with the
background mockup colors. If the logo is orange, it has to look well on
this green background. You as the seller
will only be selling the objects themselves and the client will be
putting on their logo, which is these text
and these shapes. So let's go on Pinters
and figure out what sort of niche we want
to go with, first of all, this lesson is only
for brainstorming, so you can take some notes, make different boards,
different moodboard, and figure out what
works best for you. Let's go to Pinterest. If you search for mockups, you get a lot of options. There's digital mockups
like screens, laptops, there's foods and beverages,
clothing, tote bags, cards, business related objects, so it's a very diverse field and you have to figure out
which one speaks to you. You can look into
handmade niches, home decor, fashion accessories, business tools, and much more. But I think the best
way to trim down into that one niche is to see
what's most common around you. That way you won't
spend too much time going over every
mockup out there, trying to see which
one looks good to you. Next, you need to have the
demand driven thinking. You really do need
to do your research on Etsy and see what
is selling well. For general purposes, at
S and other platforms, a lot of clients go for
ready to use mockups, meaning that all
the client has to do is open your file,
paste the logos, and it's already blended
in as a correct lighting, their colors work well on your object and they
don't have to do much. Try to think of, we're going to go through this
together, of course, but it's important that
you advertise that, hey, if you download my mockup, you can just use any of these platforms and paste
your designs easily. So you can see a lot of them
have your design here or a sample to show what the
pasting would look like. This is an example. They're showing
them how they can paste their own artwork on
different programs too. You can make a Google
Doc or just take out a notebook and write down the
stuff that are around you. For this lesson, I'm going
to focus on sweaters. I will have to look
on Pintras at C and see what examples are
out there for me. Let's search sweater mockups. And we have some examples where someone is
wearing the sweater, some were the sweater
is just hanging there, one without a background,
one that's folded, one that's on a
colored background, and a lot of other examples. I think I will go with the
half face worn sweater so that they can see the sweater style and then
make their decision there. Because if they were to
put it on their website, this would look a lot
more appealing compared to this because you can't tell how this will
look like on a person. The way that object
is being portrayed, that's going to be your aesthetic
in terms of the layout. You still have to think
about the lighting the colors and keep that consistent throughout
all of your works. Because if you're going
to be a store yourself, you should only be
focusing on, let's say, sweaters on humans with
golden hues, light coloring, or maybe you want to do studio lighting with cold
colors and full body. Maybe something like
this where we have a concrete background,
it's very streetwear. That way, you're
going to attract artists who are
designing this thing. Use the same lighting
and background style. We already learned how
to use reference frames. So we're going to in this
section of the chapter, work on that perfect reference
shot so that we could reuse it every time we want
to put out a new mockup. Next, we're going to learn how to export for print on demand. There's going to be a lot
of different aspect ratios depending on the platform and
the object that you choose. For example, if you're
doing mobile wallpapers, you have different iPhones. They each have a different size. You may be doing a four to
five ratio or a four to three, and that's something
you can easily change within Midjourney. You can upload that
image that you made and just switch
out the image size. References are key here and you should really
be comfortable using them both for image
prompt style reference and Omni reference. Okay, let's start
with our sweater. I'm going to be building
a moodboard using only Midjourney
generations because these are not made with AI, so I do not have the
right to use them. But it's good to just
get a picture of what prompt you'd
like to put in. I will do, as I said, half body minimalistic
looks, such as this one, really basic background, some neutral colors
for the sweaters, and my models are always
cropped at their nose. Which if Midjourney
couldn't figure out, you could just crop it
yourself on Canva or Photopea, which is eventually where
we're going to head anyway. So don't worry too
much if there's extra stuff in the frame and you can't crop it within Midjourney. Let's go to Explore and we're going to look
for mockups here. But now that we know we
want to do sweaters, we will put sweater mockups. So I'm getting a lot of just empty sweaters without
anything around them. But when I scroll down, I can see the exact thing
that I want to use. Now, since this person
designed it and kept it in the public domain
or the Explore tab, you can reuse their prompt. You can see Midjourney
just lets you do that and I can see who
made it right here. Let's put in this prompt. The most important part
for me is it being realistic because we're trying to convey that this
is a real person. Let's look at the
prompt, got a mockup. I will do a brown sweater. Or maybe Ell. I'm going to switch
things up a bit. The key term to use here is stock photography style
because that's going to give you the cleanest layout for any product.
Let's go over here. I will do a portrait shot, raw for sure, stylization. We can set it at the default. I'll just reset this actually, let's turn this off,
start generating. While that's happening, let's go to moodboard and make a new one. Here are my mockups. We got some good poses
in different locations. Now, this one looks pretty good, not a fan of the tattoo, so I'm going to put the same
prompt and then use this as my image prompt and then
put period, no tattoo. I think I will remove the text. We can just add that
on a different tool. E Shell sweater,
sweatshirt, remove this. And then just make another set. I'm also going to
think of colors. Let's try to think of maybe
explore here or something. I will do something autumn. And then see which color
speaks to me the most. This might be
different for search whatever color you want to use. I think this is a
nice color set. I can heard it so I
can use it later, and this one's pretty good too. You can use multiple
style references, so just heard whatever works. I course, make sure
that they're somewhat related in terms of
shade and aesthetics. There is my new selections. I kept tattoos on them for
some reason, but again, we're going to use photo P so you can just disregard
the little errors. I will do one without
the image prompt, and then once I find
something with this, I will add in the colors and we can just fill
up our moodboard. Right now we are not
creating base images. Forgot to remove this part. Let's go to Explore, scroll the way up and go to likes and then paste these
in our style reference. Then we can put in a
very simple prompt, maybe autumn aesthetics, and then see what my
journey gives us. Now we don't have the text. There's no tattoos and we got some different
styles of sweater. That, I will add this
to my moodboard, as well as some of
these since I'm just trying to think about
where my model is, do I want her on the
plain background in a different location
in her bedroom? Where should I put her? Then what is this sweater for? Is it something
they wear at home? Is it for going outside? Should they be
wearing formal pants, jeans, this is the colors. It combined all of the elements and the colors into one shot. This will be easier for me to put in as one
style reference. We went 4-1. Okay. Let's go to Moodboards and start
adding from Gallery. I will add the ones that I like. I'm just making sure to exclude the full face because that's something we
decided early on. I've got the
different locations, some with the text, some without the text,
and then we just go back to see our
populated moodboard. At the same time,
any object that can add on to this
moodboard, you can make. Let's go back to taking one of these since
it has all the colors, I'm thinking we should
do, candles and coffee. Make sure it's realistic. And just pop away with whatever
that comes to your mind. If you're doing
something techie, you can add different devices. If you're doing something
in the beauty category, you can put makeup, perfumes, and all
sorts of stuff. For me, I want to just stick
to the sweaters and not so much the pants and then
candles and coffee. There is my results. So now I have a very
nice moodboard. It even added some
dessert options for us. We got that nice some flowers. The wall is great,
love the color combo. And that's the exact mood that I'm going for with my mockups. So now that we have
our moodboard, we're going to move
on to creating the base images and
then from those images, we will combine them into one, something that we did in
the previous two projects, but we're going to be
having the model do different poses so we can have at least five different
mockups to offer. Let's move on to
Building deeps Images.
32. Mockups: Generating Poses: Now, let us create
our base images and then try putting our model
in different angles. That way, we'll have
four different poses to provide our clients. Then we're going to
take it onto Photop to create that PSD file where they would just paste
in their designs. First of all, let's
use one of our colors. I will go with one of
my coffee candles. Let's do this one, put it
in your style reference. In terms of Omni reference, look to the models you have
and upload them there. I have the default 100. We could obviously play
around with it once we've made our first prompt. Let's type in a good prompt. We want an ultra realistic
mockup of an oversized. Let's describe the sweater
that our models are wearing, oversized
Crock sweater, let's describe the atmosphere
soft neutral tones, modeled on a torso and hip only. Just the fact that I do not want her face in
any of these shots, we can do parentheses. No face cropped just
below the chin, more directions for mid journey, and then the sweater
has the folds. That's really important with
visible folds, ribbed cuffs. Then of course, I think I'll do a relaxed casual
sweater, so relaxed fit. Then we will do the models pose. The models hands are casually
tucked into front pockets. Let's talk about
the lighting next. Soft natural lighting with subtle Shadows, minimal
clean background. And basically, we're
just describing the shots that we see below. Now, our first pose is the model tucking their hands
into the pockets. I will generate this with raw, make sure that's available, and then choose
your aspect ratio. Let's try the first one and then see what remixes we
need to do next. If you're going to be doing any model mockups where
there's a human involved, always start with
a pose that's to the camera and then go
for different poses. If you want to do a back
pose, a sitting position, don't go for that
first because we want to first imagine what
the model looks like. Standing normally and
then have her or him move in different angles
and positions. Here is my model.
Looking pretty good. It does have a lot of similarity
to our omni reference. What I will do is
use the same prompt, increase the variety for sure, and then we're going to go to new reference,
add that on there. Instead of 100, I
will do 25 because I want to try
different hairstyles, different models basically. Then for style reference, if this is too much,
we can do SRF. Let's do ten. It inter. Now we got all of these things. I'm going to try one more thing. Let's do style reference 50, just bumping it up
and then more chaos. I'm just experimenting
with the parameters to figure out that ideal
first pose for my model. Now you can see that we're
getting different pants, different hairstyles, and
it's still the same sweater. That's the most important thing. I asked for a relax
fit Kroneck sweater and that's exactly
what I'm getting. T is what they all look
like. One thing that I want to do is
increase more detail. Let's say that the sweater
is made of cotton, I will have to mention
the word and ask my journey to really focus
on the cotton details. This one looks a lot better. I love this one. Yeah,
this one looks silky. But this one we're getting the fabric texture,
which is perfect. Here's our first neutral pose and we even have the
folds that we asked for. This is definitely a winner. I'm going to put this in
and use this image as our image prompt because
this is our pose. With a pose, you want to
put your reference as an image prompt and at the
same time as a omni reference. Omni reference to 25 and I'm going to add
in these colors. Let's try reducing
the style reference since last time the
colors were a little bit too potent and I'm going
for more neutral color. Maybe 30 should be good. From here, I could
tell that it's already working on that texture, so there may not be
a need for me to mention cotton or
fabric texture at all. But if for yours is still silky if you're doing some
sort of clothing, that is, just add in the word cotton and fabric texture so that you get the
exact same lines. Okay, so that is my first pose. I'm going to just see
which one looks better, and then don't worry too
much about coloring. I'm getting a
little bit of blue, which is not from
my style reference, but I could fix that
within Photo P. So once you have
your first look, look for any sort
of hallucinations, and then we will remix that
out with the edit tab. Once you have that photo
that looks good, go on it, zoom in and make sure to spot any hallucinations so that
we can clean that up. For example, here,
something is up with the neckline and
the necklace looks fine. But here we're getting something
else with the neckline. This one looks okay and then
here we get a weird fold. First, let's choose the
one that we want to use. I like this pose. Let me just check her
fingers real quick. Okay, her fingers look fine. I will go with this picture. Let's heard it, and then
bring it up to edit. Keep the same prompt
and reference. All you're doing right now is redoing the portion
that needs fixing. For me, that's going
to be her neck area. I'm just going to zoom
in so I could see it better and a little
bit here, I would say. Scroll down. If you missed
anything, just add that in. If not, just submit that edit. It's going to regenerate that
small section and try to blend it in considering the style and image
prompt references. There's my edit,
just go to view all. It's hard to see, but you can see the neck line
has been fixed. This was, we can't
see it before, but it's a lot more smooth, especially this one,
so I'm just going to upscale it to
gallery. All right. That's my first pose
and we're going to use this guy as our omni reference
now for the other poses. In total, I want four
mockups with the sweater and we're just going to switch up the prompt
a little bit. The core elements
are still there, but we're going to just maybe describe the way she's
holding her hand, the way she's tilting her head. Whatever pose you want to do. Now grab your upscaled image. Going to just like it for later, grab it as an omni
reference and image prompt. There's no need for
style reference anymore and I'm going to
copy the same prompt, but change a few things. Adjust this part,
this line, basically. Instead of her putting
her hands in her pockets, we're going to choose
a different style. Let's do the model is standing. In profile with one arm
hanging naturally by the side. Minimal background, change
the lighting as well. I do suf directional
natural lighting. And then add on the same
stuff for some variations. We're just going to
put raw version seven, no Sf since there is
no style reference. Let's see what pose
we're going to get. I have two more, but I'll
see what I get here. I may reduce it to three, but it's basically
the same process. You're just changing
that one sentence. If you want to do maybe she's doing the piece sign
or she's looking down, maybe she's facing the camera
at a three quarter angle. A pose like that will work. So here we got some
nice sweaters. This one looks really cool. I'm going to look for any
sort of imperfections. I don't see any, so I'm
just going to upscale this for when we
export to Photop. Next pose, we're going to put the same prompt omnireference, image prompt and
simply do let's do, where's that Crop
just below the chin. Let's remove this part since
I want to do a back post. The chin is not even
in the picture and let's redo this part until here. Since now I'm doing the
complete opposite direction. So detailed view
of the back yoke, which is the neck part, ma relaxed, drape,
thin line texture. Same lighting and everything. I'm going to add in some
decorative slides just to tell the client what
mood my mockups have. Choose one of these
decorative pictures. I will choose this one,
let's download it. There is my backslide.
My back pose. Looks pretty good. Let's
try some more variations. I'm not a fan of the neckline. It's a different color, but I could fix that in
photo B, so it's fine. I do want to try another pose. Let me first look at these.
This one looks okay. Let's upscale this. Then
while that's happening, let's do a crossed arm pose. Let's switch out for this one. The model is crossing her arm. There is my upscale shot. Download that. And let's see if we could
get one of these. This may be hard to do and we have to look out
for hallucinations, but still worth
experimenting with. Okay, we got some pretty soft
poses, which is perfect. She even has a ring here and
it's the exact same sweater. I think I will go
with this ring one. Let's upscale this and now
I have all of my poses. The next thing we're going
to do before taking it to Photop is once
again upscaling our photos with Topaz AI because we mentioned
about the importance of quality and we want
to just upscale this further and enhance
those fabric details. That way, the client's
logo or design is going to fit in perfectly and naturally
onto these sweaters. Zoom in and make sure
her fingers look fine. I can see some weird thing
on one of her fingers. Let's edit this
and use our brush. Zoom in here to redo this part of her
finger, and maybe this too. Submit the edits. That
looks a lot better, let's upscale this
one to Gallery. And then download them. Now I have four pictures. I'm just going to show
you them one more time just so that we're
all on the same page. Let's go to organize. There are my poses. We have this one, this one, this, and we did three.
I thought we did four. Let me go back here. We forgot to add this one to the list. These are the poses. Pretty good. I'm going to
remove this yellow one and then I think this
is the one that we didn't fix yet. All right. In the next lesson,
we're going to go to Topaz AI, upscale these guys, and then put them onto Photop, where we're going to create
a template for that client. We're going to fit
in a sample logo just so that people can see how their logo is going to
fit onto our sweaters and then provide them with
the finalized files. At this point, you should have your four different images. You can do three
and depending on what niche and subject
you went with, it may look a little different. If you're doing humans,
focus on different poses. If you're doing technology, focus on different devices. If it's soda, try
different bottles and give some variations
to your packages.
33. Mockups: Upscale and Adjustments : We have four poses
for our model. I just downloaded them and now we're going
to upload them in Topas labs where right now I have seven left
on my free plan. You guys should have
something there too. If you finished your plan, you could just easily upgrade
with the button right here. But since I have the space, I'm going to start uploading
my poses and then upscale. Let's start with pose one. We're going to go with 50, make sure everything is
selected, and then render. I'm doing sharpen and not
upscale because I already set the aspect ratio and I
upscaled it in Midjourney. What I really want
is more detail, not so much the
size of the image. Let's do a before and after. This is before, after. We got a lot more detail, and I'm just going to download. Let's repeat that
with the other ones. I got all four of my images and notice how they were
exported as PNG, which is a high resolution file, nothing is compressed here and that's perfect for Photopea. What I'm going to do is import my image, the upscaled one. You can also click and
drag just like I did here. The first thing we're going
to do is play around with the colors and I want that to be universal with
all my other poses. Just like we did with
this first image, drag the other three onto this canvas and they're just going to pile up
on your layers panel. If you're not seeing
the layers panel, just go to Window and make
sure this is turned on. Now I'm going to name my poses. That's really important
because the client doesn't really know what Image
seven or image eight is. This will be neutral front pose. This is going to be
arms crossed pose, three quarter tucked hands pose, then we got a pack pose. Let's go ahead and first of all, get a nice cooler for
all of these guys. I'm going to convert
all of them to a Smart Object so that
I'm able to edit as I go. Right click on each of the layers and convert
to Smart Object. Not much is going to change, but you can see in my history, which if you're not seeing
it, again, go here, it tells me that it's
creating the smart objects. That's going to take
a little while. But now I get to add
in my different edits. Let's start with our
neutral front post. Go to image. I think adjustments is better. What I want to do is play
around with the first of all, lighting and then
maybe the colors. To add a curves layer, you can see it goes below because this is now
a smart filter. And we're just going to increase
the folds a little bit. Bump up the highlights and lower the shadows to create more
contrast, then hit okay. Is before or after. If it's too much, you can go on this little icon on the side
and just reduce the opacity. Now you got a little boost. On the same note, I'm
going to go back to adjustments and let's see, perhaps add a
vibrant adjustment. Lower that a little bit. The yellow is too
intense for me, so that's why I'm
just going to reduce it to some degree.
Before or after. I want more of a
neutral minimal look, not so much summer
wipes or something, so not too yellow. Now when you go to
the adjustment, it's going to apply the same previous thing
that you made. Now I can just put
in 177 and then 189. The same numbers from
the last adjustment. Like, repeat with the others. Then of course, we have
to change the opacity. So 177189, same number. You can also make
a custom filter, but I'm doing it this
way in case I want to switch out certain things
for each of the images. If you see the same numbers don't look good on
one of the images, you have the option
to switch it out. Then we're just going
to reduce it to, I believe we said 55 55%. That will click, type in 55, hit Enter, and repeat
with all the other ones. Oops. Next is vibrance. We added negative 28,
negative negative five, which should be saved
from over here, but if it's not, just
type in negative 27. You can use tab to quickly move to the next part,
the next lighter. Usually on the real Photoshop, you're able to just apply your latest adjustment with the same numbers
onto your image. It's a lot quicker than photo P, but this is still fine. We're just typing in numbers. Now all of my images have
that styling set for them. I'm now going to start
putting in that sample logo. Let's hide everything for now. Just click on the eyeball and I'm going to
grab the text tool. Basically for each of the post, I will do a separate graphic so that if my client
has a full pattern, one line of text,
maybe a quote, a logo, they can just imagine how that will look based on my
four different designs. The first design, let's
make a new layer. We're going to click
once and let's, of course, increase
the size here, put in your your logo, Command or Control A, change the color, so it's
easier to see. Then we can choose the font. Because I went for a
more fall aesthetic, I'm going to choose
a Sans a sera font, which means it has the
little handles on the edge. This is the style of the font. I will do a bold 150, let's put it in the
center, hit the checkmark. Grab the move tool and just
put it right in the middle. This is going to be
put your logo here. We're giving directions to our client once you've
made your logo, your fake logo, you
can right click and convert that
to a Smart Object. The reason why we're
doing that is because if the client wants to just
replace this with their logo, they can double click
opens a new file. They can delete this layer and let's say they drag
their own logo, they can simply replace it. I'm going to show
you how that works. Let's delete that. But always put directions for your clients. This logo, let's put it on this guide because
it's the easiest thing. Let's put I'm going
to close these. And just focus on the first one. Ideally, you would want to have a separate PSD for
each of these poses, but I'm just going to do them in one so that you guys can see how I do how the process is
for each of these poses. But once we're done,
I'm going to have one PSD file for just a neutral front
pose with the template, another one for this one, and basically four PSDs need
to come out of this project. Let's put that over here and we're going to go to
the neutral front post. First, let's put it
where it needs to be. Woops. Let's lock
the base layer, put the lock on it so we
don't move it by accident. Grab the logo and position
it on a nice spot. I'm going to put it right
in front of the folds. That way we can see how the blending is
going to work out. We also want to clip
this onto this so that the client doesn't
accidentally put it here and then they
lose customers. Command or Control Z, just put back the logo. We're going to grab
this base layer. Let's hide the logo for now. While I'm at it,
I'm just going to make the three other projects. So you can right
click and duplicate that into a new
project. Same name. Just do it like that,
and there's my image, and I could just
delete it from here, so things are a lot more
simple for you guys. New project, same name. Delete it from here,
one more time. Okay. Now you should
have two layers, your background image
and your template. Go to your background image
and using the selection tool, just grab the subject. Just click on this guy. I okay? It's going to give you an ad, but at least it's
free. Close that up. Normally, you would use the negative brush to
remove from that selection, but for some reason on Photopea it just jumps out completely. I'm going to hit Q and use the old fashioned way
which is with a brush. If you guys have Photoshop, I would recommend
doing it there. Oops it to get black and remove her face
and any other detail. Going to increase the hardness. I'm only going to be doing
this sweater template once because the process is
going to be the exact same. What we're doing for this pose you would be doing
with the others. You're basically just
selecting the sweater, removing any other detail and putting a template above it. So I'm going to
clean this up and then I'll be back
so we can continue. Here's my final selection. Everything that's not the
sweater needs to be red. So make sure you take your time with the white and black brush. A hardened brush
would be better here because there's a lot of
big folds in the sweater. When you're done, hit
Q on your keyboard and we're going to
make a new layer. Command or Control
J on your keyboard, we'll make layer one, which is just a sweater. Let's call this sweater. Now that we have our sweater perfectly taken out
our upscale sweater, I should say, it's
now time to make that template for the client to just drop in their designs. Even though we have
four different poses, I'm going to show
you the four designs on this project only this pose so that it's a lot more convenient for
you guys to follow along. But ideally you
would want to set one design for
each of the poses. We may just copy
paste it in the end. Let's see how that goes. Let's move on to the
next lesson where we're going to put in the
designs and first of all, build the base for a clean and natural
design on the sweater.
34. Mockups: Templates and Export : Okay, right now, you should have your surface
separated in Photo P, turn that into a separate layer. So you have your base photo
and right above it is that canvas that the client's design is
going to go on top. Once you have that selection, just right click, convert
it to a Smart Object. And then we're just going
to blur the surface and turn it into a
displacement map. Grab this layer, go to
filter, blur, Gaussian blur. And you can see it's only
affecting the sweater. We're going to blur it out a
tiny bit, maybe five pixels, hit okay and then
we're going to save this as PSD to go to file, save as PSD and I will call
this displacement map. Now just go on the same thing, hold down Command or Control. Click once on the
thumbnail to get the exact same selection and
just turn this into a path. When you click that, you
have this work path. I'm just going to
name it sweater. This is just in case
we need to come back and use the same selection. Now we get to delete this layer because we
don't need it anymore, and I have my design. Logo. Now, I'm going
to download some of the patterns we made in my journey for our second
style of graphics. Go back here and let's
remove the light. Choose something that would
go on top of a sweater. Maybe a cute little
flower design. I'm just going to
download it as it is. Obviously, you can upscale
it use any other platform. This is my second graphic. My third graphic, I'll do
a long quote like that. Let's grab T, click once and type in your
favorite quote maybe. Let's put quotation
marks and then a author. Command or Control A and
shape this like it's a quote. We're actntly making it
within the work path. Let's click click away from the path and make
our quotation now. Quote your favorite
quote goes here. Then author. Let's make this italic and position it where
we want it to go. I'm going to make
it a little bigger because I wanted
to stretch nicely. 75 is good. This is my quote. That's our second design. The third design is going to be the image that we
just downloaded. Just click and drag that on top. It's really tiny
for some reason. Let's make it full size so that it goes on top
of our sweater. Then do your pattern goes here. Then make sure that all of
your items are smart objects. Like this one, let's hide this. The last thing, I think I will
just quickly make a icon. In Midjourney, let's do cute pumpkin Baker icon with handmade vibes and tout style. I definitely like this guy. Let's download it. Very cute and put that onto our sweater. If our subject is really
into character design, this could be an
option for them. Then do your icon goes here. For this one, I'm
going to separate the subject from the
background. Got another ad. There it is and just
make a new layer. Oops. Command or Control J. Now we don't have
the background, you can delete that one and name this your icon goes here. Turn this into a
smart object as well. Let's start with our first logo. We're going to make
sure that these are all smart objects.
That's really important. Now let's go onto our paths and actually bring
back this selection. Command or Control
on the thumbnail and we're going to
make another copy. And you can see how easily I
can retrieve that selection. Now, clip this onto sweater, right click and let's
do a clipping mask. This is only visible within
the sweater selection. Going to lock sweater in place. You can see the logo goes out whenever I leave the sweater. Unlike Photoshop, this
doesn't let you bring in a displacement map. I actually opened it up
like this by accident. What you can do is just
duplicate sweater, bring it out of here. Call this displacement map. And then make this
into a smart object, add in the same blur amount. That's the only difference.
I'll just hide this. I go to your logo and
then try that again. Displacement map and
turn this into 50 50. I think it's too much. Let's
reduce let's wrap around. Then play around
with the sliders, 50 is obviously too much
for my little logo. Just try to match the folds. This is good for me, so I'm
just going to hit okay. Next, we're going to change
the blend mode to multiply. Then we're going to
go on the same layer, go to effect lending options, go to background,
alter option on this triangle and slowly
move it to the side. So now we took a
regular text and we molded it according
to these sweaters folds. Now, say you were the client and you want to
put in your logo. I'm just going to make
a quick logo here. Just take this guy, very
simple logo, download it. Now if I were the client, just tap a click
on the thumbnail here until you go into this new window and you
should only see that logo. Notice here, it's not
distorted at all, and that's exactly why we
made that smart object. Drag your logo here, resize it. Try to get there try to resize
it and I'm going to rotate it just so that it
fits or better yet, we could just put in
a different text. If I were the client, let's say I made logo, and let's use let's use
a text tool actually. I will do a whole
different thing, my shop. And I will choose a
really decorative font so it stands out
from what we had. Let's say this is
my client's logo. They wanted to put it in.
All they have to do is hide the base logo and maybe
put in a background color. Let me just choose
something Nice here. I'll go with a slight blue. Use the Bucket tool. Long click on this guy, you should see the
Bucket tool and just click once on layer one and then you're going
to hit Command S or Control S to
save this project. Smart Object updated. If you go back here,
there is the client logo, so it's fitting perfectly in that mold we made and they can
put in whatever they want. Now the same thing
applies to these guys, it's the exact same process. I won't really show
you guys these two, but I do want to show you how to do a full pattern
on the sweater. With this one, things
do get a little tricky because it has to meet
every edge of that sweater. First thing you want to
do is bring your design, smart object on top of
that sweater selection, right click and then
clip it to the sweater. Now we're going to displace
it the exact same way. Oops. Displacement map, wrap around and
then with this one, you can be a little bit extreme because you're dealing
with a lot more room compared to a small logo. Let's try 44. The little stripes you see
is the fabric texture, so that's completely fine. Hit okay and then change the
blend mode just as we did. There is my sweater and then to make it blend
in even further, you can go to blending options. Just grab the sliders slightly. So if it's too dark, you can just add a little color
to the same smart object. To go to it, image adjustments, maybe brightness
and contrast and just give it a little
bit of a lift. There is my full design. You can play around with the
fill if you want it to be a faded look or just go
full opacity. I'll do 84%. They just have to dabble
click and change the design. Let me make a crazy
look for this guy. So now that you know how
to do displacement maps, just copy the same process
for each of these arms, each of these poses, actually. Going to delete that one, close lose basically repeat the same template for
each of these poses. I'm going to keep
this full design, let's delete the logo and the other designs and to
hide things for your client, since they don't really
have to see everything, you can grab all three with Shift Command or
Control G. Oops. You can grab these
two, group them, call them background items. All they have to do is go
here and we can even right click and change
the color into red. That's the only thing
they need to be touching. Now to save this as the format, you need to be
giving your clients, go to File Export As, do a JPEC which is a compressed image
format or a preview, full quality, let's
call it preview one. And then a save as PSD. We got Pose one as PSD and JPEG. Those are the
things you would be uploading to Etsy
then of course, you have to give it a
title, a description, and all the other things. That's how you can make mockups using
Midjourney and then use tools like Topaz and Photopea to make it ready
for your clients. In the next project, we're going to take a completely
different route, which is how we can make Children's Book or any
other book for that matter, using Midjourney
again, GBT and Canva.
35. Trendy Poster: Concept and Inspiration: For the second project, we'll make a series
of AI generated posters inspired by
real design trends. The tie into this is that posters are perfect for print on demand services that
are available on websites like Red
Bubble or Society six. The first thing
we're going to do is figure out a trend
that we want to replicate and we can do that on some of the platforms
that I have opened. Pintris is, again, a great place for you to get some inspiration, see what's trending,
that sort of stuff. If you have your own aesthetic, you can go ahead and just um, continue that for the
rest of the chapter. But you can go for things
like retro 90s poster, witted K, maybe Boho
minimalism, whatever suits you. But if you don't
know what to do, you can just come over to one of these platforms and
do trending posters. If you want to use any of
these as a reference image, make sure you search for
the ones made with AI only. If they don't have the tag, then it's not made with AI, so make sure you do
not use those images. But you can still look at
these to get some inspiration, maybe see what these are called and then
try to replicate it. So this could be a
glass film placed above the subject's face with
the text engraved into it. So you would describe that in Midjourney to get
a similar result. Or with this one subject is placed behind a glass
jar filled with flowers. Her eyes covered by the
blossoms or something. Now, you can go ahead and
make another moodboard. I will be jumping straight into creation because I already have a style that I want to do. But if you don't have a style, feel free to do what
we did last project, which was to go over Midjourney, create some inspiration shots with those in a moodboard
and then proceed. Redbubble is another
platform that I will be explaining in further
detail in the next chapter. But if you look
for AI retronine, which is the sort of
direction I want to take, these are some examples. The most important thing, we already talked about providing bundles when it comes
to these outputs, and we're going to be
creating a bunch of posters, but the style and the
subjects need to be cohesive. What I want you guys to
do is first figure out the style that you want to proceed with and
then the subject. Is the subject
going to be robots? Is it going to be a human? Will it be an animal? Even an object would
work like a PC? Think about which
subject you want to focus on for the
rest of this project. So for me, I just took a
picture of my rubber duck and I'm going to be using that as the subject in
all of my posters. My brand will be
based on this guy. Then I'm going to look for a poster that I like,
which is this one. This one, I'm not sure if it's made with AI doesn't say it, but nonetheless, I'm going to look at the
features that it has. It's definitely a retro design. It has good text. There is a cyberpunk
aspect to it, like a cityscape as well. I'm just noting down the
different elements I see in this image so that I could
describe it to my journey. This in general, would be a
retro Sci Fi poster style. If I just look that up in Pintras I'm going
to get a lot more. And this just confirms that I've found the correct
term for the poster. If you see anything
you like, just go in it and try to find out the term. Got a digital poster. This is vapor wave Avis. So just note those down. So now what we're
going to do is first put in our duck and try
to make him a little bit, I guess, cooler since he will be going in
this sort of poster. Upload your image. Let's put them in the
Omni reference actually. Then we can look for the
style that we're going for. That would be a retro
sci Fi poster maybe. Whichever color or style you prefer just drag it into
your style reference. Let me remove poster. I like this style so just click and drag and you
can put more than one, make sure that all your
styles are somewhat similar. I do like this work too. I think I'll stop at three.
You can go for more, but try not to go more than four because it might
confuse the model. Now I'm going to describe
my duck in the environment. I think what I'll do
first is try to replace a subject in one of these
places with the duck. Let's see. We could the duck
go? Something like that. Then describe the duck here. A giant pink rubber duck. Reimagined as a cyberpunk icon loading above a Neon
Cityscape VitrociFiPoster, and then choose your colors. I think I'll do bold
red and yellow, maybe. And then submit that. While that's happening,
I'm going to go over here and look for something
cyberpunk too. These work, it's a little
bit too crowded for me, but there are some
that are minimal, like this guy or this. Yeah. I'm going to
do the same process, but with this tile and then
if that's something I like, I can combine them together. Going to see where my There
we go. A new reference. Then I think I'll
leave this out, but then go here for
the same prompt. This is what we have
from our first set. You can see it is
my rubber duck, but here they are part robots. We have the full duck
with, I guess, headphones. It maintained shape of the duck, the curve and all, but just
made them into a robot. If we were to put this
in the image prompt, we're going to get
the same position, and that could be
something worth trying. Let's put in the same prompt. I'll add this as image prompt, omni reference, and then add a mixture of these guys
as my style reference. Let's grab, this and
then maybe this. Here's my second set, which is meant to be a
lot more cyber punks or a lot more digital
stuff over it. It maintained the duck
really well here. Of course, if we upscale it,
it would be even better, but you can see the resemblance from the real duck
to the fake duck. If you create
different variations, let's do a subtle and a strong. We could get some pretty
interesting results. This guy is with my image prompt being that
original image. This guy. You can see how in all four, I have the same position and the same size even
amongst my ducks. The only difference is
that the background is techie and Japanese because I think we
added one image Okay. I think it's just taking
it from this image. This is the variations
that we made from the guy. So it's all looking pretty good. Now we're going to
upscale this even further and note down the
things that we don't want. Remember, you can
use the no command. For example, if you saw that there's attempts at making text, you can just put no text so
that Midjourney doesn't even try putting together the little text
that it's trying to do. Now I'm going to
upgrade my prompt. Again, this guy has
omni reference, and I'll increase this to
100, which is the default. Style wise, I will add in This method, I like the
flatness of this a bit more and then we'll put
in our prompting again. So I essentially describe my duck first what
it looks like, and that's from your
Omni reference. Whatever you put in here, just describe it for this first half of the prompt and then
reimagined as your style. If you're doing cyberpunk,
you can put it here. If you're doing it as
an anime character, you would mention anime
and just build that up. Then what is that duck doing? It's floating above a
futuristic neon cityscape, retro Sci Fi poster style based on this guy,
bold graphic design. This is bold and
then the colors. I think I'll do magenta
instead of red. Surreal technical dystopia. That's exactly what
we're seeing here, symmetrical composition, meaning that the duct
should be in the center. Detailed geometric
lines, high contrast. Now let's put in
some parameters. I will do SL reference
100, chaos ten. We will do Persian six
just to try it out. I'm going to try one without
a text and one with a text. Bold graphic design of the
text Duck nation maybe. Let's see how it does
with text in general. Version six, and then I will disable raw because
we're going for a flat graphic style rather than a real photography style.
Then let's send this. I will do the exact same thing, but then without the text so we can see which one
works better for us. Here's the first set. You can see that it tried to replicate the
position of the duck. And it does put in
dcnation pretty well. I wasn't expecting that.
Has it on the walls on the towers in different
fonts. It's pretty cool. Then here it doesn't
have the text, so that's the only difference. Because we're dealing with a
lot of random styles here, I think I will make a
moodboard and what you can do is actually
select it from here. Midjourney only creates things based off of that moodboard. It's more than just for
your inspiration purposes, but it actually helps
us make certain styles. So let's make another
moodboard and I'll call this my Sci fi Duc we'll add from gallery the
stuff that we've liked so far. I like this one.
The robot is cool, maybe this guy.
That's pretty funny. Add those in, and then I want to do some stuff
from the Explorer. We're essentially
remixing our work and works from the Explore page. Hold down command and click once, and that's going to open
it on a new window. Then just go over here and copy the image URL so that it's
easy to add to your moodboard. Just go over here,
add from Link, paste it and blue. This is too detailed
for us actually. This one's a better match. You do want to be
consistent with your moodboard so that you don't further
confuse Midjourney. This one's good in
terms of the text, not so much the subject. I do like the contrast
between the colors. I Okay. Now I have a bunch of stuff in my moodboard and
what you can do to increase it even more
is go on the one that you like and then create
different variations. I really like this one. My
only problem with it is that the duck is fully a robot. Let's do a subtle, strong, and then I will add this
perhaps as a image prompt, but make my omni reference
my original duck. Maybe add this as
a style reference. Let's put in our last prompt and then mention
all the parameters. Okay, so we're getting
some cool stuff. I really like this
one, and that was just a strong variation, meaning that it goes away
from the original picture. Was subtle just keeps the
picture that you created a variation from and just
makes it slightly different. But out of these guys, let's see which one
looks the best. This one looks cool. I will add this to my moodboard later. Here's a combination where
we have more of that duck, but it's the position
of the robot duck. Let's create variation
up of this guy. I'll create strong and subtle. The text needs more
work in my opinion, but it's definitely a duck in the correct position
with the colors, the cityscape that we asked for, the technical side, and
the rest is all there. Okay. So now we have a
cartoonish version of our duck, which could be
another style that you could add on
for your bundle. We have this style. Then add
those onto your gallery. Okay. Now, when you're done adding stuff
to your moodboard, just look at it from afar and see what's not
really fitting in. For me, that's these guys
because I think we use the raw parameter and it made
it really photorealistic. Let's just delete these two, and that leaves us with
a flat cyberpunk style. Now that we have an idea
of the concept and, of course, our moodboard, we're ready to create the
different poster designs and see where and how our duck looks like in
each of these spaces.
36. Trendy Poster: Variety and Upscale: With your moodboard, you're now going to create
different positions, styles or accessories
for your subject. I will be doing
different accessories. The first thing I worked on is my duck without anything on. Then we can try a
cowboy hat, sunglasses, maybe a little mustache, and that's going to be my bundle that I will put out
as the poster pack. When you put in the
same prompt, this one, just make sure that
you go on this P, turn it on and your moodboard is the sci fi duck and
not the other ones. Click away, add your
Omni reference. I have mine set at 100. And let's just generate without any sort of other inputs and see how well the moodboard is involving itself
into this output. You can see that now I have
global version seven profile, which is the profile
that I made when I rated those 200 images, and then we have the
moodboard listed here. Here is my rubber duck without
any sort of accessory. I think that looks really good. I'm just going to
remove the text because I could just add
that on Canva later anyway. Let's repeat that with Omni reference the same
settings such as click on them. And then just put after last last word in
a prompt. No text. And then generate.
The good thing about using the external tools is
that if you, for example, had your own font, you can import it onto
Canva or Photopea or some other Adobe program
and put in your own touch. Even if you don't
have your own text, you can just put in something a lot better than
what we're seeing now. You can put on your logo,
little signature icons, whatever it is that can give you the um basically make
these your brand. Even though I asked for no text, I think in the moodboard, we have a lot of text shots, so it's not really
registering that. But I think that should be
fine because we can use Canvas eras tool or
photopiece eras tool too. Because these are
just random symbols and not really text. So I think what we could also
try is using the Edi tab. I'm just going to see
which one of these I like the most. I
really like this one. I like that it's a real rubber
duck floating on water, and it's just like this
techno cityscape in the back. I do want it to be a
little bit more flat. Let's use the same image as
the prompt omni reference, and then the flat version
that looked cool. I think One of
these should work. Let's put that as
style reference and put in the parameters. Similar to our mockup project, we're going to first focus on the subject without
anything a plain subject, and then add on the accessories, different poses,
different styles. Whatever you're trying
to do that will make your pack
slightly different but maintain that main branding
for you. Here's my duct. It's a little bit flat but
not as flat as I want it. Maybe we should use a different image and then
mention the word flat. Put this in as image prompt or actually
we chose this one, the cell reference,
omni reference, but mention the word flat. Retro flat, maybe
two D Sci fi poster, so no three D elements, and I think I'll remove this and put in one
of those flat stuff. This one is a little better. I'll go in my
moodboard to see if I have any three D element.
Maybe it's the guy. Let's remove that and
everything else is flat. I will also remove the. I feel like these could be
disturbing the generation. Let's run this one more time now that we have an
updated moodboard. This is looking a lot better. We do have some attempts
at Japanese text, but of course, they
don't really make sense, and that's just the
funny thing with AI. Maybe we could try this guy. I'm going to wait for
the other set and see if I find something
that I prefer. Let's like this for now, and I will definitely try to remove this text and
put in my own font, and then we can try
the curly braces to maybe add a cowboy hat, a bow tie, and then we
can do a shawl maybe. Here are some more designs. This one looks funny. We have the retro sun element
here, which I also like. Now let's add this guy as our image prompt and the rest as style reference because this
is the exact style we want. Omni reference for sure. Same prompt, these parameters. I have this turned on, but I don't think
we need it anymore. Then we'll do the curly braces to put in the different
accessories for our duck. A giant pink rubber
duck with curly braces, cowboy hats, bow tie,
and then mustache. Curly braces, it enter and now you get three sets. You
can do more than three. I'm going to just keep it
lower and then you basically, if you wanted to do more, do curly braces for all
the other components. If you wanted to
maintain your pink duck, but in different styles, then you would put
a curly braces in the section that mentions
the Neons scape, retro flat two D,
that sort of thing. So that's how you can just
incorporate the curly braces. This guy is our cowboy hat, and we can see for three of
them, at least, we have that. This is our bow tie, and I think mustache is a little hard to do considering
the big lip. Let's see what can I do instead? What can we put on this
duck? Maybe sunglasses. Let's put in all of these guys again and then do sunglasses. But now for the cowboy, it looks pretty funny. This guy looks like a detective. We have a smaller head.
I don't like that one. This one's my favorite, and I will upscale it to settle. If you want, you could
create different variations, but for me, that will do. Then this one is our second one. I just want to
make sure the duck looks remotely the same. I think these two
look more similar. Let's upscale this
to subtle and then hopefully we get the
sunglasses. All right. Again, I'm going to compare
the lip to the other ones and make sure that at least insinuating that
it's the same duct. Got the lip, so we have to go for the plum one,
maybe this guy. And then upscale this. For our third poster. Now I have three upscaled shots. I'm going to attempt removing this with the edit
tab. Let's try it. It may not work, but
it's worth trying. Better yet, let's try one
without any of these inputs. Hopefully that will ease
the distraction element. This is perfect. I'm going
to upscale this two gallery, but now I have a space to add real text so something
that actually reads. That was a cowboy one.
Let's take this guy, edit it, remove all the s, hit the X, we have
some more text here in the little just below the headline
watch out for those. You could remove
the text up here. I'm going to leave those B because the attention
is on the duck anyway. We don't need to worry too
much about those guys. Upscale and then repeat
with the third one, which is the sunglasses. Got some up here as well. An upscale to gallery. Don't know if you pressed it. Okay. So basically
with this moodboard, you have a system of generating more versions of this concept, the rubber duck in a
cyberpunk environment. If you, for example, get a
order for a pet chicken, you can just use
the same moodboard, same prompt except
change out the word duck for chicken and
you're all set to go. Once we're done, we're going to download it as we did
with the previous lesson, and I will be showing you guys a different platform because
by now for sure you're done with the Topaz labs free
trial items not this one. We were downloading
the wrong ones. Okay. Since Topas labs only had ten free images
for you to upscale, I will also recommend
this platform. It's free to use and it's
pretty straightforward, just upload your images as a badge or actually
not as a badge, get to do it singularly and
then just download them. Let's go with our Cawboy hat. Let's increase the ratio
to either 200 or 400. I will go with 200,
upload and start. It may take a while to process
because it's a free tool. I'm going to upload the
other two the same way, and then once it's done, you can just download it, and then we have
an upscaled image. I'm going to repeat that and
then we're ready to bring it into Canva to
finalize the posters. We will do that in
the next lesson, take your time, upscale, edit anything further within the Midjourney editab and then come back for
the next lesson.
37. Trendy Poster: Editing and Exporting: Now we're here on Canva
to finalize our posters, put some text on it,
and then export them. With Canva, we can go ahead
and make a portrait shot, which is three to four poster. I believe that's the same ratio we created things
with three to four. That's why it's really
important for you to check the ratio before
you make your work. For some reason you chose
something by accident, maybe you did landscape
instead of portrait, you can just re run that same prompt with
this new image size. Let's say I found out that I
need to have a square image, just switch it like so, put in the same prompt and use this guy as your image prompt. Another trick that
you can do because every time you hit
submit the prompt, the little error here, it's going to give you
different variations or different understandings
of that prompt. There's also a command
that lets you create multiple outputs and
that's dash repeat. Case you want to use that,
I'll show you how to do it. Put in the prompt, the same
stuff that we went over. AR was supposed to be
one to one, my bad. 121, Version seven and
then do dash dash repeat. However many times
you wanted to repeat. Let's say five times. Hit Enter and now we got five things happening
at the same time. That's a helpful
parameter to know, just so that you're not hitting the Send button five times. Okay. Then you can see that everything here
is indeed a square, so keep that in mind. Let's go ahead and
make a poster. There's a template right
over here and I'm going to upload all my shots
for each of the pages. I have three designs. I will do three pages and
let's upload our work here. Here are my three ducks. I'll just grab them on there. You can click and
drag to move to the other page or just
Command or Control Z, delete it from here and
paste it in the other page. Here's a quick preview of what your poster
will look like printed. Obviously, we want to do
a little bit more than just this and you guys can change this up and
do your own designs. But I want to increase it and give it a nice border around. Resize it like that and
repeat with the other ones. I'm just going to check
the width and height here, 16.7 22.1 and make sure
it's in the center, so the border looks pretty good. Now I can put a different
border on the back. Now, because the
image has a border, we'll come back
to crop that out. But for now, just
go to elements, get a shape, and make sure that shape fills
up the entire screen. Choose a neutral color and then do Command or Control left
bracket key to send it back. Now we have a cute
little border. Grab that element, the
shape, Command or Control C, and then Command or Control V. Send it back one more time and
repeat with the other one. Now let's go on to each one
and delete the border that there originally because not every one of these images
have the same border. Just by simply
grabbing this side, the longer rectangle, you
are able to crop your image. Just slide it down
until you're not seeing the Midjourney borders. Now I have the borders and my poster is looking
pretty good. I will change the
background color and send it more towards
the warmer tone. We could even sample
something within the poster by grabbing that key. Maybe something like
that could work, but we will make it
a little lighter. Maybe that color. Then
we're going to copy the style and paste it on
the rest of the shapes. Next comes the text because we empty this space for something
that's actually readable. Go to text, add a heading and
choose your favorite font. For me, it would have to be a techie font just to fit
in with the background. We have these fonts, you should be able to
see the same one. I'm going to just see if
I prefer this or this. Maybe try putting in something. Let's do the Duck Sheriff. There's different styles too. Let's do the black
one and then we'll do a neon yellow color just to
add to that retro effect, and then we can outline
it with the effect panel. Go to outline and we can grab one of the colors
in the image. Let's see. Maybe this pinkish
color could be cool. Click away and you can decide the thickness
right over here. They usually do about 40 or
something on the even side. You can also change the
text by going over here. I'm going to just squeeze
these two together and perhaps we can add a glow
effect as well, something neon. I think it's actually
better without the red. Click away and then we're
going to put in a little text, since this is a poster, let's do a subheading,
put it over here, and let's do something weird. Whatever text you
want to put in, and then I'll utilize this font to maintain
that techie aspect of it. We can even do a glitch one, actually. Let's try that out. We could do yellow or grab one of those pink
ones from the background, I'll compare them both. That's hard to see. Let's
try this neon color. Going back to effects, let's offset the
glitch a little bit more and choose a
different color set. When we zoom in, it has
some glitch effect. Think if we change the color, the glitch will be more
noticeable that way. Maybe white. Let's put it here
and make it a lot smaller. Next, I'll fade it out,
so it's mysterious. Now, let's copy
the same text and just change the wording
for our two other ducks. We can put this guy, hold down shift to move it down a straight line and then
repeat with this guy. Let's give these to a name. Let's do the Let's
try to be smart here. The duckom the gentleman thing. And then we'll do
the cool uncle, like a duck uncle. Something like that, and then
this would be my package. Of course, you can put
in some more elements because Canva has a lot of free stuff that you can add on. Maybe you can try some graphics. There's a lot to choose from. You can search for stuff, but I'm going to keep
mine the way it is. If you have a logo, just make sure you find a spot for it. I do want to have that
tiny paragraph situation that some of the posters do. I'm just going to go to lowering Ipsum IO where I could get a dummy paragraph and pretend that's the
poster specifications. Let's go for a serifont
that has the handles. We'll try like Alice maybe. Make sure it's white and then make it really tiny, maybe 30. Then close the borders here
so it all fits in the middle. Then we can drag it down, put it right underneath. I think even tinier, maybe 20. That's exactly what
I'm looking for. Put that over there, increase this so the little
decrease can be more visible. I'll actually try
15, really tiny. Then just copy and paste
on the other ones. I'm going to have
to move the text here to this guy and
the one behind it. Make sure you hold down shift, so it all moves in
a straight line. All right. Now, let's
talk about exporting. When it comes to posters, which format should you
put should you use? First of all, before we do that, if you plan on using Canva
for your poster production, make sure you give it a
name and you name each of the pages so that whenever
you get an order, you know exactly
which page to go. Right now with three pages,
it may seem obvious. But when this takes
off and you have 50, you want to make sure
you are very organized. Hit Share and we're
going to hit Download. There's a lot of
file types here, but in case you're not familiar with them, I'll
go through each one. Well, the ones that matter. JPEG already gives
you a description. That's not something
I would recommend for the poster itself, but for the preview, just like we did
with the mockups, you can put that in
as JPEG and then the real poster needs to be PDF for print and PDF standard. This way, you're
giving the client the print option and a
digital print option, you will be more
in demand for it. You need to have three
outputs with posters, print standard, PDF standard, PDF print, and then JPEG. You don't really need to do
PNG or any of the other ones. I will do a PDF print for
now, including all three, or you could sell
them separately, whatever you choose to do and make sure this
guy is a CMYK. Then you just hit
Download and upload the files that you
extract for EDC, Red Bubble, Society six, and a bunch of other ones that I will go over in Chapter seven. Let's also download JPEG, all three, and
then standard PDF. As you saw, standard PDF didn't
ask for our color profile because that's a whole
different format when it comes to print, there are certain colors
that need to be translated. What we see on the screen
is a mix of RGB or red, green, and blue colors, while C and Y K are
what's used for print. That's Cyan, yellow, magenta, and K is black. You just have to follow
those standards. Now I will have
these three files as well as the files
from last chapter, the last project, actually, in the resource pack for you guys in case you
wanted to look at it, get some reference or use
that as your starting point. Feel free to do
that. There's a lot of other resources as well, and now that we
have our posters, we can move on to
our next project. Which will be a storybook. We're going to work
with characters again except now we have faces. We will put our characters
in different positions, maybe come up with
a funny story, a few pages, of course, so that the lesson is not
too long and you can take that idea and turn it into
either a storybook or comic, manga, whatever you're into. Let's move on to
the next project.
38. Storybook: Pictures and Story from ChatGPT: For this last project, we're going to be
making a storybook where we're going to have to
make consistent characters, consistent styles, and pair
it with the story line. You can take this any
direction that you want. I will be making a character and have maybe two to three
pages worth of a story. To be utilizing Canva again, make sure that you're
comfortable with it this time and this way we will have a easier time laying out all the components in
our final storybook. First of all, we have to think about our character
in the story. You can decide whether this
is going to be a human, an animal, an object, something unconventional,
whatever you want to do, and work around shaping that
character for that style. If you're going to
do a two D thing, which I do recommend, what character is
going to fit well and how should they look
in this two D setting? So cartoon characters are rather popular here
on Midjourney. We have anime styles like that, something really flat
and simple like this. You can even do highly
detailed ones like this. I wouldn't recommend
starting here though, because keeping all of these details consistent is
going to be a little hard. First, let's go ahead and create different pieces for
the moodboard and then we'll use that board as one of our inputs for the prompts. Back here, let's
call this storybook. Go to create and just type in something really
simple book character. Right now, I don't know exactly what my character
is going to look like, what's their species
and all that. I'm just putting
in book character, and I'm going to do another one, increase the variety
a lot more and maybe weirdness to get some fun ideas that I could further explore. So we got a little boy here, we got a priest. We have an adventurer with a briefcase and a
girl with ponytails. Over here with more
chaos and weirdness, we're getting some other stuff. We get this guy with a
cane, this special style. I'm not sure. It's like a two d style and she has all these ribbons coming
out, which looks fun. We have a three D, realistic small figure and this art style, which I really like this. And we're getting more
three D elements. Right over here, I can decide
if I want to do something two D or three D. Once
you made your decision, just put in the same
thing and before book, put in two D or three D, and I will add these settings. For weird, we can do 100, something lower and
slowly formulate what our character looks
like so that we can make that base position, then have them move in different directions
and do various actions. We got all two D
except this guy. I think if we reduce the weird, we should be getting
something more like this. I'm going to create different
variations here, strong. Since that's my favorite style, it's a watercolor look. Here's our two D book
character with weird 600, so it's all three D. Going to ignore these and this one
is looking pretty good. So I think I will go
with this style and just add one of these as input
for our style reference. And while we're here, let's
do Omni reference as well. Now we're going to
do book character. Well, we can first explain what this character looks like. Female book character with
pony tails and long sleeve. It's like a yellow top. Yellow top holding a basket
of muffins, let's say. But in the chaos, Version seven, you can try Version six
as well and then submit. I'm going to try a different
hairstyle while I'm here, put in the exact same things.
We're getting weird again. Let me reset this and
we run this prompt. The weird always ends up
making your characters three D. I'm just running the same thing, but
without the weird. Let's try our
different hairstyles, Kos ten version seven. We can do long curly hair. Then I'll do maybe a blue top. This is with the weird. Again, it's very three
D. Now it's a mixture of three D and two D. Not a fan. This is the same prompt, but we have her hold
a basket of muffins. This one's more ideal. For some reason, it
didn't listen to the yellow top, but that's fine. I think if we remove
the omni reference, that's probably what's
influencing it right now. Maybe we can lower it to 25 and then do one without
the omni reference. This one's pretty good. We do have a problem with her fingers, but we can fix that in post. I'm just going to heart
this for now so I can add it easily to my
moodboard later on. Now you can see just by taking
off the omni reference, I'm getting a whole
different character style. We have a mixture of the
green and the yellow. I think this is
cute. I'll create some variations from this. It's hard that as well. She has that hair, she's holding something and that's her outfit. We could switch it up and
we'll describe her outfit. Put in these parameters. But long sleeve yellow
top and pants maybe. Blue pants. I'm going to add this as omni
reference as well. Just make sure to
lower the influence. Ow ten maybe. So these ones are
again going towards the three D look slightly. I'm thinking of combining
this style with the characteristics
in our later prompt. Let me first see
what we got here. All right. These
are not that bad. It's a whole different
style, though. We have a longer limb situation. Here, we got ginger.
We didn't really say what color her hair
is, that's fine. I want this style to be preserved because of
the smaller details. So let's put the same prompt. And this time for
Omni reference, we're going to add this guy. Add her here, and
then for style, we're going to go
with one of these. Since this is full body, make sure to add one
of the full bodies for style reference. Okay, so I ran a few more, and here I described
some of these styles, so cute Illustration style, storybook, children's book,
and this is what I got. And then I went ahead and
describe the environment. Where is she exactly? I said tall stone
towers with people passing by and she's standing
in the middle of it. I think this one
looks pretty good. Let's create some variations. It's as you can see, the
pants is there or yellow top. This green element is coming
from the omni reference, the long curly black hair. I'll leave it as black. You can change the color. If there's any imperfections, don't worry about it right now because we're going
to fix it later on. We have some people in the back. I think I'll leave them
B. But this is what our character is looking like
so far. I'll like this too. Now we're getting her in
different head angles. Here she's looking up,
looking to the sides, turning back, that
sort of thing. I like the environment
of this one a lot more. Of course, we have a
little hallucination, but that's something we can fix. Now we're going to go to upscale the two images
that we want to combine. I said I like the
background here, so we have to upscale it subtle. Then this is the character, so we upscale this too. What I want to
describe further is, I guess for now we
could leave it B, just combine these two. I don't think we need
a third element. Maybe we can leave that for
the pieces other pages. Let's take the one that has the basework which is this
guy, going to compare the two. It's pretty much the same. That's very consistent already. On second thoughts, I won't
combine this with this. But either way, let's
edit this because we do have some hallucination
in the hair. I'm going to delete
the style reference, get my erased brush and
just redo this section. If there's anything
else that looks weird, I think these two people looked a little bit combined
with the lamppost. I'll redo that and this will be the first
look at our character. But we do have a third
handle. I missed that. This is the hair and that those people removed completely.
This one looks better. On top of this,
we're going to make another selection to
remove this part. Like that. While I'm here, let me check on other things. Her ear looks weird for sure. Just make one dot there. Her shoes are fine.
People in the back. I'm not sure what that is. Let's remove that. These
are just regular people. Submit the edits.
For my storybook, I'm going to do a
square image size. You can change that up with
any of the other ones. Since we're doing it
on Canva, anyway, you have a lot of freedom as to what your book looks like. Now we got different looks. You can also increase the size here as we learned,
feel free to do that. This one's my first shot and I just zoom in.
It looks look fine. So let's go to the
moodboard and add all of these things that we
made this last set. Grab these guys that's still
developing, let's put these. Basically, we started from these guys and ended up
with this character. Let's go back and I will
add this last one too. Okay. Now that I have
my base character, I can start putting her in either different locations or having her do different poses. Now for the actual story, we could use the
help of ChatGPT. I'm just doing this for
demonstration purposes. Obviously, a story
written completely by AI may not be the
funnest thing to read. Unless you explicitly say
that this is an AI storybook, then that has its own audience. But we're just going to
go over here and ask for a very simple three
page children's story. Write me a three page
children storybook about a young girl with muff
a basket full of muffins, walking through the
cobblestone town. Let's see what her
mission would be. She's looking for the
best picnic spot, maybe. Make this let's do make
this funny and fun. Use rhymes and provide
maybe four lines per page. Now we get these
short little poems. This could be our first scene. But if you don't
know how to describe basically the position that your character needs to be
for page two and three, you can also ask
ChatGPT for help. Write knee a Midjourney, prompt for page one, two, and three to generate the
visuals for the pages with an aesthetic storybook,
illustration style. You can do like colored
pencils, not colored pencils, actually, Let's do with vibrant colors and
two de textures. Here it gave us a
different dress, so wearing a bright dress. We could switch that
out for something else, basket full of muffins. And now we just
got a copy paste. I'm going to mention she's wearing change the
prompts so that she's wearing a yellow
top and blue pants with long curly black hair. So we got all these three. We have our moodboard, and all we got to do now
is put in the references, put in those prompts, and start building the
different pages for our story.
39. Storybook: Layout and Export: All right. Welcome back.
We made these prompts and this whole story in ChachiBT last lesson and made our
moodboard with the following. Now we're going to
start generating the prompts that
ChachiBT gave us based on the
characteristics that we already decided
for the character. So first page is this
prompt, copy it. Command or Control,
go to create, paste, make sure this guy is turned on and you're
choosing storybook. Kaos ten version seven. Let's add this as
our omni reference. And we can add the
same thing for style. Let's hit Enter. Here's our first set. I go to zoom in to show
you the differences. We do have a lot
of hallucinations. This one is fine.
Got an extra flower. If you wanted to use these guys, just remember you got
to edit it in here and have Midjourney
regenerate the weird parts. But this one looks pretty good. I'm still going to zoom
in and see if there's any sort of odd detail. Which we don't have. This is my first scene. Let's upscale it, subtle, and then begin with
our next prompt. Put in the exact same thing, then put in the second prompt. I believe all the other
details are there. We forgot chaos ten. Chaos ten, OW 25, and then the P is red, which means the moodboard is being used here,
profile and all. There's our upscale shot. Let's heard it and download
it for further upscaling. I did notice that we
do have this area. I just came to the edit tab to fix up a few of the
hallucinations. Mainly, let's delete this. This area with the hair. It's merging in her
shirt. Submit the edit. I'm going to check this
door. The rest is fine. I think it was just that area and we could see it trying
to fix it in real time. I'm going to repeat
the same process with the third scene and
then I'll be back so we can see what
edits we need to make. I made all of my shots and I was editing
them in the edit tab. Once you're done, as
always upscale gallery. And I have three scenes. For this one, she is
wearing a different top. Well, not this one. This one. We're going to attempt to redo the top hopefully it's going to listen to
the yellow aspect. Going to just make
a bigger brush and go over the top area. But the other two scenes, they were upscaled and edited
and it's all ready to go. In the meantime, let's
look up the dimensions for a children's book
and that's 11 by 8.5 because we want it to be landscape, so
it's the opposite. On Canva, you could just
create your custom size. In inches 11 by
8.5. There we go. The first thing
you want to do is split this down the middle. Get a line, rotate
it, click once, then hold down shift to
get it to the perfect 90. Then you can see this pink line directing us as to
where the middle is. Once it's in the middle, click one on this circle, hold down shift, and then
stretch it all the way down. To change the thickness, just go on this guy and I
guess change it to two. That should be maybe one. There we go. To make it
stand out a little less, let's go for a gray color. And then lock this in place. Now grab your frames. I did a square image, so I need a square frame, put it here in the middle and then grab your text and put
it on the other end. Or we could do here, one here, depends on how
you want to lay this out. So I will do
something like this. Make sure it's in the
center, grab both, alter option, then shift, drag it to the other side. Now I have two
images and two text. We're getting some
interesting tops here, but whatever you
see that is better, just build off from there. Go to this one and begin erasing one more time because it
already figured out the bottom. I think what I will do is
I think this is too much. I'll do text here, one image on the
other end. Yeah. Let's see. This one
sort of looks better. Again, I'm going to
make my brush really tiny and keep doing this until we get
that perfect yellow. There we go. Let's upscale to gallery and we're done
with our three scenes. This was our upscale. I'm going to download it
and then wait for this guy. Now she's wearing the exact
same color, same hair, same basket for the muffins, and she's just going around trying to find
that perfect picnic spot. All right. Now back to Canva, we are going to start uploading the images and
putting them into the frames. Just duplicate this two more times or let's first
actually put page numbers. I'll do page one. Make this smaller, maybe 15. Watch out for
margins, Alt, shift, drag on the other side,
that's going to be page two. Now just duplicate
this two more times. One, two, and then
change the numbers, three, four, five, and then six. Since this is a book cover, we're going to do the front
and back cover as well. I think I will
duplicate this one more time and just delete
these components. Drag in your scenes and just
drop them into the frames. Going to copy this,
delete it from here, and just fill up the pages. Oh wait. That's for that. This page two, that's page
three, and that's page one. You can grab the frames and maybe round them
up a little bit. Let's stew 15, grab the copy style and paste
it to each of these. Now let's go back to this poet, this short rhyme, copy it and put it
where the headings are. Since it's a cuz vibe, we want to go for
a serifont course, not as bold and choose
an appropriate size. Then of course justified. Now we have the
story, the image, make sure that the
quotes are fully visible so that it's
easier to read. But as for the rest,
it's all good to go. The last thing we
need to do is design a fun cover that's going to fold into the front
page and the back page. This one, you can
do a landscape. Image size. Let's
go here, landscape. I think a four by three,
maybe a three by two. Choose a three by two, put in the image that we use
for the omni reference, and then the same for style. Now let's ask for a prompt
for the front and back cover. Well, just a front cover because that's also going to
be the back cover. For this one, I'm going to
play around with the towers. This could be the front side of the cover and the towers and buildings could
be the backside. Let's grab something
like this guy, scale it up first as always
and then edit it where we erase half and bring in the
same image or maybe this guy. This one looks cooler.
Let's upscale this two. We're going to merge these two together in the
edit tab and then put it here where it looks like it's the
front and the back. Let's take the front side, edit, and then add our other image. Open this in a new tab
and copy the image URL. That way, I could go to add
and do it a lot faster, so just paste it and
there's your layer too. Using the MVTol situate
this in a way where half of the scene is her and then on this end we're
just getting buildings. And just try to
match the bottom. With both images, Midjourney will fill up the spot for you. Then grab your paint erase and erase the edges for
both of the images. Layer one, layer two, and then we'll have
Midjourney stick it together. I'll do another one
where we don't have an omni reference since we
have our character already. That's our first building.
Let's take a look at the rest. This one looks pretty cool. So I will upscale this
to gallery and then bring it onto
Canva. Scale it up. Hopefully it if it doesn't
can go back and resize it. But for me, I don't mind
cutting a little bit. Just make sure you push it
back so the fold line is visible and it's always
lock that guy in place. Now we're going to give
this a nice title, so grab a heading and
just write something like Molly Mae as
the main title. Choose a vibrant color and
set it right in the middle. Like that. Let's give it
a background as well. So Molly Mae, and then we
can hold down Alter option, click and drag to put in
another part of the title. I'll break it up
into three parts. Let's put this here
and then Alter option, click drag for the third word. Okay. Now you can fill up the details
regarding the author, which is just chat tri But, I'm going to label that here. There's always some
stuff in the back. There's always a logo and a
bunch of little texts talking about the contributors
and all that stuff. Let's put in write a short description for
this book's backcver. Put that in right here
and this is the text that people usually read
before they purchase. That's why we're
putting it there. I will go with another font. Maybe a sand seraph
will do better. Make sure that it fits in nicely maybe 15 is a good number. Scale it down and
then if you want, you could give it a
nicer background. I'm going to center
it and I'll try. We can do something
that makes it a little easier to read perhaps a shape. Okay. And then finally, the most important
detail the author. I will remove that and just
choose a different font. Yeah, that one's good. 30. We'll just put Cha chibT there. Right. Okay. So there is my storybook. All you got to do now is share it sorry download it
as a PDF for print. You will have the printer
print this for you. They will handle the folding. If you want to print
this on your own, just make sure that
you're fitting the design onto the standard. Some printers prefer
a full page design, some printers like half. It really depends on where
you plan on printing this. If you want to do
something digital, then it's not that important. I would just be your preference. I just downloaded this as a PDF standard since
I'm going for an eBook, but feel free to do
a PDF for print. I will have this in the resource pack as well in case you want
to reference it. That's how we can
make our storybook using HHIPT
Midjourney and Canva.
40. Mistakes to Avoid: We've covered a lot of
ground in this course. We started from
the fundamentals. Then we moved on to
prompt essentials, experimenting with different
image types to creating complete projects where we designed our dream room and
then our fantasy character. This last chapter is about
tying everything together, helping you avoid
common pitfalls and showing you where to go next so you can keep growing
as a mid journey creator. This lesson, we're
going to talk more about the common
mistakes to avoid, starting with what not
to do with Midjourney. A lot of beginners especially stumble in this area and knowing these ahead can really save you a lot of time and even
weeks of frustration. You put in that prompt that
you thought was perfect. Maybe you used ChatGPT. But then you keep getting
some weird results. Maybe your subject
has 20 fingers, it's not even a human and
you don't know what to do. Let's go over four
mistakes in this lesson and these will really help you better understand where
things could go wrong. Mistake number one is
overloading your prompt. Sometimes we try to fit ten different ideas
into one prompt, which may sound really awesome, but it actually confuses Midjourney and gives you
different hallucinations. Just like a human, the
AI gets confused too, so the results, which
is the understanding of the AI from your
prompt is really messy. Always think about this example. If you have a complex
idea and you were to write it in a book and
make someone read it, is that person going to be able to visualize what
you're talking about? If a person cannot
understand your ideas, then AI will definitely
not understand it. The best thing to do is to build prompts in layers and
iterate as you go. With our course, especially
with the fantasy character, we started a while back from here where we
had the smoke background. I didn't immediately
go in and put in my entire all my ideas
with the sword, with the moon, the
glowing spaceships, the force like background
into one place. Instead, I built the base
workork which was my subject, and then built
different iterations. With that one, we
actually started here and then we ended up here. Take things slow, build the
main idea in one prompt. Once you have a good
result from that prompt, add that in as an
image reference and then build up upon it. Mistake number two is
ignoring aspic ratios. That's the AR that we're
putting down here. In this bar, it's image size. If you're designing
for social media, wallpapers, posters, aspec ratio really matters because if you're
putting everything as a square and it's time to use that square
image as a wallpaper, you're going to lose a lot of quality because you
have to scale into that photo or perhaps even
crop the top and the bottom. So don't settle for the default square unless it
fits your project. When I reset this, it's a one to one ratio, which
means a square. I need to make sure
that I'm matching the ratio for whatever
output I'm trying to make. And if you, for example, want to post on TikTok, you would go to portrait
and on their website, they tell you the
ideal aspect ratio. You want to make sure that
you set that before you even start generating and
using your credits. Mistake number three is
skipping references. In our examples, we used
a lot of references because we're trying to get a nice mix between
different ideas. This is apart from the prompt. For example, we wanted
to add the subject, the background, the
sort, and the moon. If I do the same prompt, but without this inside, I will get a completely
different image. I'm just going to show
you what that looks like. But basically, when you
describe something and you add a reference photo that's
related to that description, what you're doing is that
you're grounding your results, meaning that they're
going to look more consistent the more
you build up upon it, especially if you have a
unique character in mind, you want to make sure
that that character is actually within your results. Right here, even
though I do have my female warrior, the moon, the glowing stuff,
I don't really have my character. These are
some random people. This one looks like a game. This one, her back
is turned to us, and this is a whole
different character. But then when we
added the references, we got that ideal angle, the right colors,
the right moon size, the right sort, and so on forth. Make sure to add at
least one reference when you're trying to build complex
generations like this one. Finally, mistake number four is forgetting post production. The raw Midjourney output
is rarely the final step, especially if you want to do this in a more professional way. Throughout the course,
we use Photopea, Canva, and Topaz AI to make our
work look more professional. Starting from the colors, we added some filter
to the overall image. We used Photopea to make
certain selections, maybe change the color
of the armor only. We use topaz to upscale beyond what Midjourney
is capable of and then Canva to put it in
a nice little poster to share it or maybe print
it for our bedroom. Make sure you're also
exploring those ideas. The tools that I mentioned through this course
were all free. Only thing you're paying for is Midjourney Photopea
is free to use, Topaz AI is free
up to ten images, and Canva, of course, is free unless you use
the paid elements. If you avoid these
traps that I mentioned, you progress much faster
than most new users. Again, make sure that if
you're using references, you are allowed to
use that image, meaning that it's
not an original work of art where an artist drew it, but a AI generated image. When we went on Pinterest, we used for AI modified
tagged images only and that way you are avoiding
any copyright issues. I though we know what to avoid. Let's talk about how
to stay sharp as Midjourney keeps evolving
every month and every year.
41. How to Stay Current with Midjourney: We've already gone
from version one to version seven in
just a few years, and it's only going
to get better. Each update changes
how the prompts work, what the model is best at, and also what features are
available to you as a user. In order to know
about these changes, you have to be within
the right communities, and that's the thing that we're going to be
discussing today. I will be going over
three tips for you to stay current with the
new Midjourney features, and that way you will not miss any of these big announcements. The first one is to
join the Discord. That's where all the updates
are announced first. Right over here, I have it open. You just search
Midjourney Discord and if you have a
discord account, you're just entering
one of the channels. We have different
newcomer rooms. We have a discussion,
prompt craft, you can showcase your work and this is where the announcements
are going to be made. New features like remix mode, blend or motion are often
dropped here before they actually come to the
Midjourney web interface. When you go over here this stuff down for image style prompt, this is the remix feature
where you get to use other outputs like this one to make your own work up here, the blend feature, which is in the Edit tab,
we've used it before, and then the motion, which is the way you turn
images to videos. Within Midjourney, there's also a chat feature which
is similar to discord, the layout looks the same. This is also where you get to share your work, get feedback, discuss things
with other people, find other beginners
and maybe some friends. You go to your images and get
some inspiration as well. You can re run their
prompt, use their prompt, and do the same things that
we've been doing and explore. If I really like this
person's artwork, I could reply and give them some compliments
or even some feedback. The second tip is
to always check the documentations from time to time, actually, not always, which is basically if you search Midjourney documentations
on their website, they have the most
relevant information, most, I would say the biggest
things about Midjourney. How it all works,
a parameters list that walks you through all the commands, what they're for. These are the things
that may be hard to combine when you're only relying on Discord
or the chat feature. The Midjourney dogs
like this one, and also slash Help commands are really lifesavers when
something new is added. You can do dash help on Discord or just go to the Help tab here, where there's actually
an AI assistant ready to help you with anything
regarding the tool. Third place is following
other communities that are not Discord
or Midjourney itself. These include reedit Twitter, YouTube creators and they post experiments within hours
of a new feature release. Let's say you learn
about the Version eight and you just don't
know how to navigate it. Maybe there's a whole
new tab over here. The best way to get familiar
with it is to go on YouTube, where you're going to get
a demonstrated explanation of how that version works. There's also a Reddit community, which is R slash MIDJRN here, you can also learn
about the updates, see the other people's work, comment, engage,
and do all of that. But I would say if you're still a beginner and you're
trying to learn the tool, especially the new
ones that come out, YouTube is the best place to go. Learn from their testing
instead of starting from scratch yourself and
then wasting your credits. Once you learn from
these communities how the structure works, how these new releases work, you can actually
start experimenting yourself and find
your own workflow. You will avoid all
the beginner mistakes and just go right in
applying what you've learned from YouTube
credit in this course to actually building things with the new versions
and the new tools. Now you're equipped
to keep learning. Let's actually recap
what you've gained from this course so that later you will know exactly where
to take this next.
42. Where to Next: Think back to where you started. You might have only known how
to type a simple product. Maybe you've used
another AI tool before and we're wondering
how to use Midjourney, or perhaps you didn't even know anything about AIR
tools to begin with. But now you've built
two full projects, which was the dream room
and a fantasy character. Later on, we had applied projects where we built
three different things that belong to certain categories. In addition to those projects, you learned how to refine, combine and present your ideas
in the most creative way. Let's first take a look at our skill path recap and see where you started from and where you
are right now. The first thing we
did was learn about prompt structure
and the keywords. Then we moved on to image
types, which included people environments, abstract
art and even motion. We then looked at
some creative tools which was in the Edit tab, which included the inpainting, blending, references, and remix. Finally, we applied those
two polished projects for us to showcase. Now that you know all the basics and how to apply them properly, you can think about
the fields that you're interested in and
begin brainstorming how you can apply Midjourney and all the tools that we
learned about into making your workflows easier and giving you that creative
boost that you need. Finally, let's talk
about your next steps, where you should go
after this course and what are the things that
you should look out for? This course was meant
to be your foundation, but keep in mind that AIRt
is moving really fast and soon you will want to keep leveling up from where
you are right now. Here are some tools that you
should take a look at next, look them up if you really enjoyed using
Midjourney as a tool. The first one are runway AI, which is great for videos. It has stable diffusion, which is one of the
models for more control, and there's also Photoshop
AI for hybrid workflows. Both of these tools are paid, but they're excellent at
what they're made for. With runway, you can even
generate images and then turn them into videos with scripts with different
camera angles. You can even make a fake camera within that platform and move it around to make some
intense fighting scenes and many other things. Then if you're familiar with Photoshop and you've used it, which is similar to Photopea
which we used in the course, you may want to
look at their beta, which includes the AI
portion and that really combines this whole
AI generated images. Into your regular
Photoshop tools. That's a really cool
and evolving feature that Photoshop has been working
on in the past few years. Next, you'd want to,
as we mentioned, stay active in the
different channels such as discord channels, credit groups, and even
design collectives. Collaboration really
pushes your creativity. If you have a friend who's
really into AI artworks, try to combine your ideas and see what amazing
works you can make. Lastly, make sure
to subscribe to AI art newsletters or follow
the creators on YouTube, reedit and the
platforms we mentioned, who are going to be sharing prompt ideas and best
practices for different tools. With the Midjourney
course completed, you can think of it as the start of your creative
toolkit, not the end. Try to put other AI tools
within your toolkit, and that way you will be able to express your creativity
in many ways. To wrap up, avoid the common
mistakes that we mentioned, stay current with
the new features and keep on practicing. Your projects from
this course are just proof that you
can go from an idea in your head to a full
professional level visual which you can
choose to monetize. The next step is
to keep building, whether it's your portfolio, a small business,
social media channel, or just your own personal
creative journey. You don't have to
share your work anywhere if you don't want to. That's just extra step
that you can take. I hope you enjoy this
course and I can't wait to see what things you guys
create with Midjourney.
43. Class Project: Now it's your turn to create. For the class project,
you're going to be creating an entire visual piece using the techniques and foundations that we have learned so far. You can choose one of the
following project options. Create a dream space
or environment, design your very own fantasy
character or portrait, design some brand
mockups or posters, or do a fully illustrated
storybook or even a series. Start by planning out your idea. Think about the mood and the story you want to
tell with this piece. Then write your prompt carefully using styles, references, parameters, and
consistency techniques you have learned
from the lessons. Generate your images in
maturny, refine them, and if you'd like
to enhance them, export them into Canva or photo when you're ready, upload your project into the
class project gallery alongside at least one
of your prompts and a little explanation as to
what the pieces is about. Remember that there's
no correct answer to this project
because you're just experimenting with
Midjourney and seeing where your
creativity takes you. I will be reviewing
your projects and leaving some feedbacks. You can also look at what the other students of the class have created and make this a learning experience
with everyone else. Sharing is a part of growing and I can't wait to see
what you guys create.
44. Congratulations! What’s next?: Congratulations. You've
reached the end of the course. You've not only learned
the basics of Midjourney, but you've learned
how to apply styles, references, consistency, and even implemented them
in real world projects. As you move forward, keep on experimenting with
different designs, iterate on your ideas and explore new combinations
because that's going to be the key part of learning
how to use Midjourney properly as all of these new releases are
going to be coming out. If you enjoy this class, feel free to look at our other courses here
on Skillshare. Thank you for learning
with Skillademia and thank you for creating. I hope to see you guys
in our next classes.