AI content creation - Video, Audio, Photo, Text + Automate Blogs & Social Media

1. 1.1 Introduction: Welcome to the smartest choice you could make right now. Similarly to how you're missing out if you're not using Cha chi Pet or another LLM. You're missing out if you're creating contents like text, photo, video or audio without using AI. Let's have a look at some AI contents. To start off, I want to play some music that I made in a matter of seconds. Feel rhythm. Move with heat. City of Angels calling me the Sunshine from the west side. They make me feel Oh. This literally took me seconds to create, and it's all fake. It's all AI. Let's have a look at some images. Once again, it's all AI, and you can do this in seconds. Whatever you need if it's realistic looking people, such as these, or logos for your company like this or vintage photos or album covers. Whatever you want to make, you can make it with AI, and you can make it fast. And of course, AI video, which is slowly just getting better and better and more accessible. But it's not only about generating video. You can also automate the entire process of creating a video with clips that are real. In fact, the voice that you've been listening to for the past 10 seconds that was made by AI two. By the way, the background music for this video I made it with AI. But don't worry, I'm going to do this whole course myself. I'm not going to use AI to make this course. According to Forbes, 33% of companies that produce content right now take help from AI. If you make any type of content without letting AI help you, you are behind those 33%. I'm here to fill that gap. Just to be clear, by content, I mean any kind of content. Music, voiceover, images, video, blogs, or any other text or any content that you're creating and putting out somewhere in the universe. After watching this course, you will have knowledge about AI automation in content creation that few have. Will know how to produce high quality and realistic audio and image content in seconds, and of course, save tons of time in your content creation. You will be ahead of all the people who are not leveraging AI for content. We're going to learn how to make content with AI, and we're also going to learn how to automate the process of creating content. We're going to go through all the leading AI platforms for creating image, video, music, voiceover, and text. Then in the end, we will look at how to automate the entire process of creating content, creating blogs, music, or other content, and then publishing it without you having to do with. After taking this course, you will know what platforms to use for what purpose and how to use them. 2. 1.2 Concepts to understand: Before we get into the course, there are some concepts that are good to understand. Creating is now faster than buying. Repurposing content, text controls every other medium. AI enhancement is recreation, not refinement, and of course, AI usage rights. To start off, creating is faster than buying. Before people have paid for stock footage, stock photos, logos, administrative help, royalty free music, sound effects, voice recordings. I promise you that it's currently cheaper and faster to create these for yourself rather than paying for them. This is not a prediction for the future. It's here now. It exists. People are using it, and if you're not using it, I'm sorry, but you're behind. To demonstrate this, I made this photo. This is the death of stock photos and paying for royalty free music. By the way, I made this photo with AI. Perhaps you need a logo for your company, like this one, or like this one, or this one or this one. I mean, really, you can choose between all of them because it took me about 3 minutes to make them. Or maybe you want some music for your next video. Let's listen to a few music tracks that took me about 10 seconds per song to make. Just feel it. Feel pym Move with the heat. City of Angels calling me the sunshine from the west side Dam in the ray Wi Wi AR ray while won that place. I'm cruising las bay in this LA sun Dad They make me feel Oh. O. If I don't really like these songs, then it's not a big deal. I can just spend a few minutes and make a few more of them. Or if I want, I can remove the vocals or the drums with AI. Let's talk about repurposing content. Text controls every other medium. Chachi PT and other LLMs doesn't just make your life ten times faster and easier. It also helps you in producing photos, video, music, and voice because all the other mediums can be converted to text. Once they are text, then you can edit them, and then you can turn them back into image or audio. The best thing about all of this, of course, is that it can be automated. That's why this course includes two full sections on Chat EPT and text because your usage of LLMs, CTEPT will improve all your content creation. It's crucial that you understand Chat EPT and that you use it. What does it mean to repurpose content? It means you take content that already exists and remake it as your own version. Here are some examples. You take other blog posts and make your own blog post from the information in them. You take a video, turn it into a text script, and then rewrite it into your own version, and then you turn it back into a video. You take a photo. Turn it into a text prompt for an image generator, generate a new image that looks the same. You take a song, you analyze the style and lyrics into text, and then you turn it back into your own song with the same style. Or you take a voice sample, make it into an AI voice, and then you make your own voice over with that voice. Text can help you with all types of content creation. How about this one? AI enhancement is recreation, not refinement. What I mean by that is that when you enhance something with AI, it is creating a new thing from scratch. It's not editing the current thing. When you edit a part of a photo with generative fill, it creates new pixels. It does not edit the current pixels. When you create a song with AI, it creates new sound from scratch, sound that did not exist before. When you generate an image, it creates new pixels from scratch, brand new pixels. It is recreation, not refinement. AI usage rights. Commercial usage is allowed for content made by these platforms, mid journey, stable diffusion, 11 labs, Music fi, TO, Sono, in video, synthesia, image, video and audio platforms that we will be discussing in this course, these are the major platforms for AI content creation right now, and they allow you to use whatever you create commercially. So there's not a whole lot of boundaries. Of course, AI usage rights will vary by platform and country. This is not legal advice, but most platforms, when you create AI music, voice, photo or video, you have full free usage of whatever you create. As we said, while the AI models were trained on current content, they are not recreating things that exist. They are creating brand new images, brand new voices, brand new music, that is not the same as anything that currently exists. To finish it off, here are some AI statistics from A Forbes advisory survey. According to this survey, this is how businesses are using AI tools in the following ways. 56% are using AI to improve and perfect business operations. 51% are using it for cybersecurity and fraud management. 47% of businesses are using AI as a personal assistant, which I guess is the most obvious use case for AI. Then there's this one, of course, 35% are leveraging AI for content production. That's what we want to see. If we go ahead and just remove everything else, and we just look at this one. 35% of businesses are leveraging AI for content production according to Forbes. All you got to do is search for it, and you'll find out for yourself that this is true. So one out of three companies that produce content use AI. Now, will this number go down? No, it will not. It will go up. That's why I've made this course because I think when we look at the statistics in a year, it's going to look more like this. It's going to be more like 66% or two out of three companies that produce content will use AI. That's it for this video. I will see you in the next one. 3. 2.1 AI Audio Introduction: Hey, guys, in this section, we're going to talk about AI audio. How to create music and voices with AI. I'm really excited about this because compared to let's say AI video, AI audio has really come a long way already. And there's a ton of use cases where you can already use these AI tools for AI audio. Some of the things that you can do is you can create voices that speak for you. You can give the AI a text script, and it will read it out in whatever voice you choose. You can change your own voice into another voice, just like I'm doing right now. So you can basically record an audio sample by yourself, and then the AI will turn that into Another voice of your choosing. You can use voice samples, or you can train your own voice. You can clone your own voice. So maybe you have a cold, but you still want to make a podcast. That's not a problem. You can create music out of FI air, Music from text prompts, just like you do with photos, AI photos. You can describe very briefly the type of song that you want, and the AI will create it for you. Then there's other specific use cases for AI audio. You can remove vocals from a song, or you can remove only the guitar from a song. Let's say you only want the guitar from a rock song. Without the drums and the vocals. AI can make that happen. Okay, now I'm gonna play you a couple of songs that I made in about 30 seconds. Brody u day. Feel the weight on my back. L a beast, burn my spirit up, tack every step of my tak. Feels like walk in the shales rela less bloods. Testing my son. Every night of screen, my voice lost in knee abyss, on it by the Devon trapped in this black abyss. Eas the golden sun. Feel the ssh bloom the fib colors make come my life. Softly sway Whispers in the captivated. That was AI audio. Quite exciting create voices that speak, change your own voice into another voice, create music, remove vocals from a song. In this section, we're going to be looking at some AI audio platforms that you can use for different things, creating voices, creating music. There's already a lot of AI audio platforms, and I will not be able to cover all of them. Want you to know that it's very easy to find these platforms. You just have to search for them, test them. Now, these platforms that I have listed here, they're some of my favorites. There's some of the most popular platforms, but more platforms are going to show up and there's already a lot of other platforms as well. You don't have to feel that you have to stick to these platforms. You can find other AI audio platforms that might be just as good. However, these platforms, they're good. Just like with video and photo, we're going to be going through them one by one. 11 labs is the leader and has been the leader for a long time for just generating normal voices with AI. I got to say they sound pretty realistic. Music i is a platform where you can change your own voice, and you can also make music. But this one is better for converting your own voice like I just did into another voice. And you can use these voices however you want. Even if they sound like Trump or Spongebob, it's not actually a sample from Trump or Spongebob, so you will have the rights to use them freely and commercially. Sono is a new platform that is really good for music. It's like an image generator. You put in some text and it creates a song for you. The music is pretty good. DO does the same thing as Sono, creates music from a text prompt, and this one is really good as well. We have adobe podcast for Adobe users, really good for noise removal or voice enhancement. This is more if you want to improve the quality of an actual voice recording that you have. And then we have platforms like la.ai, where you can extract vocal or instrumentals from a song. Let's say there's a song from the 80s that you like, and you want the vocals, well, then you can extract all the instrumentals, or maybe you want the guitar solo from a song without the vocals. That's something that AI can do for you. This is very exciting because AI audio has already come a long way, and, let's get into it. I'll see you in the next video. 4. 2.2 (Audio) Elevenlabs: Okay, we're going to start off with 11 labs dot IO, the best and most widely used AI audio generator. Here you can create voices and download them as audio files. It does not create music. These voices will read your text. 11 labs use their own model. It can be freely used for commercial use. If you have one of their paid plans, you get 10,000 tokens for free. Which is quite a lot. That's the equivalent of about 10,000 words. Here's the website 11 labs dot IO. Once you make an account, this is the page that you're going to land on. To try it out, you want to go up here to the menu and click on Speech. Enter a text into this box right here, Text to Speech, and click on Generate Speech. Only about 2 seconds later, we get our generated voice. Let's play it. With a map in one hand and hope in his heart, the man stepped into the unknown, ready for the adventure that awaited. As you can hear, it sounds realistic. It's actually pretty hard to tell this apart from a real voice. On this button right here, you can change the voice. As you can see, if we scroll down here, they have a bunch of voices. You can also create your own voice. Let's test out a couple of other voices. Bill, generate speech. With a map in one hand and hope in his heart, the man stepped into the unknown, ready for the adventure that awaited. Okay, Bill sounds pretty good. Let's test Nicole. With a map in one hand and hope in his heart. The man stepped into the unknown, ready for the adventure that awaited. Okay, that was kind of creepy. Obviously, they have a lot of different voices and a lot of different styles. But I mean, this voice, even if it's a little creepy, it can certainly be used for different creative purposes. Here, if you click on speech to speech, you can record your own audio file upload it here and turn it into one of these voices. In the menu, if we click on Voices, we're going to land in the voice lab. Here you can create your own voice. Maybe you have a podcast or a YouTube channel, or you want to make an instructional video. You could create a voice that sounds just like your own voice, and then you can make scripts and cha GPT or something like that. Give it to 11 labs and it will read it for you, and that can certainly be a faster workflow than recording videos by yourself. Here's an interesting little article page on 11 labs. It's a feature that they have not released yet, but they have a wait list for it. So basically, it's generating sound effects with AI. Here they have generated some sound effects for different video clips that SRA made because all of SARS video clips that open AI released, they did not include any audio. So 11 labs used their new feature that they have not yet released to create sound effects for SARS video. Let's have a look at those. Yes. To be. That's really impressive. That sounds really good. A lot of the times when you're creating a video or a movie or something like that, you'll have a video clip, you'll have to add in the sound effects and find them online, but now you're going to be able to just generate them out of thin air, save a bunch of time. What about commercial usage for 11 labs? Can I publish the content I generate on the platform? The free plan does not include a commercial license, but the paid plans all include a commercial license. Of course, this is also going to vary by each country's loss on AI audio. But basically, you can use their voices however you want and you own all the audio samples that you create with 11 labs. There's not a whole lot of restrictions here. That's because these voices are trained on a bunch of different voices. They're not representative of any one person's voice. They are actually a new voice that 11 lab has created. In the menu, they have a section called dubbing, translate your content across 29 languages in seconds. If you have content for social media, you can dub that content into a bunch of different languages and post the same video in different languages on your social media platforms. Map in one hand. Once you've generated your audio file, you can click right here to download it. And just like that, we have our audio file. And it is hard. On our computer, we can make a video out of it or whatever you want to do with it. Creating audio literally has never been easier than this. There's not that much more to it than this. The platform is super easy to use. It's very easy to sign up. It's easy to test it for free. If you like it, you want to start using it more, then you can pay, get one of their plans. If we have a look at their pricing, there's a free plan and that one even has API access. That one gives you 10,000 characters, which is about 10 minutes of audio. If you want to jump up to the $5 per month plan,'ll get 30,000 characters, which is about 30 minutes of audio, and then it goes on and on and on, the more you pay, the more audio you get, of course. Really, really good platform. I would highly recommend everybody try this out. Even if you're not interested in making audio, even if you're not going to use it for anything, just go in here and test it. This is a revolutionary tech, and I think it's important that everybody understands how easy it now is to create audio that sounds realistic, not only for the purpose of creating content, but also for the purpose, if you hear audio somewhere or maybe you receive a call, where somebody you know is asking you to do something, it's good to know that it might not be real. It might not be created by a real person. It might be AI. Okay, that's it for this video. I'll see you in the next one. 5. 2.3 (Audio) Musicfy: Okay. Next up is a platform called Music f dot LOL. Here you can change the voice of an audio sample that you already have. You can create your own voice. You can create music, mainly instrumental music. They have a voice library of famous voices, such as Sponge Bob Obama, M&M, Shrek, Taylor Swift. Their voices are not direct copies, but made by mixing different real voices. According to Music fi themselves, this gives you the ability to use their voices commercially. Here's the websites, music f dot LOL. You can sign up for free and make 15 audio generations for free. They have an affiliate program and an API call. Here are some of their voices, Juice World, Billy Ish, Sponge Bob, M&M, Donald Trump, Patrick the Str, Peter Griffin, Schreck, Travis Scott, the Weekend, Mickey Mouse, Joe Biden, Obama, Riana, Homer Simpson, Drake, Adele, Katie Perry, Stevie Griffin, Ariana Grande, Luigi, Ted Bart Simpson, Candace Flynn, Finnis, and a bunch of other voices. Once you've made a free account, this is where you're going to land. There's a few options here on the left in the menu. We're going to start off by going to create and then convert voice. Here you can upload an audio file, or you can record audio yourself. You just want to try it out. You can record something, say some words, and then convert it into another voice. Here if you click these buttons, you can remove instrumentals from a song or you can remove reverb or Echo, but we're not going to do that yet. To test it out, we can try out this audio sample that we got from 11 labs with the voice named Bill. It sounds like this. With a map in one hand and hope in his heart, the man stepped into the unknown, ready for the adventure that awaited. Okay, so let's drag that one into music five dot LL. We have now uploaded an audio file. Then we're going to come up here, click on Select A Voice. Let's do Donald Trump. We have the voice, Donald Trump, we have our audio file. Now you just have to click on Generates. Your generated samples are going to show up right here on the right, and up here you will see how many generations you have left. It also shows up down here so we can play this sample. With a map in one hand and hope in his heart, the man stepped into the unknown, ready for the adventure that awaited. So now that we know that it's Donald Trump's voice, it makes sense. It sounds like Donald Trump. But if you would just hear this, you probably wouldn't think that it's Donald Trump. It's not just enough to have an audio file and then convert it into somebody's voice. The tone and the pitch of your audio sample should also be matching your speaker's voice. The voice of your speaker that you have selected, if you want a more realistic pinpointed result. Let's click here on Switch Voice. Let's test out Peter Griffin. Peter Griffin, here's the audio file again. Click on Generate. Map in one hand and hope in his heart, the man stepped into the unknown, ready for the adventure that awaited. Okay, so Peter Griffin from Family Guy, I must say this one did sound like him. If I just heard this sample, I would certainly be fooled to probably think that this was from Family Guy. Okay, let's try something else. Here is a song that I generated with AI. Let's play it to see what it sounds like. Death Glens field of Fido five Colors. Let's go back to create Convert Voice. Drag this song into Music F, and select the voice of Let's do Travis Scott, rap artist. Click on Generates. Let's play this one. Tf call and Sun Feel the Fs pom calls make com Soft sway in. Okay, I wouldn't say that sounds very good, but it certainly did something. Of course, you can't just expect to get a good result every time. You're gonna have to put at least an ounce of creativity on your own into this. But let's try another voice for this song. Switch voice. Let's do Billy Ilis this time. Click on Generate. Let's play this one. Icons make the soft sway in. Okay, that sounds pretty good. And because this was a song, and we did not check this box right here where it says Remove instrumentals, it removed the instrumentals for the song and recreated the vocals with the sound of Billy Ish. So if we play this original song track again, Tic and son. F who the flowers. Of course, we can hear that there's some guitar and other instruments in there. But when we play this new version with Billy Ish. The new. He. It's only the vocals. At this point, I think you get the idea. This is very powerful, can do a lot of things with it, and it actually sounds good. Let's test out one more feature at Music Pi. Right here under Create, we can click on Text to Music. Here's the box where you can enter in a description of the kind of music you want to generate. Here are some suggestions. We can just try one of these. Create a low fi hip hop beat, relaxed and mellow, incorporating vinyl crackles and soft piano chords. Okay, generate that one. Let's play that one. It's not the best music I've ever heard, but it sounds good. Now if you're making a video instead of paying for music or looking for royalty free music online, it's probably actually faster to just come here and generate music yourself if you need background music for a video. Mic, really good platform. Same as 11 labs. It's easy to sign up. It's free to test it out. I would highly recommend that you come here and play around with their tools. See in the next video. 6. 2.4 (Audio) Suno: Next up is a platform called so.com. Here you can create music from text. The music that comes out of this platform actually sounds surprisingly good. You can create songs in any style. You can create instrumental songs. If you need background music for a movie or a video or something like that, this can be really good. You can use your own lyrics, and then you can extend your songs once created. Once you've made the songs, you can make them longer if you like It's free to create an account and try this out. There's really no reason why you shouldn't because this platform is awesome. This is Sonos dashboard once you've created an account and logged in. This is where you're going to land. The website is super easy to navigate here. You can listen to some songs that other people have made. You can go through some genres, categories, listen to songs, and you can even copy the prompts. If I click on these three dots, you can click on this button here that says reuse prompt. If you find something that you like and you want to make something similar yourself, that's a really good way to do it. We're just going to go up here on the left and click on Create. Over here are some songs that we have already made. And in this box for here's where you create your music. There's not a whole lot of settings that you need to be aware of. Can basically just type in a description here and it will make a song for you. But there's a custom mode. We're going to look at that in a second. If we type out here a country song about trucks, country music. All we got to do is click on Creates, and that's going to start creating two versions for us. It's also going to give it a name, this one is called the Open Road. As you can see up here, I've generated two other songs, but it's given me four songs in total, and it's also going to generate a little image for each song. Let's listen to both of these. On Dusty H. Under set song. I'm behind the wheel meaning my red ****. Rolling down the back rows, kicking up some dirt, but the windows rolled down in my favorite country song. She's got the still guitar twin and that whiskey soap draw singing b good on time, That was the first one. Let's listen to the second one. All dusty under the setting sun. I'm behind the wheel meaning my red ******. Rolling down the back road kicking up some dirt. One. I got to say that sounds surprisingly good, and that took me about 10 seconds to make. It's giving us two versions and they're both very similar. The lyrics are similar as well. Over here on the right, once you click on a song, you can scroll down and look through the lyrics here. Here you can click on Extend if you want to extend it. This one that it made for us is 1 minute and 37 seconds long. Of course, if you want to download it, you just click on these three dots and click on Download, and that will give you the audio file. On Dustin under. It's so simple. I love it. If you go to download here, you can even click on Download video. And that will give you the song as a video file. Dust and under son on behalf W sneak. With the image that are generated for you, with the image that are generated for you, the title, and the lyrics. Alright, you can also click on Instrumental. So if we check this box instrumental and we keep the same prompt, click on Create. Let's listen to the instrumental ones that I just gave us. That's the first one. Here's the second one. So I think it's pretty fair to say that the days of Royalty Free Music platforms is over. In my opinion, right now, it's currently faster to generate your own song. For a video rather than to look for it online. Then we have this one up here, which is called Custom Mode. If you click on Custom mode, that's going to enable you to enter in your own lyrics. You can also make your own title for the song, and then you'll enter in the style of the music. What we can do here is we can move over to Chat PT and tell Chapt to write a two minute country song about a Deer in the Headlights. Here's some lyrics from Chachi PT. Let's copy them. Paste them into So Style of Music. We're just going to type out country music. For the title, we'll name it Deer in the Headlights. Click on Create. Let's listen to these. So he Am cultural frozen is that globe midnight star. Shine so bright. I think you get the idea at this point, it's super simple. Of course, it's not quite the same quality as Drake or Riana or Taylor Swift or any top artist. But if you need music or background music for a video or for anything, this is the new way to do it. This platform hasn't even been out for that long, so it's only going to get better from here. That's it for this video, I will see you in the next one. 7. 2.5 (Audio) Udio: Next step is UDO. This one is very similar to Sono. I would say that it's equally good to Sono or maybe even better, but Sono is a little faster at generating the palms. Same platform as Sono, the platform is very similar, basically the same. Udo.com, you can test it out for free, create an account for free, log in, and this is where you're going to land. Same thing here you can listen to some songs that other people have made. Roll down to look through some genres. If you click on these three dots for other people's songs. On DO, you can actually download other people's songs. The prompt box for UDO is right at the top up here. You don't have to click anywhere to access it. If we click on the prompt box, here they have some settings. If you want to make your own lyrics, you can click on the Custom box here and write your lyrics in this box. Here's the button for instrumental music, and Auto generated is just a normal song that's not instrumental and without custom lyrics. Here they also have some genres. For example, we can click on Hip hoop. That's just going to enter in hip hop as the prompt. Here we can add in hip hop, a rap song about Los Angeles. Click on Creates. That's going to start generating your music. If you go over here on the left, click on M creations. Here you're going to see the music that's currently generating. Usually, it takes a few minutes. Here I have a song that I made with the exact same prompt. Let's listen to that one. In the cost with the bob we turn up City lights, we ignite we burn up, chase dreams down to one, no one. Same thing here, you get two alternatives per generation. Let's listen to the other one. I cruise in Les Bay in this T s Day. Go this down tonight ba. I got to say I'm so amazed by these platforms and how good they sound. I don't want you to forget that generating music is just like generating text or photo. It takes a few tries before you get a really good one. Same thing with Cait you might not get the best answer right away. So, you know, don't knock this off if you try it and you don't like the result that you get. Keep trying, and I promise you you're gonna get something good. Let's have a listen at another song that I made. All I do is mix a little by the I'm traveling in the get more de tax sounds like I'm buying. Des on my mind like my custom want some. You can let in lose chickens. You can sit the tunes. It sounds really good. One thing that you quickly think about is the lyrics. They might sound a little bit weird. This is AI generated. So, you know, if you just type in a simple prompt like this, you might not get lyrics that are comparable to a real top artist. But how amazing is it that you can make this in a few seconds. If I just heard this in a car or somebody played it, I probably would think that it was a song from Spotify, a real artist. And there's really not that much more to it than this. There's not much more for me to cover on these platforms. They're so simple to use. So that's it for this video. I will see you in the next one. 8. 2.6 (Audio) lalal: Next step is a website called l.ai. Here you can separate vocals from instrumentals, and it works really well. You can download them separately, so you can download the instrumentals or the vocals. All you got to do is search for.ai. Click on this website. You're going to land on this website and here you don't even need to create an account to try it out. All you have to do is click on this yellow button where it says select files. The default setting is going to be vocal and instrumental, but if you click on this gray button. Here for example, if you have an audio sample where it's voice and noise, you can remove the noise. But we are going to try it out with this country song that we made earlier in So. Just going to play that one quickly for you so you can hear it. A dusty song on behalf. That's what it sounds like. Click on select files. Drag in the audio file for this country song. Upload. It's going to upload on the sites. You don't have to do anything else. It's just going to take care of it for you. Generating previews. And that took about 15 seconds. Now we can listen to the vocals and the instrumentals separately. Let's play it. On dusty under sun on behalf we sn Those are the vocals. So over here on the right, we can mute the vocals and unmute the instrumentals, play it. Then we can test out separating only the drums. So over here on the left, we click on drums, and then we click on Create New Previews. That's going to generate some new previews for us. And that's going to give us the same setup here. So right now if we just play it, we're going to only hear the drums. Then if we unmute this one, where it says without drums, and mute the drums. Play it. Roll down back kick in some dirt the wind rolled back. That's our song without drums. This is simply just incredible. If you want to download the samples, you can pay for one of their packages. Or, of course, you can just use an audio recorder, like the quick time player on Mac, and then you can just record the internal audio on your computer and play the play the music. Under this. That way you can get it for free. That's it for this video. I'll see you in the next one. 9. 3.1 AI Video Introduction: Welcome to the section on AI video. This is going to be very exciting for a lot of people. The value is in understanding what's on the market so that when you have a need, you can choose the right tool. When it comes to AI video, it's not just AI generated videos like these on the right here. AI can help us in different ways with video. There are platforms that actually generate video, similarly to how Mi journey generates a photo. There are platforms that put together video for you, so they leverage a bunch of stock footage, and then with the help of AI, they can put together an entire video for you. Then we have platforms that can edit your video for you or edit parts of your video. Like photoshop generative fill. You can think of this instead of having a green screen, you can use AI to change different parts of a video. What different AI video platforms are out there? If you know anything about AI video, you certainly know about SARA at this point. SRA just like Chat GBT is made by Open AI. In this current moment, I would like to say that it's by far the best AI video generator on the market. But it's not available for public usage. These clips on the right that we're looking at here, they are made by SRA. Then we have a platform called Runway. Similarly to SRA, Runway can generate video. But that video is not as good as SRs video. But runway is like an overall AI video platform. There's a lot of features within runway. For example, you can remove the background. You have a motion brush. You can give runway images and make those images move. Then we have a platform called Synthesia. They focus on making AI avatars. On the right here is Synthesas website, and this here is what an AI avatar looks like. Synthesia is more for instructional videos. And with synthesia, you can make your own AI avatar. You can basically clone yourself. They currently offer 160 different avatars with different looks like this one, and they can speak 160 different languages. Then we have a platform called I Video. This one puts together a video for you. On the right here is in videos website. Here you can enter a text prompt and it will create a video for you. But it's not AI generated video, it's real video. Stock footage clips, and they will put those stock clips together and make a full video for you. They choose between 16 million stock photos and videos. In video can also generate audio and a script for your video. On the right we have D scripts websites. Their features include making your eyes always look into the camera. If you film yourself and you look off camera, they can edit your eyes to make it look like you're always looking into the camera. Be useful if you're going to read a script or something. You can edit your video by editing text. They have AI voice cloning, and they have a green screen effect. If you're filming yourself in your living room, but you would rather make it look like you're sitting out in space, D script can do that for you. Then we have some AI features in regular video platforms like Premiere Pro. For those who are not video creators, Premier Pro is Adobe's video editing software. It's been a leading platform on the video editing market for a long time. They have added in a few small AI features like extending an audio clip. If your video is 5 minutes long, but your soundtrack that you're using for that video is only 4 minutes. You can use their AI feature to extend that audio instead of finding a new song. They have an auto captions feature, and I'm also going to show you another trick how to automatically cut up a video in Premiere Pro. How should you take advantage of this knowledge? Well, most of these platforms honestly are not super good yet. They're in an early stage. But as they become better and better, they will become more useful. If you are an early adopter, you will be in a very good position for the next year. But there are still tons of useful tools already for using AI when creating video. Of course, if you are aware of the AI video platforms on the market, then as they evolve, you will be in a good position to start automating video content. Be automating video content is already a thing, and it's only going to become more of a thing. We're going to cover some of these platforms one by one, SRA, runway, synthesia, talking about the best AI video generator, SRA, talking about runway and all its AI video features, talking about synthesia, and making AI avatars, talking about in video, creating full videos for you. Of course, there's already a lot of different AI video platforms on the market like D script, and they all specialize on something. For those of you who are associated with video editing on a regular basis, if you're not already aware of the AI video features in Premiere Pro, for example, this can be very useful. Real quick, if we head over to Google, search for AI video, scroll down past the ads. The first results that we're seeing are synthesia, in video, SRA, which are the platforms that I've already mentioned. But then of course, the list just goes on and on and on an endless amount of AI video platforms already. Lastly, I just want to say watch what you're interested in. You don't have to watch what you're not interested in. But it's still good knowledge for anyone in the professional world at this point. If you know that you're not interested in AI avatars, you don't have to watch the video on synthesia, but you might still gain a lot of helpful knowledge about AI avatars by watching this video that can help you in your work, or maybe you'll recommend a friend to start using synthesia. I'll see you in the next video. 10. 3.2 (Video) Sora: Okay. Let's talk about Open AI SRA. First of all, you can just search for SRA on Google, go on OpenAI's website. The first thing that you will see if you scroll down is some different videos that have been made with SRA and then you can flip through them and watch them. SRA is created by Open AI. It's not available for public usage at this point. I think we can all agree that it's currently the best video generator at this moment. The quality is just way way better than any other AI video generating platform on the market. SRA training. SARA is an AI model that was trained just like My Journey or Chat GPT. It was trained on an unspecified amount of data. Open eye is rather secretive about this. They don't want to really give out what they trained SARA on, but it's a lot of data, and it's basically video and photo content that they either own or don't own. A supercomputer called Nvidia Age 100 is a popular computer right now to train AI models. One of those computers sell for about $40,000. It's estimated that for SRA training, they needed 4-10 thousand of these computers to train SARA for one month. So let's say they used 6,000 of these computers, that would have cost them $240 million. Yeah, Open eye is secretive about the details. At least at this current moment, they're not really giving out too many details about how it was trained or if it will be released, when it will be released. Of course, there's a lot of safety concerns surrounding this. Here's an interesting fact. According to this study, SRA can generate about 5 minutes of video. Per hour of NVDA age 100 GPU power. Basically what that means is to generate 5 minutes of video like this one. You would need to run one of these supercomputers for an hour. Here is an interesting thought for the future. Before a production company would buy an expensive camera, camera equipment would cost $40,000 or even more. Perhaps in the future, a production company will instead buy a supercomputer or spend this money on renting GPU power in order to generate video instead of producing it themselves. The main point, of course, is that if SRA was released to the public, it would make a huge shift. Not only would it change movie production, but of course, it also creates this whole thought process of, can you now trust videos that you see online? If you see a video of a person doing something, how do you know that it's real? How do you know that it's not AI? Well, that's an issue that hasn't exactly been solved yet. There's really not that much more to say about SRA. You can find SRA videos on Open eyes website. If you haven't already, I would really recommend going to the site open ey.com slash SRA. Just scrolling through this page, looking at the different videos. There are really a lot of astonishing videos here that will make you question what is real and what is not real? What are the possibilities for video in the future? If you can automate something like this, all you're going to need is your imagination if you want to make a movie. Simple as that, I'll see in the next video. 11. 3.3 (Video) Runway: Talk about runway. Quick pitch for runway. It's like a user friendly Hollywood studio in your computer. The website is runway m.com. While this platform might not have everything you need right now, it's certainly a platform to keep your eyes on for the future as AI evolves in video. What can you do in runway? Well, they have a lot of features, but the main ones are the green screen feature, remove or separate background from your subject. Here we can see a little preview of that. You can generate short AI video clips. With quality, the quality is not the best. It's nowhere close to SRA, for example. Think of Runways AI generated video like Adobe Firefox or Dolly, but slightly better. You can change the style of videos. You can make videos look like a painting. You can turn a photo into a video. Basically, you can make a photo move. Image to image. You can transform the style of an image. Think of this like a photoshop edit, but you're doing it with a text prompt. Runway is free for starters, you can sign up, create an account, test the AI video features for free, creating AI videos costs credits, and you run out of credits rather fast. This is the website app runway ml.com. Once you create an account for free, you'll land on this page. If we go over here to the left in the menu and click on Generate Videos, three options. But we're going to click on this one, text to Video. That's going to bring you into this user interface. Up here on the right, you can see that we get 105 seconds of video for free. This is your text prompt box. If we type in here, a guy running down the streets of New York, you can click down here on Generate four S. That means generate 4 seconds of video and then this box is going to pop up on the right. Your video is generating and will be done in a few minutes. Now, if you ask me, I will say that the main use case for runway right now is their green screen feature, separating background from subjects. This is where a lot of video creators could actually find this platform very useful. Our video is finished, let's have a look at it. Here we can see that they're not even moving. We have some flaws going on here with the cars. I would say that it looks okay, but obviously this video is not really working. Videos of people, not the best. But if we try a video without people in it, let's type out a drone shot of the Alps in Switzerland. Generate that. It's going to open up a second box down here. I got to say this looks pretty good. There's a little drone up here, so it didn't really understand that we meant that it was a drone shot. It actually put a drone in the shot. But besides that, I can see this clip being used in a YouTube video or a corporate video or whatever. It looks good. Let's test out the photo to video feature. You can do that in the same usurface. Up over here, you can drop an image. Let's test out this image that we got earlier, drag that in there. Remove the prompt, click on generate. That's going to open up a third video generation down here. Let's have a look at this one. Once again, we can see that it's not so good at generating videos of people. It's certainly making them move a little bit, but it's also making them look unrealistic and warped. It's not good for making videos of people. But what if we try a simpler video where there's no face and it's a simpler background, there's only one subject. Let's drag that one in there. Click on generates. I got to say this one does look better. There's a little glitch down here on the foot. But besides of that, I don't really see any flaws. The conclusion here is that people don't work, but landscapes work and images where there's a simple background and a simpler subject works rather okay as well. Let's go back to the main page and click on the removed background green screen feature down here. Up here, we can see our previous generations. Let's try it with the Spaceman. Over here on the right, we have some instructions, click on an area to start masking. So I'm going to click on three different places like this. As you can see, that places out some green dots. Just like that, we have separated the background from this guy. Now there's still a little space up here that's not selected. I'm going to press on that. Now the whole guy is selected. Now we can go down here and click on Replace background. Throw on a little effect, perhaps. Now once you've done this, you're able to export only this spaceman without a background as a video or as a PNG sequence. Then you can put it on premiere and lay it on top of another video, or maybe you want there to be some text in between your background and your subject. That's another good use case for this. To summarize runway, it's not good for creating videos of people, but it is pretty good for generating videos of landscapes or simpler subjects, and it's good for removing backgrounds. That's it. I'll see you in the next video. 12. 3.4 (Video) Pika: Okay. Onto the next AI video generation platform. This one is called Pika. Can find it on the Url pka dot art. It's a very simple to use AI video generator. Pk have developed their own model. They're basically using their own system to generate video. The quality is not super good, but it has good potential for the future, and it's open for anyone to use. In Pika, you can generate three second video clips. You can make about 25 videos for free. Their platform is only for generating video. They don't really have any other features like editing parts of your video or something like that. Can be used in discord and it can be used on their website, pka dot art. This is discord, and this is what Pka looks like in discord. They have a bunch of chats or servers here. Here you can see all the videos that other people are currently generating. We can click on one of these and this is what it looks like. Here's another one. But we're going to be looking at it on their website instead because that's easier. Part, you can just click on T Pka. Once you've made an account, you will land on this page. Here you can hover your mouse over some different videos to look at different videos that they have created with Pa. As you can see, there's not that much movement in these videos. It's almost like taking a photo and making it move in runway. Down here, we can see that they're 3 seconds long. As we can see here, Pika can generate videos of people, but they're not going to be super realistic and most of them have the same frame. It looks like a close up shot from a movie when you generate a video of a person. Right now we're in the Explore page. If you click on my library, it will take you to the place where you see the videos that you've made. And down here is also the prompt box where you can generate new videos. Here I wrote a guy running down the streets of New York. This was the video that we got. Doesn't look very good. And here I wrote a drone shot of the Alps in Switzerland. This is what we got, and I must say it looks a little bit better. This one I can almost see being used in some sort of video. Let's test out some other prompts. Cinematic shot of a cat in the desert. Click this button to generate the video. Let's try another prompt, the Sahara Desert. Generate that one. You can see here that you're able to generate several videos at a time. Here's a cinematic shot of a cat in the desert. As we said earlier, it just looks like a photo that they gave movement to. Here's the prompt, the Sahara Desert, and of course, that's a shot of the desert. It doesn't look great, but it looks okay. To conclude Pika, I can't see many use cases for it right now because the quality is not good enough. But if they improve their model, Pika has great potential. It's easy to use open for anyone. And if you just want to try to make some AI video, this is a good place to go. That's it. I'll see you in the next video. 13. 3.5 (Video) Genmo: Here's another AI video generation platform. This one is called enmt AI. It's pretty similar to Pika. They use their own model. Doesn't have the best quality, but it's very simple to use. It's free to use, and this one is also good if you just want to try to make some AI video. You just have to search for GenMo create an account or sign in with Google and you'll land on this page. Here I wrote a man running down the streets of New York. This video is not terrible. It's not good enough to use in an actual video, but you can see that it's getting there. I can definitely see how in six months, this could be way higher quality. Here we have the prompt of the Alps in Switzerland. Okay quality, but really doesn't look that good. All you have to do is type in your prompt in this box, click Submit, a cat catching a mouse. This is the video we got from the prompt, a cat catching a mouse. Even if you don't end up using these platforms right now, it's good to know about them. It's good to know what's on the market for AI video generation. Because in a few months or a year, if these models have become better, you will be in a good position if you are an early adopter. 14. 3.6 (Video) Synthesia: Okay. Next up is a platform called synthesia.ai. This platform focuses only on AI avatars. You can create tutorials or instructional videos. The quality is good, but it's not quite at the point where you would want to use it for your business in my opinion. You can test synthesia and create free test videos, but to actually use synthesia, you have to pay. So let's have a quick look at a couple of videos that I made with Synthesia. Hello. Today, we will talk about apples. Apples taste good a fruit that many humans like, but I'm a robot, so I don't know what they taste like. And here's another one. Hi, Anna. Jake wanted to say that you are a rock star. Your contributions are a big part of the team's success. Thanks for all you do. So there's a couple of things to take away from this. First of all, obviously, this video doesn't look 100% realistic. You can tell that something is wrong. He feels and sounds a bit robotic at the current moment. But what are they doing in this video? They're actually adding in some text animations in the background. They added in a little animation of Steve Correll popping up in the background. They have a pretty nice template here with the guy and the blue background and it's moving. I got to say if you watch it with a sound turned off, it looks pretty good. Same thing here if we watch this video with a sound off, I could probably be fooled that this was a real video if I didn't look closely at it. The point here is that these platforms they only getting better. Soon enough, you're not going to be able to tell that this is fake. I think synthesis one of the platforms that has a lot of potential. Here we can make an entire instructional video without editing, without recording anything. Chat GPT could basically write the script for us, and then you throw that script into synthesia. In a few minutes, you have a video that is actually useful that would have taken a long time to make the regular way. I'm very excited for this platform to up their quality a little bit. Acid becomes just a little bit better. I think we're going to see a lot of businesses starting to use these videos, especially when they can make videos like this with pop up, sub celebrities, animated text. These videos can actually be quite engaging. Here's the pricing for Synthesa. In the starter pack, you get 120 video minutes per year, and that's going to cost you $264 per year. Of course, right now, that's a little bit expensive. But if these videos just become a little bit better so that you can actually use them in your business, then I think this is pretty cheap. Let's say you make a video like this for your business. It's a minute long. If you made it the old way, it would have cost you maybe $100. Now you can make it for $260 and you can make 50 of these videos. And it's way faster than making a regular video. The website for Synthesia is synthesa dot O. And if you go to their landing page, you'll find this button right here, create a free AI video. I would recommend you click on that and play around with these different layouts. Test the avatars, see what you think of it, make your own little script here. Okay, that's it for this video. I will see you in the next one. 15. 3.7 (Video) Invideo: Next up is in video dot IO. This platform does not generate video for you. Puts together a video for you. Using stock photos and stock videos. It generates a script and audio for you, and you can use your own script. This is the website in video dot O. You can create videos with text prompts, enter any topic and in video AI gets to work. It generates a script, creates scenes, at voiceovers and tweaks the video at your command. You can sign up for free, and this is what it looks like once you've made a video. Let's play this video and have a look at it. Ever wondered why the Tynosurs Rix, fondly known as the T Rix, has held our fascination for over 100 years. This prehistoric predator, with its massive size and equally massive reputation, continues to captivate us. But what makes the T rex so special? We'll stick around as we delve into the fascinating world of the Tyrannosaurus rex. First up, let's talk about size. The T rex was one of the largest carnivorous dinosaurs that ever roamed the earth. These giants stood at a staggering 20 feet tall with a length of up to 40 feet from head to tail. That's roughly the size of a school bus. Imagine seeing that in your rear view mirror. Next, we turn to the TRx's most distinctive feature. It's teeth. The T rex had teeth that were up to 9 " long, the size of a large banana. Their teeth were not just long, they were also incredibly strong. In fact, they had the most powerful bite of any terrestrial animal that has ever lived. It's believed that a T rex could bite down This video took me maybe 30 seconds to make, and I can already see this being posted to YouTube. There are some flaws in it, of course. For example, here, he's comparing the TRX size to a school bus. In video IO has now chosen a clip of a school bus, which isn't really correct. But what you can do is click down here and edit, and it will open up this editor that will show you all the different stock clips and stock photos that in Video has used to create your video. Then you can go on and replace different clips, upload your own media. But as a ground total, I think it's doing a pretty good job. Maybe you notice that there are watermarks on every single clip. Here it says story blocks. Story blocks is a platform for stock video and stock photos. Here you can barely see it, but it says stock by Getty images. In Video has collaborations with the stock photo and stock video platforms, and they have access to over 16 million different stock photos and videos. Of course, when you pay for this platform, the water marks go away. Once you've created an account on in video, this is the dashboard, the Landing page. Here you can click on Create AI Video. All I did to create that video was to write out a fun instructional video about TRxs. Click on Generate video. It's going to give you some options here. It's going to give you a target audience suggesting the target audience. Right now, it's suggesting dinosaur enthusiasts. You can choose to look and feel of the video, bright, inspiring, professional. Can choose what platform it's for, YouTube, Facebook, Linked in. Let you just click on, Continue, and it's going to start generating the video for you. Now, as it says here, this might take a few minutes, but you really don't have to spend a lot of time on it. I spent maybe 30 seconds creating that other video. The other thing that you can do here, of course, if you click on Create AI video, instead of typing out a short prompt like a fun instructional video on T axis, and then letting in video create the entire script for you. We could go over to Chachi PT and give Chachi Pita prompt, write me a three minute YouTube video script for a fun instructional video on T axis. Here's a script that Chachi Pit made. We can copy that one. Come over to our notes, paste that script into our notes, and then we can remove the things that we don't like and edit the script. Then we can copy the whole script again, come back to in video. Paste it into the prompt box. Down here, you can see that you have 3,600 characters to fill into this box. You're able to make a video of several minutes. Then you can generate that video. I must say even though this platform is also in the early stages, I'm pretty impressed with the results that you get. King was a colossal creature that roamed North America in the late cretaceous period. Despite its puny arms, it was a powerful beast with a hardy appetite for meat and a bite force to match. It's top speed remains a mystery, but it certainly wasn't the slow poke of the dinosaur world. In video, Great example of a platform that's already at a point where you can actually use it. Maybe not for professional videos, but certainly for social media content. This is a good example of how you could automate a YouTube channel. It will be very exciting to see this platform become even better and more useful. But if I tried to manually put together one of these videos, it would take me at least one or 2 hours. Here I could do it in a few minutes. That's it. I'll see you in the next video. 16. 4.1 AI Photo Introduction: All right. Let's talk about how to generate AI photos. First of all, I just want to say that there's a platform for all skill levels. No matter how much time you want to put into AI photos, there will be a suitable platform and solution for you. Before we get into it, let's go to mid Journey's page and have a look at some beautiful photos here. None of these photos are real, they never existed. They are all just generated by AI. Considering the very high quality, I think this is absolutely amazing. Of course, this has already transformed the entire business of photography. Me being an old photographer, this is my favorite use case of AI. Can really see that it's starting to look realistic and the quality is very good. Here are the platforms that we're going to be discussing in this whole section on AI photos, mid journey, Dolly, stable effusion, Adobe Firefly, and Photoshop generative fill. Perhaps you already know about all of them. Perhaps you've never heard of any of them. Don't have to remember all these names right now. If this information is a little bit overwhelming, just be assured that we're going to cover all the info that you need to understand this. Let's quickly go through the platforms. Mid Journey is the number one platform for generating AI photos. The company is also called Mid Journey. Dolly is a very simple AI photo model, created by Open AI, ChagpT is created by Open AI. So you're conveniently able to use Dolly within the chat GPT chat. Stable diffusion is number two when it comes to quality. Stable diffusion is for more advanced heavy users, and the company that created stable diffusion is called stability AI. Then we have Adobe Firefly, a generative AI photo model created by Adobe. This one is still up here, but we're not going to be talking about it that much because it's not that good. Then we have photoshop generative fill. Also created by Adobe. This one does not generate photos, but it's really, really good at editing your photos. What platform should you use and what should you use it for? If you want quality, you should use Mit journey. Mit Journey produces the best quality, that's it. If you want to edit photos, edit something within a photo, then you should use Photoshop generative fill. If you just want to try to generate a photo, but you don't care about the quality, you just want it to be easy to use, then you should use Di within Chachi BT. Or if you're building a platform and you want easy API access, Dolly is also a good option. But we're going to talk about that later. If you want a lot of customization, maybe you want to build an app that generates AI photos. You want to build a specific website that generates YouTube thumbnails, or you want to make AI photos that look like yourself or your friend, or you want something very specific. You also want high quality. Then stable diffusion might be the good option for you, but it's going to take a little more effort. It's a little bit harder to learn how to use it. If you're a content creator or you just want to create high quality photos, you create any sort of content. Then mid journey and Photoshop generative pill are going to be the best options for you. Mid Journey combined with Photoshop generative Phil is a very powerful combination. Both of them are also pretty easy to use. I would say that this combo at this moment gives the most value. If you're a beginner, you don't really care about generating photos, you just want to try it. You don't care about the quality, then Dolly is for you. The only exception, as I said before, is if you want to build an automation or a platform or a website or an app that generates a lot of photos quickly, then you can use Open AI's API, and then Dolly would be a good option. If you're a heavy user, You want customization, you want quality. Maybe you want to build an AI photo platform. Maybe you want to build an app, maybe you just want a lot of control over the photo that you generate. Then stable diffusion is a good option. Mid journey clearly wins the entire competition for generating high quality photos. Quality is going to be the main factor most of the time. Mid journey takes a little bit of learning, but it is easy to use once you get to know it. For heavy users, however, stable diffusion combined with a platform called automatic 11 11 wins. But basically, if you combine stable diffusion with a pretty technical platform called automatic 11 11, this is how you can really customize your AI photos. Stable diffusion within automatic 11 11 can create high quality photos, has a lot of customization options, and it's open source. What about mid journey versus stable diffusion? Mid journey has slightly better quality, it's slightly more realistic. Stable diffusion still looks pretty realistic as well, and it certainly has more customization than mid journey. Before we get into it, let's quickly have a look at some photos from these different models. Because now we are in the visual world, explaining won't do much. Seeing the quality with your own eyes, that's going to make you understand. Right now we are on mid Journey's website. We're on the Explore page, and as you can see, the photos just look astonishing. What I really love about mid Journey site is that you can come up here and search for prompts. If I search for Einstein, I'm going to see a bunch of photos that other people have generated. The best thing is that this is free, and you can download these photos and use them yourself. The person who created, for example, this image, don't hold any rights to this image. Let's search for man on horse. Instantly we get a bunch of results. The good thing about mid journey is that a lot of the times, you don't even have to generate your own image. You can just come here and search for it. Take another great image that somebody else created. If we search for portraits, we can see here that the quality is extremely good in mid journey. People's faces that mid journey generates. They look very realistic. Let's have a look at stable diffusion. All right. These photos are generated by stable diffusion. As you can see here, they look pretty realistic, pretty good as well. Here's one that's called epic realism. Another photo that looks pretty realistic. Usually you can tell that there's something that just slightly off about the photo, or they look similar. They have a similar look in their eyes. This one, for example, it just looks fake. Anyways, stable diffusion, as you can see, also has very high quality. Let's head over to Open AI, which has created D three. Let's have a look at some Dolly three photos. These ones are mainly concepts or art. They're not good at all at generating realistic photos. But if you need something like this, then Dolly could generate that pretty quickly for you. A lot of Dolly photos have this cartoony weird vibe about them. Lastly, let's have a look at Adobe Firefly. Similarly to Dolly, a lot of these ones are going to look a little bit cartoony. Here's a photo of a woman, and it looks pretty good. It looks almost a little bit realistic, but there's something weird about it. Here's adobe Firefly creating a leopard. It certainly looks really realistic, but still there's something a little bit off about it. Same thing here's a photo of a burger. It looks really good, but I could almost tell that it's made by AI. I don't know how I can tell, but I can just tell. Then a lot of the photos from Adobe Firefly are just going to look like this. It's more like concept art. They'll have some common floss like this, the eyes will be looking in different directions. Just to quickly compare them. Here's me using Dolly to generate a photo of a fish eating a carrot. Looks okay. Could be used as a cover or something like that. Here's the same thing in Mid Journey, I ask Mid Journey to create a photo of a fish eating a carrot. As you can see the main difference is just that it looks more realistic. You're definitely going to have to make more photos to get the one that you want, but there's a clear distinction in quality. Let's go through the pricing for the different platforms. Mid journey has a basic plan of $10 per month. Here you can try out creating some photos. This is going to be enough if you just want to generate a few photos. If you want to start using it more, you can pay $30 a month. They also have more expensive plans than this. But unless you're a business looking to generate a lot of photos, $10 basic plan should be enough, or perhaps you want to upgrade to the $30 plan. Dolly is included within the chat GPT subscription, which is $20 per month. Right now there's not really a separate user interface for Dolly. You just use it within the Chat PT chat. If you're paying for CagPT, then you can use Dolly for free. If you're not paying for Chat PT, I would absolutely not recommend that you buy it just for Dolly, because Dolly is not good. Stable effusion is free, but if you use Stable effusions API to integrate into your platform or something like that, then you're going to pay about $10 per 5,000 images. Regular people don't really use this a lot. They make their money mainly off of businesses, integrating their model into their own websites or their own apps. If you've ever seen an app that generates photos of people's faces, they are using stable diffusion. They are not using mid journey, they are not using Dolly, they're not using Adobe Firefly. All the apps and websites that are not any of these companies are using stable diffusions API to power their AI photos. Adobe Firefly is free to use. I think you can generate up to 25 photos for free. Then you can pay $5 a month if you want some more photos. Or if you have Adobe's all Apps plan, which a lot of content creators would have, then you get a lot of photos included in that plan. Photoshop generative fill, the cheapest option for getting it is $20 a month, and that's if you buy the photography plan through Adobe. 17. 4.2 (Photo) Platforms to use: Okay, Mid journey, Dolly, Stable effusion, Adobe Firefly, Photoshop generative fill. We are going to have a look at where you can use and interact with these tools. Of course, if you already know how to use these tools, then you can skip this video. Or maybe you just want to jump directly to the more advanced stuff, where we talk about the features of mid journey and photo creation. Anyways, let's start off with where to use Mid journey. We're going to start with Mid journey because that's going to be the major one that you're going to want to learn. So the place where you use mid Journey is discord. Unless you're already familiar with discord, which a lot of people are, it's not super user friendly. Discord is a social platform for chatting with other people. And Mid Journey have a collaboration with discord. So this is what Mid Journey looks like in discord. It's important to make the distinction that this is not mid journey. This is discord. In order to open Md journey on my computer, I'm going to search for discord and open discord. They are completely separate. But within discord, here you can see the mid journey bot. This is a chat within discord that we use in order to access mid journey. There's an exception to this. If you've already generated more than 1,000 images on your mid journey account, then they currently grant you access to use their image generator on their website. This is currently like a Beta feature. Right now, as I'm making this video, you're not going to have access unless you have generated 1,000 images already. But if you're watching this course one month, two months, or a few months after I made it, then I'm hoping they will have integrated their image generation feature on their website because that one is a lot more user friendly. This is what it looks like using mid Journey on their website. As you can see, right now, we're just on mid journey.com slash Explore. Up here on the left, if I go over to Creates. I'm going to see the images that I have recently created. They look cool, but I promise you it requires no skill to create them. If you have access to using their website feature, right up here where it says Imagine. This is where you enter your prompt, so here I can just type out a cat in a bathtub. Send that off, and that's going to start generating within this Create tab. Here we have it a CAT and a bath tub, 33% complete. There we go. Finished. Cat and a bathtub. Easy is that. Using Mid journey within discord looks like this. You go to the Mid Journey chat within discord. Go down here to type a message. Then you're going to type slash, Imagine, and this little box is going to pop up and you want to choose prompt. To choose it, you can click on Enter, and then you can write your prompt. A Cat in a bathtub, send that off to Mid journey. All Mid journey is doing is that they're using discord to communicate with you. When you send off a prompt like this to Md journey, all you're doing is communicating with their AI model. In discord, you can click on this image. You get four images. If you want to save one of them, you're going to have to click on one of these buttons in order to upscale it to make it bigger. Then you can right click on it and save image. Then that's going to go to your computer. Pretty simple. What about stable diffusion? Where do you use stable diffusion? Well, what you have to learn here is that it's open source, so it's different. There is no one place to use stable diffusion. Mid Journey is the company and they've created their own AI model, and they've been using discord for you to interact with their model, Md journey. Stable diffusion is the model, and stability AI who made stable difusion, they have not made any collaboration with any apps like discord. Instead, they're making it open source, and open source just means that their model is available for anyone to build on. Anybody can build their own company using the stable diffusion model, which means that we can use the stable diffusion model on different online sites. We can use it on different apps. Then there's a platform called Automatic 11 11, which is like the main platform and user interface for stable diffusion, advanced usage. If you really want to customize your images, automatic 11 11 is the place you want to go. When you find stable diffusion within different online sites or apps, usually they don't have a lot of customization. Because companies that are building these sites, they are just taking the model and making it very user friendly. But if it's user friendly, it's not super advanced. For example, if we Google stable diffusion use free, Scroll down a little bit here, pick one of these websites. Here's a website where they have built an integration for the stable diffusion model. Click on Try now, what to draw? A cat in a bathtub, create that one. Just like that, we have a photo of a cat in a bathtub. Of course, this is a lot more confusing than just going over to the mid journey site. That's where I want to show you automatic 11 11. This is the interface for automatic 11 11. This is like a program that you can install on your computer, or you can run it with Cloud hosting. Now, this is just a photo of it, but you can still see that here there's a lot of different settings. Here's where you enter your prompt. You can use negative prompt. There's a lot of settings like width, height, scales, sampling steps. You can do image to. There's a lot of different settings. I hope that clears things up a little bit. I'm not going to be getting into how to install automatic 11 11, because in my opinion, that's a bit too technical for this course, and you're probably better off using mid journey anyways. But it's still good for you to know how stable diffusion plays a significant role in the market of AI photogeneration. If you want to become a more advanced users, you should definitely look into stable diffusion and automatic 11 11. Where can you use Dolly? This one is obviously super simple. You can use it in the chat GPT chat. All you do is tell Chat GBT to generate an image of something. So we're going to do that right now. Generate a photo of a cat in a bathtub. It starts creating the image right away. All you have to do is prompt Cady Pete in order to understand that you want it to generate a photo. Here we go. C at in a bathtub obviously doesn't look as good as M journeys, but still pretty good. What about Adobe Firefly? You can use Adobe Firefly on their website, firefly.adobe.com. The entire user interface is on their website. This is the website for firefly.adobe.com. If we scroll down here, we can see some different images that have been made using this model. Here we can type in Generate a CT in a bathtub and click on Generates. That's going to bring us into the user interface for using Adobe Firefly, and these images, they look pretty good. Now, I don't really see the point in using Firefly, because if you're going to use an external party to generate your images, and you want to take the extra step above Dolly, but you don't want to go all the way to stable diffusion, then you might as well use mid journey instead of Firefly because mid journey has higher quality. Where do you use Photoshop generative fill in photoshop. Generative fill is just a tool within photoshop. It's not an app, and it's good to have some photoshop knowledge if you want to use it, of course. So if we open up this image of the cat in the bathtub in photoshop, basically the way to use generative fill is we can highlight an area like this. Then this little box here is going to pop up where it says generative fill. If we click on it, we can type in remove the plants. It's going to start generating and regenerating and editing the image. It's going to start creating new pixels for this selected area. Bam, just like that, the plants are gone. This just takes photoshop to a whole new level. This is what people have been doing in photoshop for years, but look at how easy it is now. If we do it over here, remove and make the background empty, send that off. You can see that it just removes that whole water faucet. The best part about it is that we had actually selected a part of the bathtub here as well. But the bathtub remained intact. That's because the generative fil model doesn't just remove things, it actually creates new pixels. We didn't actually remove anything. We actually created something. Down here, we have the layers for photoshop. If we remove the original layer, we can see that we did in fact not remove anything. We created two new images that are laid on top of the old one. I think this feature is just fantastic. Let's give this cats a hat. Bam, just like that, we gave the cat a hat. Over here to the right, you can choose because you always get three different variations. I'm happy with it like this. We removed some unnecessary items in the background and we gave the hat a little party hat. Beautiful. That's it for this video and that's it for the cats in the bathtubs. See in the next video. 18. 4.3 (Photo) Midjourney Basics: Let's get into and go through Mid journey, how to use it, the basics, the settings that you can use, and we're also going to go through more advanced settings and strategies for creating the photo you want. Similarly to Chat GPT, Mid Journey can create almost anything if you have the right prompt. If you're not getting the result you want, you might not have the right prompt. These might be the most important videos if you want to create high quality images with AI. So we're going to have several videos on Mid Journey. If you're not that interested in mid journey, then you can just skip these videos. Really quickly, how do you start using Mid Journey and discord? Basically, you log into discord, subscribe to a mid journey plan. Join the mid Journey server or discord. Then you go to one of Mid Journey's channels within discord. Then you use the imagine command within discord. Within the mid Journey chat. Like we can see right here, they have the forward slash imagined prompt and then they're prompting the thing that they want a photo of. I'm not going to be guideing you through the process of downloading discord, but if you want a guide for that, you can just come to this page. It's on mid journey.com, and it's the Quick Start article. An easy way to find it is to just go and Google and search, how to use Md journey in discord. The first article is going to be Mid Journey's website, Mid journey Quickstart Guide. You will have to go to discord.com. This is discords website. If you just search for discord on Google, it's going to be the first link download, and then you can just click on Download Discord, create an account. Even if it's easier for me to go to Mid Journey's website and create images right here in their website. In these videos, I'm going to be showing you how to use Mid journey within discord because one, most of you are not going to have access at this current moment to generating pictures on their website. Two, you have a little bit more customization currently when you're using Mid journey in Discord. But just to clarify on the website, you don't have to type out the Forward Slash. Imagine prompt. Here you can just directly type out what you want. A dog running on the beach. Click Enter and it's sent off. You can see here that four different images are starting to generate. Here on the right, we can see the prompt that we used, a dog running on the beach, and they are finished, we can click on them and look at them one by one. I also want to point out that on the website, they are already up scaled. But as you can see here within discord, when you generate images, you get all four images within one image. And then you have to click on one of these buttons that says U one, U two, U three, four, and that means upscale image one, upscale image two, and so on. If we want to generate an image in discord, we're going to type out forward slash. Imagine, click Enter, then this little box here is going to pop up, type out your prompt, a dog running on the beach. Send that off. Now we sent off our prompt into the chat within discord, and we can see that it's starting to generate 17% finished, and it's finished. Now we can click on it, look at the images. Let's say we like the second one here. Now we're just going to click on U two, scale image number two. It's going to start a new process. It's going to be a little bit faster than generating four images. After a few seconds, that one is finished, we can click on it. You can also double click on it, and then save image, save it to your downloads folder, for example, here's the image. If we check out the statistics on this image, you can see that it's 1024 pixels by 1024 pixels. So it's not super high resolution, but it's still a pretty high resolution. This quality is going to be good enough to use for most use cases. We can access our settings page within discord by typing out forward slash settings in the chat and then pressing Enter. That's going to open up these settings. The first setting that I want you to be aware about is the speed of the mode. We have fast mode, we have relaxed mode, and we have turbo mode. Now you don't have to care that much about the turbo mode because we're not really going to use that one. It's not super necessary. Let's compare the fast and the relaxed mode, how long it takes to generate an image with both of them. Currently, I'm on the fast mode. If we type out forward slash imagine dog on the beach, and then I'm going to change the set up a little bit here. I'm going to open up a timer on the left. Start the timer, head over to my journey and send off this prompt. Let's see how long it takes for this image to generate on the fast mode. The image just finished and we're at about 1 minute. Now we know it takes about a minute to generate one photo in fast mode. I'm not going to show you the slow one because it's pretty slow, but Mjony also has a separate article on the different modes, fast relax and turbo modes. Here you can see that they're telling you that wait times for relax mode are dynamic, meaning they vary, they differ, but generally they are 0-10 minutes per image per job. I've created over 6,000 images, so I can definitely confirm that sometimes slow, sometimes it takes about 10 minutes for one of these to generate on the relax mode. So if you start using Md journey, what you'll notice is that your workflow will be something like this. You'll send off an image to be generated, and then you'll switch to another tab. Maybe you'll go to Chat GPT, ask ChagT something, chat a little bit with Chad PT, and then after a minute, you'll come back to mid Journey and get your image. In the next video, we're going to have a look at all the settings or all the parameters for mid journey. I'll see you in the next video. 19. 4.4 (Photo) Midjourney Parameters: This video, we're going to be going through some of the parameters and settings for mid journey. First off, I just want to show you some random photos that I've made in the last few months with M Journey, just to show the quality they can really reach. Obviously, some of these, they look a little bit AI. When I say they look AI, I say there's something about them that looks a little bit off. You can tell that maybe they're not real. But still, I think the quality is just great, and you can really do so much with this stuff. Getting a photo of a concept that you think about in a few seconds, it's a whole game changer for this industry. I feel like this photo, this is something you would see at a banks website. They probably had to hire some models before. Now you can just create it in 30 seconds, 1 minute. What are parameters in mid journey? They're basically additional instructions, like technical instructions. These technical instructions are going to affect how your image looks. In the last video, we made spaceman Picasso style, and you know at first, it comes out as a square. But then I made it in a different aspect ratio, 16 by nine. Now all of a sudden, it's a wider image. This is like the most obvious settings, but there's also a lot of settings that has to do with the style. There's a chao setting that changes how varied the results will be. There's a stylized setting, influencing how strongly Mid journey defaults aesthetic styles. So I'm going to put the link to this parameter page, the settings page, in the text section of the video that you're currently watching. The way to use these settings when you're using Mid journey is that first you type out forward slash. Imagine, press enter to enter into the prompt box. Then you type out your prompt. I'm going to type out Spaceman. Then you want to type out two dashes in the name of the setting, and then a number to adjust that setting. So let's start off with Aspect ratio. Simply described by Chachi PT aspect ratio is the proportional relationship between the width and the height of an image or screen. So for everybody who doesn't know aspect ratio here are some examples, this is a one by one aspect ratio, right? It's a square. This is the default for journey. So one by one could be 500 pixels high and 500 pixels wide. So this one on the right, for example, it's taller than it is wide. That's because the first number is bigger than the second number. Here are the two most common aspect ratios. These are the aspect ratios for standard YouTube videos, standard movies, standard TikTok formats. If you're making a YouTube thumbnail, you want to make it 16 by nine. Then of course, if we just flip it around, we have nine by 16, that's going to be the opposite, social media stories or videos or TikTok. To demonstrate this further, I'm going to type out a prompt saying Spaceman, and then I have two dashes and then AR, and then five by 20. Then I'm going to flip it around and make a spaceman that's 20 by five aspect ratio. AR, space 20 coll five. Once again, here is the format for creating the setting for the aspect ratio, and then AR, and then a little space, and then the first number, and then a colon, and then the second number. Let's have a look at these and it actually turned my five by 20 into one by four, but that is, of course, the same difference in ratio. That's the same aspect ratio. Here's aspect ratio one by four. You can see it's very, very vertical. Here's aspect ratio four by one, very, very horizontal. Let's have a look at the next setting, which is chaos. This one I actually like to use quite a lot. The chaos setting, the higher value that you have, the more unusual and random the results will be. If you have a lower value, you'll get exactly what you asked for. The value goes 1-100, and you can type out either C or chaos, and then the reference number. Here's an example, we wrote spaceman C 100, chaos 100, the highest chaos value. If we have a look at these four images, you can see that they're completely random. Sure, they all have something to do with spaceman, but they're not quite what you would expect. They're all different variations of the word, spaceman in different scenarios. This is a great tool to use if you want to be creative. Maybe you want to come up with a concept. You want to base it on a spaceman, but you don't really know what you're after. This can be a great way to find visual ideas. On the other hand, here's the same prompt, spaceman with C one, Chaos value one, the lowest value that you can have. And you can see that it generates photos that they all look the same. They all have a similar style. They're all of a spaceman. Here, I want to show you a practical example of a good use of the chaos value. So here we have a prompt that is pretty specific. There's a lot of different instructions. It generates a photo of a woman in a specific dress in a specific setting. The setting in the background is pretty specifically explained within this prompt. The style of the photo is also specifically instructed. If you have a lot of specific instructions like this, then you're better off using a low chaos value like one to get exactly what you asked for. Because here you can see we have the same prompt, but the chaos value is C 100, the highest value. If we have a look at the results, they vary a lot. Some of them still look good. They might still be interesting, but it's not really what I asked for. It doesn't have that specific style. It's not in the same specific environment. Here's the same prompt again with C 50. Right now we're on half of the chaos parameter because it goes 1-100. These results, they're starting to look more like the style that was described in the prompt. But still they're not quite there. They vary a little bit in the style and the look and the type of photo that we get. But here we have C one. We're getting exactly what we asked for. I just want to show you really quickly within discord that you don't have to type out imagine and then press enter every single time. If you're making a lot of images, it's more time efficient to just write forward slash, and then this menu is going to pop up. The first option will be the option that you used last. A lot of the times that's going to be imagine. A lot of the times in discord, it's enough to just type forward slash and then press enter. That's going to open up the prompt box for you. If imagine it's not popping up immediately, you can just type out I, then it's definitely going to be the first option, you can just press enter, and then you're taking into the prompt box. What you can also do is you can copy your whole previous prompt. If you have a long prompt the last one that we just talked about, that has a lot of descriptions, but maybe you want to test out the different chaos values. Instead of sending this off and then going here with your mouse and highlighting it and copying it, What you can do when it's written out is that your text editor is going to be at the bottom here. Either you can just click this thing three times and it's going to highlight the whole prompt and Forward Slash imagine, or when your text editor is at the bottom, you can hold command, shift, and then press the up button. That's going to highlight the entire section, including Forward Slash, imagine. Then if you copy that, you can send that prompt off and then paste in what you copied, then you can just come in here, edit the chaos value to let's say one and then send that off again. If you're generating a lot of images, this is going to save you a lot of time. The next setting that we're going to look at is the no setting. This is basically negative prompting. All you have to do here is type out, no, and then the thing that you don't want. Here's an example of that. Here's a prompt saying beach in the Philippines. If we just type out beach in the Philippines, we get a beach in the Philippines and what are beaches in the Philippines have? They have palm trees. As you can see here, all these four images actually include some palm trees. Then down here, we have the same prompt, Bach in the Philippines, d, no palm trees. We're doing a negative prompt here within this prompt, negative prompt, no palm trees. If you have a look at these images, you can see that it's still a beach in the Philippines, but there's no paldres. The next setting is stylize. As you can see on Mid Journey's website here, you can type out stylize or S, and then a number 1-1 thousand. This parameter influences how strongly Mid Journey's default aesthetic style is applied to jobs. The other thing to note here is that there's a default setting for stylize. If you go to our settings for slash settings, your setting is probably going to be set to stylize med or stylized medium. There are four stylize default settings here, low, medium, high, and very high. Now if I remember correctly, they go something like this, 50, hundred, 251000 for the low, medium, high, and very high. Here's an example. Here's the prompt. Kodak disposable camera photo of a man in the office. This is the photo that I think was the best out of all these results that we got from different styles. Here I didn't even include any style parameter. What's going to happen when I'm not including a style parameter is that it's going to put the stylize at around 100 or something like that because my stylized default setting is set to medium. I would recommend keeping it at medium. Here we have another one where I put the stylized setting to one, S one, and these photos still look good, but you can see that it's taken all the style out of it. This photo, which is the default medium, it has a few more colors. There's at least a little bit of style. But here it's almost like there's no style. It's just a gray photo. Now here's the next example, which is style 500. Here we can really see that there's more things that are starting to happen in these images. There's a lot more colors, there's a lot more items. There's a lot more things going on, more little things to look at in these photos, which could be really good if that's what you want. Personally, I still think this one without the stylized setting, represents my prompt the best. Here's the last example with stylize 1,000 1,000. Here there's really a lot of stuff going on. You can see that the colors are becoming more vibrant. It looks less realistic. It's starting to look a little more like concept art. This is really good to know about. If you're going for concept art, if you want a lot of things going on, if it doesn't have to look super realistic, then you can bounce up this S stylize setting to 500 or 100. If you want it to be very bland, if you don't want any style at all, then put it at S one. But for most of your photos, I would probably recommend keeping it at a default and not using this setting. That's going to make sure that the photo looks the most like the prompt that you give M Journey. Kodak disposable camera photo of a man in the office. I think these photos perfectly represent that prompt. Right. Let's finish this video and go back to the page with all the parameters. As you can see here, as I'm scrolling down, there's a lot of different parameters. If you're interested in Md journey, I encourage you to go read every single one of them. Test them, see what they do. You're going to learn by trying, by testing, by doing yourself. After creating about 7,000 images in Midge journey myself, these are the settings that I care about. The ones that I just talked about, aspect ratio, chaos setting, negative prompting and stylize. These are the settings that I would start testing out first. Once again, I'm going to put the link for this article with all the parameters in the text section of the video that you're currently watching. You're interested in M journey, I encourage you to go have a look at all these parameters. That's it for this video. I'll see you in the next one. 20. 4.5 (Photo) Midjourney Adjusting Style: In this video, we're going to talk about how to make realistic people and how to prompt to adjust the style of a photo. Over here on the left, we have some notes of things that you can reference to change the style of a photo. This will also affect how realistic the person in the photo looks. Here we are in discord. I'm going to be using the example of prompting for a 30-year-old man. And then down here, I'm going to add these different sentences to that prompt and show you the difference. Before we get started, here's a list of ideas for things that you can reference in your prompt in M journey that will change the style of your photo. The first thing and my favorite is to reference a photographer, like a literal photographer. Off the top of my head, I don't know any professional photographers, but you can easily look some of them up and choose one whose style you like and then just reference their name. Which is what we're doing down here, Patrick De Macharlier. That's a photographer that I found. I like his style. We are referencing him in these photos here. You can reference a movie director. You can reference a specific camera, type of lighting type of photo, and of course, the style itself. To start off, we have a prompt here where we have simply just said a 30-year-old man. If we have a look at these photos, these people, they look good. They look pretty realistic, but it's very boring. Like this one down here does not look realistic actually, I think it looks fake. You can tell that their skin texture is a little bit off. They don't have that deep quality, that deep skin texture that we're looking for, and they don't look hyper realistic. Next up, down here, we've put in the same prompt, a 30-year-old man in the style of Patrick Demacarer. I'm probably saying his last name wrong. Real quick, if we copy this photographer's name and go to Google and search for it, you can see that all of his photos are in this black and white style. When we ask Mid Journey for a photo of a 30-year-old man in his style, that's precisely what we get. I mean, compare these photos with these photos. It's a massive difference. Next up, we are referencing a movie director, Guy Ritchie, a famous movie director from England. If we search for Guy Ritchie movie shots. This is the kind of picture that you would expect. This is the kind of picture that Mid Journey will give us when we reference Guy Ritchie. Here are the photos that we got. Now, these guys are not holding shot guns. It's not 100% comparable, but the style is definitely different. They look a little bit shady, a little mysterious, just like his movies. This one was a 30-year-old man in the style of a Guy Ritchie movie. Next up, we are referencing a camera. The camera that we are referencing is a cannon R five, and then we're also adding in aperture 1.2, which is a camera setting. Here's the result that's not quite getting the aperture right. But it does look like a photo taken on a canon R five. The Con R five is a pretty standard high quality digital camera. I think Mjne is doing a good job of referencing that camera here. This prompt was a 30-year-old man shot on Con R five aperture 1.2. You might notice that it says style raw at the end. That's because the default setting is set to style raw, which it should be. Next up, we are referencing type of lighting. We are referencing Golden Hour lighting. The prompt goes a 30-year-old man Golden Hour lighting. Of course, we get a man with Golden hour lighting. Perfect. Now, the lighting is, of course, a really big part of any photo. I just want to show you a couple more examples of lighting here. Here we have a 30-year-old man soft Studio lighting, and it's really nailing that soft studio lighting. This will create a great portrait photo, just prompting soft studio lighting. Same thing down here, a 30-year-old man soft natural light. I think these photos they look really good. The only difference between these photos and these photos up here, which I don't like at all, is that down here we asked for soft studio lighting. Next up, we're going to reference the type of photo. For that, we have two examples, we have Kodak disposable camera photo, and we just have analog photo. Here's a 30-year-old man Kodak disposable camera photo. I think this prompt the Kodak disposable photo. It always makes the person look really realistic because it takes away that sharp AI look. It gives you that soft film look. If you want to make realistic people, would really recommend using this prompt down here, dated posable camera photo of, and then just describe whatever person you want it of. Here we have analog photo, a 30-year-old man analog photo. I think these also look pretty realistic, like the people, the humans, they look realistic when you prompt it to be in the style of a film photo. Because as I said, the film look isn't so sharp, it doesn't have that AI look to it. Just referencing the analog photo style changes also the look of the human or the person in that photo. Last but not least, you can reference different styles. Here we're going to put Retro music video album cover style. 30-year-old man Retro Music Video album cover style. Wow, these look so hip, so cool. They're not super realistic, but they are cool. I don't think anyone can deny that. The main takeaway here is that every single word in your prompt affects everything. Every single word in your prompt affects the look of the person. This is what Mi Jerne thinks a 30-year-old man looks like, a guy with a beard and a T shirt. But when it's a 30-year-old man in a guy Richie movie, he's wearing a suit or a tie glasses. He looks a bit more cool, a bit crueler. When it's a Kodak disposable camera photo of a 30-year-old man, he looks more chill, more flowy. Maybe he's a tourist like this guy. He looks a little bit more down to Earth because I guess that's what Mi Jerne thinks a guy that would be in a Kodak disposable camera photo would look like. The setting environment, clothing, look of the person is affected by each word in your prompt. That's it for this video, I will see you in the next one. 21. 4.6 (Photo) Midjourney Upscale Options: Okay, let's quickly cover these settings, the options that you get once you have generated an image. What you'll see here is U one, two, three, four, and then this re roll button, and then V one, V two, V three, V four. The ones that you're going to use all the time are these upscale buttons, u one, u two, u three, four. That just means upscale one, two, three, four. So one upscale one, that's going to be this image and then the second one is the one on the right. This bottom left one is third image, and this is your fourth. What happens when we press on scale? Well, as we know, for starters, we have all four images within one image. If we click Open browser on this image, we can see that it's just one photo. It's four photos in one photo. But once we have clicked on this upscale button, at the bottom of the chat, we'll get the full image. What about these V buttons? If we click on it, you'll see that it says remix prompt. That's why I'm calling it the remix button. Even though it says V. Here you can change the prompt. But you're not rerolling the entire prompt that you can do by clicking on this button. If you press the V button or the remix, you're remixing this specific photo. Let's do that without changing the prompt at all, and then let's do it one more time, but we'll type out in nature. Add that to the prompt. Now it's a 30-year-old man soft studio lighting in nature. Send that off. Here's the first upscale of this photo. Down here is the remix of that photo. You can see that these guys, they don't really look like the same person. They're very similar, but they do have some different facial features. They're looking in slightly slightly different directions. It's not the same photo. They're just making new versions of this photo, which can be really good if there's a little tiny detail in this photo that you don't like, but you like the photo itself. Then you can just remix it, click on the V one button. You'll get four more options that look just like it. Out of those four options, there's probably going to be one that's good. What about down here, we remixed it and we added in nature to the prompt. Now they've definitely changed the photo more. Obviously, we can see that it's now in nature or outside, still has the same lighting. The guy still looks pretty similar, but obviously now he looks more different compared to this guy. When you upscale a photo, this is an upscaled photo, then you get all these other options here. If you remix a photo, then you just get the same options again. This one here, the circle or button. That's just a rerole. Then you're just remaking the whole prompt. But what about all these settings for an upscaled image? Let's start off with this one, very region. What that allows you to do is to edit a specific part of the photo. If we click on it, it's going to open up this editing box. It says, drag and select the area to be changed. I'm going to select his eyes. Then down here within the prompt, after it says a 30-year-old man. I'm going to add with glasses. Now it's a 30-year-old man with glasses, soft studio lighting. Send that off. We can see that it's generating and boom just like that, now he has glasses. You can see here that we didn't get the full glasses on all of them. That's because I actually did not select the entire area all the way back to his ears. What about the Sum out options here? We have three of them. We have Sum out two x, one x, and custom Sum. Click on two x, we don't get any options, it's just going to do that for us. If we click on 1.5 x, same thing, no options, it just does it for us. It starts a new generation down here. But if we click on Custom Som, the prompt box is going to open up again, and you can see here at the end of the prompt, it's added Som two. Now, here's where you can edit the Sums. You can edit it 1-2. I'm just going to put it at Som 1.2. Submit that. If you click on the Custom Sum button, of course, you can also change other things within the prompt. Let's have a look at the results. This is the original image. Here are the options where it's somed out by two x. All it's doing here is that it's taking the original image and then it's adding more to the edges. Obviously, this one is pretty simple because it's just black. But if it's like in nature, then it's going to add a bunch of stuff on the sides as well. Here's the one where it's Zoom out by 1.5. What about these arrows? What do they do? They basically change the aspect ratio of your photo and I I press on this down button, it's going to add space underneath this photo. Let's test that out. Let's press the down button. Submit that. Let's also try the one where we press the left button. The left arrow, submit that one. Here are the results. As you can see in the prompt, it's also changing the aspect ratio. It's actually adding a AR two to three aspect ratio setting. Then it's also filling in like the rest of his shirt and his arms down here. This one, obviously, just add some black space on the left. What about these very buttons? Very subtle, very strong. Let's test them out. Click on it, Submit that. Very strong. Submit that one. What they're going to do is change this image and give you some more alternatives. This one is subtle, and this one strong changes the image a lot. Here are the results for variations subtle. It looks pretty much the same. In this one, the guy has a weird eye. In this one, his face looks slightly different. They're very similar, but once again, if you have a flaw in your initial image, you can do this and potentially fix it. Here are the variations very strong results, and they still look very similar, but they are a little bit more different. Here you know their pose and their eyes, The pose has changed a little bit more than in the other one. I want to show you another example quickly of the sum out feature that demonstrates this feature better. Here I've generated a photo of fish in the ocean in a coral reef. This is the image that we got, and then we summed out by two x, scaled one of the results. Here's the next image that we got. Summed that one out by two x again, and then here is the third result. Now we've summed out by two X three times. The cool part about this is when we download these photos and we look at them next to each other. You can really do some cool things like this, just summing out, starting off really close on something, and then you just back out. I'm sure you can come up with your own creative ideas of how to use this. That's it for these settings if you want to look at your photo on the web, just press on this web button and it's going to take you to mid Journey's website. See in the next video. 22. 4.7 (Photo) Midjourney Describe Feature: Okay, let's have a look at another feature, the describe feature. This one is super useful. This feature does not generate a photo. It generates text from a photo. It looks at an image and describes it with text. So here I gave M Journey this really iconic photo that a lot of you will probably have seen of a bunch of construction workers. Sitting really high up in the air. It's giving me four variations of text describing this photo that you can copy and use within your forward slash imagine prompt to generate a photo. That's just what I did, and this is what I got down here. How does this work? You can give it either an image or a link to an image. Here's something that's good to know about. You can copy the link from any mid journey photo. Let's go back to this photo and if you double click on it, you're going to get this menu and down here at the bottom, you'll see copy link. You can also click on Open Link. That's going to open up this photo in the browser, then you can double click on its copy image address, that's going to be the same thing. Then of course, you can also double click on it and save this image. Once you have an image saved on your computer or you have the link, you want to type out forward slash D. The first option should be Forward Slash Describe. If it's not, you can just type out forward slash, describe, press on Enter. Then two different options here are going to pop up. You're going to get one option that says image and another option that says Link. We're going to start off by pressing on the link option. Now this little box down here opens up. This is where you want to paste in the link for this photo. Now we can double click on the photo again. Copy the link, go back to the box, paste it in there, send off that as a prompt. Let's do it again, but with an image. Forward Slash D, describe up here is the first option. Press Enter, click on Image. Now we open up a folder on our computer. Here's the photo. Drag that into this box, go back to mid Journey and press enter to send that off. Now it's given us two descriptions for the same image, and each time you do this, you get four different descriptions. The nice part about this is that you usually get them in different lengths. As you can see here, this one is shorter. It's about two rows. This one is really long. This one is long, but not quite as long. So you have some different options. What I recommend you do is you copy them, go to the imagined prompt, paste it in there. Send that off to generate that image, Copy the second one, send that off as a prompt as well. Copy the third one, send that one off as a prompt. We've now sent off all four of these descriptions that we got as prompts to generate images. While they're generating, let's have a look at these descriptions because these descriptions give us a really good idea of how Mid Journey interprets text into a photo. What are the components that Mid Jerne thinks of itself when describing a photo? First off, portrait of a man with a beard, that's pretty obvious, describing the subject, the person in profile, looking away from camera. Describing the person even more. What is the person doing in the photo? Wearing dark gray sweater. Pretty obvious. If you don't do that, then this sweater could have had any color, dark background. Yeah, it's pretty dark background. Moody soft light shot on cannon EOS R five. Lastly it's describing the lighting and the camera, just like we did. If we have a look here, we can see that it's describing the camera in a lot of them here. It's also using R five camera. On the third one as well, it's also using R five. Here on the other example, also canon ES R five. There's something about this image that Mi Journey interprets as being taken on a Con R five camera. Interesting. Let's have a look at the four different results that we got. This is the first one. Here's the second one, here's the third, and here's the fourth one. Now what you can do is you can go through these, pick out your favorite. Let's say the first one is my favorite. Now you have a prompt that you know is going to work. Well. We can use it again. We're going to get a very similar result. You can also have a look at this prompt and see what is it about this prompt that makes me like the photos that it's producing. Portrait of a man with beard, dark gray sweater, profile shot, soft box lighting, 50 millimeter lens, F 2.8. It's describing the lens, Sansom camera settings, the type of lighting, man with a beard. Let's try it again with this photo. This is an iconic photo. It's been famous for many, many years. So we're going to type out forward slash D, press Enter. Click on Image, and then drag in this image. Click on Enter again to send that off as a described prompt. Let's quickly read some of these. A group of construction workers sitting on the edge of a skyscraper eating lunge, Black and White photo in the style of Latex dripping photography. Then it's also going to describe the exact aspect ratio of this photo. In this case, that's going to be 54 by 29. Let's try this prompt as imagine. What about this one? A group of construction workers sitting on the edge of an iconic skyscraper eating lunge and laughing in the style of vintage Black and White photography. The scene captures their relaxed yet brave demeanor as they rest during work hours, surrounded by city scape. Let's try that one as well. Let's try this long prompt up here as well. Forward Slash, imagine, send that off to be generated. You get the point. If you have a photo of a dog like this and you want to recreate it. Instead of testing different prompts, writing them yourself, just send this photo into the described feature a couple of times and then test all the variations, and that's going to be a much faster way of getting what you're after. Okay, here's the result from the first one. Now, as you can see here, it's not really realistic. Some things are just off about these photos. That's just AI. That's just how it's going to be. But if you generate a lot of them, a lot of variations, there's always going to be one that stands out, one that's good, and that also doesn't have any flaws that looks realistic. I mean, this one down here looks a bit more realistic, or this one here could be a real photo. If you just saw this on a random website, you probably wouldn't think about it, you would just think that it's a real photo. Same here, this one maybe looks a little bit off, this one also not super realistic. But this one here, for example, if I saw this on a website, I probably would think that it's a real photo. That's it for this video. Hope you liked the described feature. I'll see you in the next video. 23. 4.8 (Photo) Stable Diffusion: Stable diffusion, Dolly and Adobe Firefox. We're not going to go through all of them in detail like we did with M journey because you're better off just using Md journey. But let's cover the basics for them quickly so you know what's on the market. Starting off with stable diffusion. Stable diffusion is made by stability AI, and this is their website. What is stability AI and what do they do? Here we can see that they have image models, they have video models, they have language models, and three D models. The main thing that people care about stability A four is their image models and perhaps their video models. The first thing to understand is that there's not one place to use stability AIs models. They are used wherever people build applications with them. Here we can see that they have different image models, video models. If we go back to their main page, the first button that you see here is get started with API. We're going to talk more about API later, but API is basically what companies use to build their own platform with stability AIs different models. This is their main way of making money, not through subscriptions. But if we click here and get your membership, we can still see that they have a pricing plan. This plan applies to people that want to use their models for commercial use. Here we can see it says non commercial. If you don't intend to make money off of their models or the things that you create with their models, it's all free. If you do intend to make money with their models, then you should get this professional plan for $20 a month. Now, they release new models all the time. For image generation, there's a lot of different models available. These models can be good at different things. Here's a little example. Here's a website called Hugging Face. Hugging Face is like an AI community providing different models to anyone who wants to download them or use them. Here's a page that I'm going to link in this video. To a model that somebody built a little user interface for. Back to stability AIs website. Here's somebody that has taken stability AI's API and used it on this little site on Hugging Face. What this model is specifically good at is that you can upload a photo of somebody's face, and then you can upload a photo of a body pose. Then the AI will create a new photo with that face and that body pose within a new environment. That's what I did. I uploaded a photo of Obama and then I uploaded the photo of this guy standing with his arms out like this. This year is the photo that the AI model created for me. The prompt was just Obama with his arms out. Obviously, it's a Chinese setting here as well going on. Now, even if this photo is not great, the quality is not so good, and the photo is weird. You can see here that we have a use case for stable diffusion that we could not do in mid journey. Let's say you have a YouTube channel and you want to make thumbnails for your videos like this. You want your thumbnails to be a view, and you want to have a specific pose and you want the photo to be somewhere specific, maybe on a mountain. This model could do that for you. Now Mid Journey could create a really high quality photo of somebody on a mountain, but it can't get your face in there, at least not yet. Here's another site on Hugging Face another stability AI model. This one is called stable video diffusion image to video. What you do here is that you upload a photo and it turns that photo into a video. They're not showing the original photos here, but basically these little videos here were just photos to begin with. Here's a photo of automatic 11 11, which is the main platform for running stable diffusion models in. Basically, if you want to use automatic 11 11, you would be able to download a bunch of models from hugging face or another site and then choose between them up here. Then you have an interface here for generating images with a bunch of specific settings. You could take this model where you're able to recreate a face and use it within automatic 11 11. Here you have image to image. Now, automatic 11 11 runs locally on your computer. That means it's like its own program that you have to download on your computer. Automatic 11 11 itself is not made by stability AI. It's a separate platform where you can run stability AI models inside of it. What you can also do is you can run it on borrowed CPU power, borrowed computer power. The most common way to do that is to use Google Co lab. Most people are not going to want to use stable diffusion, since it's for very specific use cases. In my opinion, it's very tech heavy and it's a complicated installation process unnecessary for most people, which is why we will not cover it in this course. Besides that, reasons to use stable effusion would be you can do image to image. It's free, but you do need a Google Colab subscription, which is about ten or $20 a month, or you need a really good computer. Let's say you have an old Macbook error, you're probably not going to be able to run it on that computer. There's a lot more customization at least within automatic 11 11, and there are more models for different use cases to choose between. That's it for this video, I will see you in the next one. 24. 4.9 (Photo) Dall E: Let's have a quick look at generating photos with Dolly. As we said before, Dolly is created by Open AI, and you can only make photos with Dolly within Chat GBT. Unless you're using Open AI's API for Dolly. That would be if you're building your own application. The first thing I'm going to do is to come over to Mid Journey and have a look at one of these photos that we made earlier. Go to copy this prompt, a 30-year-old man soft Studio lighting. Then in chat GPT, I'm going to say generate a photo of a 30-year-old man soft Studio lighting. Send that off into Chat GBT, and it's going to start creating an image for me. Here's the image. It looks pretty good. It doesn't look like a real person really, but it's still a good photo. Comparatively to this one on Mid Journey, I would say that obviously this quality is way better. This one actually looks realistic, and Dolly only gives us one photo. But Mid journey, Firefly, and the other ones, they give us four different photos. We could, for example, say make his hair blonde and send that off into Cha GBT. That's going to start creating an image again. And here's the new image. It says the image has been updated with blonde hair. We can see that the guy in the photo, they looks pretty similar, but this one has blonde hair. Obviously, that's working out rather well. I'd like to remind you that you can only use Dolly on the paid version of Chat GPT. If you're not paying for Chat GPT, you cannot do this. Let's test out one other image. We're back on mid journey and here's the black and white photo of some construction workers having lunch amongst the sky scrapers. I'm going to go ahead and copy this prompt. In Chat PT, I'm going to say generate a photo of and then paste in that prompt. Send that off to Chat GPT. And here's the image, and this one doesn't look that bad. I think this photo is actually quite nice. However, it looks a bit more like concept art back here at Mid Journey. This photo looks like it could actually be a real photo. It looks realistic. This one looks more like a painting or something that somebody put together. Basically, I would say that the conclusion of this is that Dolly is good for really quick short usage if you just want a quick photo of something. Before ending this video on Dolly, we're going to go back to Open AI's website. Click on Log in. Then once you log in, you'll get the option to either go to Chat PT or go to their API page. This is their API page. We're going to talk more about this later. Basically, what you can do here in the left menu, you can go to API keys, generate what they call a secret API key, and this is what you can use to integrate Dolly into your own automation or application. Let's say you want to automate making ten blog posts every day. You want each blog post to have a photo that looks like this. It's very simple with Dolly with the use of these API keys to make an automation like that that can create photos for you automatically. Now, this is the only point where Dolly wins over all the other platforms. Because pen AIs API API access, their API keys. They're extremely simple to use. That's it for this video, I will see you in the next one. 25. 4.10 (Photo) Firefly: Okay. Let's test out Adobe Firefly. To use adobe Firefly, just go to firefly.adobe.com, or you can just search for Firefly and you'll find it. If you don't have an account and it says sign in up here, when you try to generate a prompt, it's going to tell you to create an account. But you can make an Adobe account for free and then you'll be able to test out Adobe Firefly for free. We're back in mid Journey, and we're going to try the same prompt again. Copy this one, a 30-year-old man, soft studio lighting. Go back to Adobe Firefly, paste that in there, send off that prompt. Are there salts for this prompt? Straight on shots of some guys and they don't look good at all. I guess you could probably use them for something, but, if you have the option of choosing between this and this quality, that's a pretty easy choice for me. The thing that I do like about Adobe Firefly is their platforms user interface. Up here, we can quickly change aspect ratio. Right now it's 16 by nine. Click on that one and make it into portrait three by four. Generate a new photo. If you're not familiar with aspect ratios, this could be a good place to just come to learn what different aspect ratios look like. Here's the new results, and I got to say these people do not look good. Back in mid Journey, let's try out this prompt again of some construction workers. This prompt is longer and more details. Let's see if we get some higher quality results. Clearly, it does not want to give us high quality results. These people's faces are warped. It's not even what we asked for. At this point, it should be very clear to you why mid journey is the obvious choice for generating photos. That's it for this video. I'll see in the next one. 26. 4.11 (Photo) GenFill: Adobe Photoshop generative fill for editing a part of a photo. Anyone who uses photoshop already knows about this. If you use photoshop and you already know about generative pill, then you can probably go ahead and skip this video. If you don't know about generative fill, it will be good education for how easy it is to now edit photos. For example, this is how you can use it. This is photoshop and here's a photo of a hamburger that I made in mid Journey. This is AI made. If I want to remove the tomato and this class of drink, I can highlight them and fill into this box, remove and fill with black plain background. Send that off as a prompt and it's going to generate some new photos to put on top of this original photo. Boom, just like that, they're gone and we get three new alternatives for how we want this photo to look in the specific places. Now, this is the layer for the original photo. If I remove it, you can see that it doesn't actually edit this original photo. It creates two separate new photos that are then placed on top of the original one that match the background, and it works really well. Here's another photo that I generated in mid journey. Let's say I want to turn this wall into a window. I'm going to highlight this wall and say, make this wall into a window with a beautiful view, send that off as a prompt. Just like that, it's created some new windows, some new alternatives for how this whole photo could look. As you can see, this little part here on the right, we did not highlight that part, so that one stays the same for all three alternatives. It's only this part of the wall that we highlighted, that's going to differ in the results. This is what the new image looks like without the old background. I want to show you a little clip here. Here's a video of a woman running on a golf course. What this guy is going to do is he's going to remove the background and create a new background with the help of photoshop generative fill. Then he's going to place the new photos on top of the video. He's creating this whole setting with a cliff and a little castle here. Then now he's playing this video again. This upper part here is just static. That's a photo. But this part down here is still a video, and then you can play it as a video and it looks really cool. How do you use generative fill? This is not going to be a photoshop course, but I'm quickly going to show you how to use generative fill. This little box here that I'm currently moving around, it's called the contextual taskbar. If you're in photoshop and you don't see it, you need to go up here to window, open up this menu and then down here, you'll see contextual taskbar. If I unclick it, this one is going to disappear. I'm going to click it again to bring it back. And this is where you use generative fill within Photoshop. In order to use generative fill, you need to have a selection. You can either click here on select subject, and Photoshop will automatically select the person for you, just like so. You can also come up here to the menu and choose the Marquee tool. The rectangular Marquee tool, for example, now I can drag a box here over the American flag. Once you have a selection, this box, the text in it is going to change to say generative fill. When you click on generative fill, that's when this prompt box opens. That's when I can type in remove flag, fill with plain background and send that off as a prompt to start generating Boom, just like that, we have removed the flag, and all we have left is a plain background just like what I asked for. You might remember that his shoulder was part of the selection as well. If we toggle through these results, you will see that this part of his shoulder is also changing between the three different variants that we got from generative fill. You can also use the Lasso tool from the menu here in order to make a selection, and then you can drag something out. Let's say we have one selection here. If we want to make another selection over here, what you have to do is hold shift and then draw out your other selection. That's going to add to your current selection. Up here, you'll have the option of adding to the current selection or subtracting from the current selection. I'm going to tell it. Add some devil horns, and this is what we got. I'm so sorry about this Obama, by the way. Over here on the right, you'll see your three alternatives. You can also toggle through them by clicking on these arrows in the little contextual task bar. Let's do one more example of a use case where you can use generative fill. Here's a photo of a Spaceman that I generated with M Journey, but it's in a standing vertical aspect ratio. Let's say it's from my YouTube thumbnail and I want it to be an aspect ratio 16 by nine. Then I'm going to change the aspect ratio here and drag this photo over here to the middle. One way to do this before we had generative fill would be to copy a part of the image, drag that one over here and make it bigger. But now, as you can see the shades are different. You can tell that it's not one photo that it's been edited. It doesn't look that good. Instead, it's much easier to just use the Marquee rectangular tool. Highlight this part of the image, hold shift, highlight the other part of the image. Go to the generative fill box and we don't even have to type anything in. We can just click on generative fill and then generate. Now, as you can see, we have a completely clear background. It matches, it doesn't have any weird shades or lines. As you can also see, we filled the empty background without typing in a prompt. Now the way that that works is if you don't type in a prompt, it's going to either remove the object that is selected. It's going to fill in with something very simple. In this case, we already had a white background, so it was pretty obvious that when we highlighted the other parts of the image that were empty, we wanted to fill that with more background. But if I highlight the spaceman and click generative fill and then click on Generate without typing in a prompt, the spaceman disappears. I just removes the object that we selected. Say we go back to this bathroom that I generated at mid journey, we highlight this little stool here and just click on Generate without typing in a prompt. It's going to do its own thing. This is the result that we get. It doesn't remove the object, and that's because there was a few too many things going on here. But let's say we go up here and we try to just highlight these lights up here. They're very simple. They're in the middle of this black wall. There's nothing else going on here, we click on generates. This is the result that we get. It removed those lights entirely, filled it in with black background. If you're highlighting this area, for example, where there's a lot of things going on here, It's not going to remove everything and make a plain background. Now, as you can see, when we highlighted these lights up here, and the lights are the only objects, except for this black roof, Generative Phil makes the decision to just remove them. The other thing that you could do is you could make several little boxes like this. Here we have seven different little boxes that are selected currently. Click on Generative pill, and we're going to type out make miniature planets and moons. Send that off as a prompt, and these are the options that we're going to get. It's not 100% what we asked for. We did have some selections down here as well. This tool is not perfect, but it is really amazing and it can speed up your photo editing process like crazy. It also completely revolutionizes what you can do with a photo. Just in 1 minute, this is what we started with. Here we go. I'm already ready to post this as a YouTube thumbnail, for example. So Generative fill revolutionizes photo editing. It requires some photoshop knowledge, but really it's very simple to use. You don't really have to know that much about photoshop in order to use this really great tool. All you need to do is go into Photoshop, upload your image. Come up here to the Marquee tool. Just click on this little rectangular box. Hold and drag to make a selection. Click on Generative fill, type in add a planet, click Enter to send that off as a prompt. Boom. There we go. If you're just willing to download photoshop and give it like 10 minutes to learn this, you can really do some amazing things. That's it for this video, I'll see you in the next one. 27. 4.12 (Photo) Remove Watermarks: Next step, I want to show you something that not that many people know about. Watermark removal. I absolutely love these tools. It's not any one tool. They just started showing up around the time that AI photos became a thing. People build these tools with help from AI models like stable diffusion. Basically what I'm talking about, Here's Google and all we've done is search for corporate people. Let's say you really like this image. You want to use it on your website or something. But there's all these water marks on top of it. Oh, you can't use it because of all these watermarks. Well, if you just go on Google and you search for remove Watermark. Pick the first option, Watermark remover. Online Watermark remover for free. I'm going to upload that image from Google, like that. Wow. I didn't even press a button. It just does it automatically for you. It removes all the little watermarks. This is just amazing. Click on Download. It's free, and just like that, we have the image from Google with the water marks, but we have it without any water marks. I missed one little watermark up here. But now you can just go into Photoshop generate to fill and remove that if you want. I just think this is so funny. It's such a useful tool. And Shutter stock, I'm terribly sorry, but this is going to be the end of stock photos, I think. That's it for this video. I'll see you in the next one. 28. 5.1 ChatGPT Introduction: Welcome to the first chapter on text. Text controls all the other mediums. Video, photo, audio. They are all controlled by text. And when it comes to generative AI and text, as we know, Chat EPT is the master. We're going to quickly look into the alternatives for Chat EPT as well. But if you're like me and you're a heavy AI user and you're really into AI, and all the updates, it's hard to realize that not everybody uses Chat GPT. Actually, The majority of people right now are not using Chat EPT on a daily basis. I do. I use it more than I use Google, and a lot of the people that I know, same for them. But it's important to remember that most people are not early adopters, and Chat PT is still an early platform. You would be surprised to learn that most people have not even tried AI, tried CatiPT, and a lot of people have tried it, but they're not using it. So if you're already using Chat EPT on a daily basis, this video and the next coming few videos in this chapter on text might be a little beginner for you, you know, You might already know some of the stuff that we're going to talk about here. You could skip this video in the next one. But I still think that a lot of people that are heavy Chat IEPT users can benefit from it. You might hear something that you resonate with that you didn't know that will help you that is valuable information for you. In this video, I'm simply going to introduce Chat EPT to the people watching this course who are not heavy users, to everybody who has not fulfilled the full potential that Chat EPT can give you. So let's talk about Chat GPT replacing Google. Obviously, Chat EPT has not replaced Google, and Google, I'm sure Google is going to be around for a long time. But for me in my daily life and my work, Chachi PT has currently cut out and replaced at least 50% of my Google usage. I know this is the same for a lot of people that I know who are entrepreneurs or they have jobs or whatever. Chachi PT has replaced maybe around 50% of their Google usage, which is absolutely crazy. Because of course, Google is like a rock. It's such a big company. It's hard to imagine it going anywhere and not being used. Here's why I think that ChagPT and AI will actually replace Google. That's the first sentence here. Why look for an answer when you can get one right away? If I want something, if I want some information, why would I Google it and look through several articles and look through the Internet when I could just ask CachPT and receive the answer right away? It's almost the same thing as comparing the retrieval of information between looking for information in a book or looking for information on the Internet. So those are the steps that I see. First, we have, you know, looking for information in a book. Then we have looking for information with the Internet with Google. Then we have looking for information with AI. That's the leap that we're now taking. And we have seen this in the past with other really big companies like Google. They are pushed out by innovation, not competition. What I mean by that is it's very, very unlikely that another company will create a platform similar to Google that's going to push out Google and out compete Google. That's almost impossible. But Chachi PT, not being similar to Google at all, being a whole new thing being a thing that is better than Google, a better option. That is something that will push Google away. So with that said, I want to move over to Chachi Pit and show you some real examples of how Chachi Pt is replacing Google for me. So this is sort of a silly example, but I was watching the show, Sons of Anarchy and obviously spoiler alert if you're watching it. But I asked Chachi Pt when one of the characters dies, and it told me in what episode that character dies. Here's me doing some math with Chachi Pt. So I guess I was too lazy to bring out my calculator. And I asked Cheti P, how many percent is five out of 1,500, and it gave me back that's 0.33%. Nice. Here, I had a screenshot of some job titles that I wanted to put into an e mail. So I just uploaded the screenshot of the job titles into Cheti Pat, and I told Chachi Pati write these commas in between nothing else. And that's exactly what it did. Cheti Pati wrote out all these job titles for me and then I could just copy them and paste them into my e mail. Instead of, you know, writing them out one by one, that would have taken me more time. Here's another pretty good example of just super random information that you can get faster than Googling it. So I ask put, what is a fortune teller? And it says a fortune teller predicts future events. And then after that, I ask, what do I need as a EU citizen to work in England, London? And it says as a EU citizen, you need a visa or work permit to work in London since Brexit, and I ask is it hard to get one of those? It says obtaining a visa or work permit in the UK can be challenging, depending on your skills, job offer, or immigration criteria. And then I ask, in 2024, can you stay in England for more than six months during one year as a EU citizen? And it gives me back the information in 2024, EU citizens cannot stay in the UK for over six months in a year without a visa, exceptions might apply. And I say what exceptions? And it says, exceptions include work, study visas, and family visas? I ask it, can you extend the six months? It says you cannot extend the six month Turis visa for longer stays apply for a relevant visa. These are a lot of different questions that I was able to get answers to very, very quickly. And if I had to go into like London's immigration page and, you know, read pages and pages just to find this information. That would have taken me way way longer. So this is just a good example of how you can quickly get some relevant information really fast. And I think it serves as a great example of why you would rather want to use Chat GPT instead of Google. 29. 5.2 (ChatGPT) Paid vs Free & Interface: Okay, so in this video, we're just going to talk about what is the difference between the paid and the free version of Chachi PT. And I'm also just going to tell you how to sign up and use Chachi PT. So, if you're already using it, you can basically go ahead and skip this video. So there are two models of Chachi PT. The first one is the free one, and it's called Chat CPT 3.5. And around March 2023, they released Chat PT four. Which took Chipt to a whole new level. Essentially, Chapt 3.5, the free version. They say, I think you can compare it to the brain of an insect. You can talk to it, but it's not so good at reasoning. It doesn't access the Internet. The paid version, Catipt four, I think they compare it to the brain of a mouse or something like that. That one is good at reasoning, and it can search the Internet, which is obviously a huge deal breaker. Where it says context window, four K and 32 K, that's basically 4,000 tokens or 32,000 tokens that the free or paid version of Catipt can handle. Now, what is a token? A token can be compared to a word or a piece of text or a part of a word. So basically, you can give the free version of Cha tit upwards of about 4,000 words, and you can give the paid version upwards of about 32,000 words. Obviously, this is going to change. They're constantly trying to upgrade their product and make it as good as possible. They do have another version of the paid Chat four called the Chat PT four Turbo that can handle 128,000 tokens. But that one is only available in the API axis at the moment, so we're going to talk about that later. For now, just think about it like this. The free version. You know, I can be fun to try it just to see what Chat GPT is if you've never tried it. But if you're going to use Chat PT for your work, for your personal life, if you're going to get any value out of Chat PT, you need to have the paid version. It's $20 a month, and if you're going to do this course, you're going to have to buy it. I don't make any extra money off of you buying Chat GPT four. It's just that the difference in how good it is really is that big. So if you don't have a Chat CPT or Open AI account, you can go ahead and just search for Chat PT. You can either click on Chachi PT, or you can go to the Open AI website. Open AI is the company that made Chat EPT. Click up here, try Chat PT, sign up, create an account, log in. Okay, so this is what it looks like once you're logged in. Now I'm just going to quickly show you the interface and all the buttons and what they do. And we're also going to test the free version and the paid version, so you can see the difference between them. This menu here on the left, that's all your previous chats. You're not going to have any chats here if you're just logging in for the first time. These two things up here that are called e mail marketing and SEO blog writer. Those are custom GPTs. We're going to talk about that later, but for now, you can just ignore them and also, you're not going to have them if you just logged in. This one right here, which just says Chachi BT. If you click it, you're going to open up on U chat. If I go to one of my old chats and then I click up here, that's going to open up a new chat for me. Down here on the left, you have your profile, you can click on it, create your plan, customize your Chat GPT, which we're going to do later, can create a GPT or access your GPTs, and you can go to your settings. Except for that, Chat TPT is like the simplest and most valuable thing ever. You just enter something. You just say, Hello, How are you, and it's going to answer whatever question or inquiry or thing that you tell it or ask it. Up here where it says Cat T four. If you click on it, you're going to get some other options. This is where you can choose if you want to be using the free version, CaT 3.5, and that's the one that you're going to be on if you don't have a subscription. Right now, I'm on GPT four, but let's go ahead and switch to GPT 3.5. And I'm just going to tell the free version of Chat GPT 3.5 this. Search for Banana history with B. Can see that it really quickly gave me an answer. And even though the answer is correct, it didn't search the Internet because the free version of Chata PT can't do that. But if we click up here and move over to GPT four and we enter in the same inquiry, send it over, you can see that it doesn't give you an answer right away. This black little thing is going to take a little bit of time, search the Internet, because we told it to search the Internet, and then it's going to give us an answer. And these answers are usually a lot better because they googled the topic that we asked it about. And it got more information on it. The reason that I put Bing in here is because C chat CPT four uses the Search Engine Bing to search the Internet. It does not use Google. This is because Open AI, which created Chat CPT, Open AI is owned by Microsoft, and Microsoft also owns the Search Engine Bing. So of course, they're not going to use Google, they're going to use Bing. This doesn't really matter though, because Bing still does a good job of searching the Internet for you, giving you the answers that you're looking for, and then Chat i PT summarizes them for you. That's it for this video. If you're new to Chachi PT, and you just create your account, go ahead and play around with it, see what you can do. 30. 5.3 (ChatGPT) Alternatives: Let's have a look at some Chat GPT alternatives. So there are other LLMs, large language models that you could be using instead of Chat GPT. Some of them are good, some of them are not so good. The main competitor is certainly Google Gemini, which used to be called Bard, but there are a couple of other good ones like Claude and Lama. So these are all different large language models created by different companies, trained on different data, trained in different ways. You're going to see a lot of AI tools. For example, one of the biggest AI tools is probably Jasper. Jasper is a platform that introduces itself as a chat bot, but it's built off of Chat GPT. So there's a lot of chat bots, a lot of tools that are built on Chat GPT, or they're built on Google Gemini, Claude, Lama, using these LLMs APIs. They're basically taking the Chat GPT engine or the engine of one of these other LLMs. And putting it into their own company, their own chat bot. But these that we're looking at right now, Chachi PT, Google Gemini, Claude, Lama. These are all different large language models. So it's just important to make that distinction. Overall, Chat EPT is still dominant in the space. It has the best reviews, the best customer feedback. But Google recently made an update to their LM Gemini, and it's certainly catching up with ChatCPT. This is sort of like a race, and these platforms, these companies, they are all competing for your $20 a month. They all realize that this is the future and Millions of people are already paying $20 a month to use these, and that number is only going to grow. Right now, Chat GPT has about 16 million daily users, and Google Gemini has about 4 million daily users. There is one main thing to note here, and that's that the free version of Gemini and clawed is a lot better than the free version of Chat GPT. So if you want to use AI and you don't want to pay for it, I would definitely use Gemini or Claude instead of Chat GPT. Claude cannot access the Internet, but Google Gemini can, which of course makes Google Gemini a better option over Claude. All of them are good at reasoning, but Chat GPT wins in most scenarios when it comes down to how much people like the platform. Two positive things about Google Gemini is that they do include 2 terabytes of storage in your plan of $20 a month, and there might be a really good integration coming up with G mail. Obviously, Google owns Gmail. Google has Gemini. It could be pretty powerful once they start combining those two. Let's test out Chachi PT, Gemini, and Claude real quick. So I'm asking them to plan a one week trip in Paris for a single person. List one activity for each day. This is Chachi PT, and it does a pretty good job of breaking it down. It highlights the title of the activity. And then it also describes that activity. Same thing with Gemini, the layout is a little bit different, you know, images, but basically the same thing here, does a good job. And same thing with Claude, giving us a good answer and good breakdown of the activities. So they all do a good job with, you know, basic reasoning. As I said, if you want to go for a free version, Gemini is a really good choice because it can access the Internet, and it's pretty good at reasoning even on the free version. However, for most of us that want the most value out of AI. Currently, Chachi PT is the no brainer choice. It's just overall better. It could change in the future. But as of now, if you want my recommendation, just go with Chachi PT. 31. 5.4 (ChatGPT) Custom Instructions: Okay, let's talk about customizing your Chat GPT account, customizing your chat, the way that it speaks to you, the way that it formulates its sentences and writes to you the length of the text that it gives you. This is pretty much the only setting in Ja GPT, and it's also my favorite setting. It's great. I love it. It's really good. And it serves as a really good example of how settings are going to work in the future. Before, you've always opened up a settings page and you have all these parameters, you have buttons and stuff like that. But now you're going to start seeing setting pages that look more like this. This is the setting page that we're looking at right now, and this is all that it is. You can describe to a chat what you wanted to do, and it will do it for you. So the way to find the setting is you just come down here to the left corner, click on your name and click on Customize Chat GPT. And you're going to see two boxes here. First, you're going to see what you would like Chat GPT to know about you to provide better responses. And then you're going to see how would you like Chat GPT to respond. And this is the custom instructions box. These are the only two settings that they really are. And then you just want to make sure that this little thing down here is, enable for new chats. So I'm going to read you what I've put into these boxes. And this is what I use on a daily basis. This is not prepared for this course. This is what I actually use. What I find to make Chat GPT give me the best responses in the way that I want them. So I just say I am an entrepreneur based in Sweden. I'm interested in saving time by not wasting time on reading or writing unnecessary text. I like to get the facts without introductions or outgrows. Over here to the right, you can see what Capti is recommending you to put into this box. It's recommending you to put where you're based, what you do for work, what are your hobbies interests? But basically in this box, you can just put a very brief description of yourself. The second box is really the more important one. Here we have how formal or casual should Chait be, how long or short should responses generally be? How do you want to be addressed? Should Chatbt have opinions on topics or remain neutral? This second box is really going to affect what the answers that you get from Catipt look like. As you can see right here, you can put in up to 1,500 characters in each box. So I'm just going to read you what I've put in my second box. Never write sentences longer than 25 words. Don't write introduction sentences in your answers. Don't write outrows. Only give the direct answer to the question or statements. Besides of the direct short answer needed, don't say anything else, don't write another word. Be even more brief in your answers than you are currently. Short, concise answers only. Convey the value of the text in as few words as possible. Don't write introductions. Don't beat around the bush, get straight to the point. Don't write endings. Just give me the facts that I'm asking for and try to shut up. And then I'm giving it an actual example here of the same sentence in a shorter and longer format. So down here is the original sentence, and up here is the same sentence, but I've just made it shorter. So I'm giving Chatut an example of what it looks like making a sentence shorter. So as you can see in this second box, I'm really repeating myself here. I'm really repeating the message of, you know, give me short, concise answers. Sometimes it doesn't always listen to you. So the way that I've forced Chat GPT to listen to my message here to make short answers is to really repeat that message like a lot. So once you've saved your custom instructions, they are going to apply to your regular chat GPT chat. Now, if you have other GPTs, like I do, right here, the custom instructions are not going to apply to those GPTs. They will only apply to the default chat up here. All right, so let's have a look at some examples. So I told Chachi PT, tell me about Spain. And this is me talking to the regular Chachi PT chat that has my custom instructions included. So, as you can see here, it just answered in two very short sentences. This is how I like my answers, short and concise. Saves me a lot of time. But then I asked one of my GPTs, which is the Chat PT chat without any custom instructions, the same thing. Tell me about Spain. And it feels like it gave me a whole blog article on Spain. Now, if I want to write a blog article, this could be good, but if I just want to know briefly about Spain, I don't want to read all this stuff, I would rather just read one sentence that encapsulates all this information. I just want to show you that even with my custom instructions, instructing JagibT to write short answers, if I tell it to write a 500 word blog post about dogs, that's exactly what it's going to do. So if you're telling it to do a longer post, it's still going to do that. But if you're just looking for a shorter answer, that's what it's going to give you because that's what I instructed my Chachi Pit to do. There's no limitation here to what you can tell Chachi Pit to do and how it should respond to you. Would definitely recommend playing around with this, and I would also definitely recommend telling your custom instructions to give you shorter answers because Chachi is known for giving long answers a lot of the time. Longer answers than you would normally need. 32. 5.5 (ChatGPT) Usage Cap: Really quickly, I just want to tell you about the Chat GPT usage cap. Basically, on the regular plus plan, which is $20 a month, you get about 40 messages every 3 hours. But it varies on how much compute power they have available. So you get 40 messages every 3 hours, but a lot of the times you'll get a lot more than that, and you don't even reach the usage cap, even if you send maybe like 60 or 100 messages in 3 hours. So if you've reached your limit, this is what it's going to look like if you try to send something to habit. You've reached the current usage cap for GPT four. You can continue again after a specific time that they are going to tell you right here. Or you can use the free model right here as much as you want. What solutions do we have to this problem? We can either wait, upgrade the plan, or use a separate tool. The waiting part is pretty obvious. Just go to habit and check the time that I told you to come back. Or if we have a look at their pricing plans here, the plus one for $20 a month. That's what most of you are going to be on. This is the plan that I'm on. The team plan is another $5 a month. But you need two users for this one. If we go back to Chachi Pi, click on your profile, click on my plan. You can see right here under team. They need a minimum amount of two users. But the team plan gives you higher message caps on GPT four. So they don't tell you how many that you get. They just tell you that you get more. And then they have an enterprise plan. Unlimited access to Chat GT four. But of course, they're not going to show you a price for that. You have to contact their sales first. 33. 5.6 (ChatGPT) Internet Search: There's a few things I want to quickly cover when it comes to Chachi Pits Internet access. Sometimes it will tell you it can't access links, but it does it anyways, and this can depend on how you prompt it. When Chachi PT searches the Internet for you, it will read about three to ten sites before giving you an answer. This is important to know because sometimes this is not enough. Sometimes this means that it doesn't find the information that you want by only looking at three to ten sites. But if you prompt it again, it will keep searching and eventually find the right answers. I just want you to know that Chachi Pits Internet access has been revoked in the past. Open AI has taken away Chachi Pits ability to search the Internet. And who knows? This might happen again, but it's been up for several months now. And it's obviously one of their key features, so I don't think their users would be very happy if they did it again. But you never know. Chachi PT uses Bing, not Google as their search engine. This is because Microsoft owns Open AI. Open AI created Chachi PT, and Microsoft also owns Bing. So, of course, they're going to link those two together. I just want to show you real quick that sometimes if you prompt Chachi PT something that sounds like this, visit this link and read the content, and then you put a link. Sometimes you'll get an answer saying, I'm unable to open specific links. But now I just want to show you that here's a random article. The article is about a page called copy.ai, and the headline is 0-10 million users in four years. So if we copy this link, and then if we open Chachi PT and we paste in this link, and then we just ask ChaiPT, what are they talking about here? Send that off as a prompt. It will actually access this link, and it's going to tell you the article discusses copy AI's growth to 10 million users. This is just a good example. Proving that hatibi can in fact access specific links. But if you're having problems with it, what you want to avoid is to literally type out, visit this link because that seems to trigger some kind of glitch or problem. And my other point, if we ask Catibt, how many web pages can hatibt access before giving you an answer, I will tell you hatibti can open up to ten web pages in a single Q. This is my point if you're looking for specific information and you don't get it the first time that you ask for it and Chachi PT searches the Internet for that information. Then you might just want to tell Chait, keep looking until you find more information on what I asked for. You can also tell Chachi PT, be creative in your searches. It doesn't just search for the same thing over and over again, but it searches for a wide variety of things giving you a wide variety of answers. That's it for this video, S in the next one. 34. 5.7 AI content detection: All right, Let's address this question. Does AI generated content hurt SEO? If you're not familiar with SEO, basically, that just means search engine optimization. That basically means what is being showed on Google. What we're asking here is if we write content, a blog, or whatever, with AI, and Google recognizes that that article is written with AI, are they going to not show it to people? The answer to the question is no, it doesn't matter. Google only cares about quality? Same thing goes for Instagram and several other platforms. However, for example, the video platform, TikTok has started marking AI generated video. If you're scrolling through your feed, you might see a stamp on a video where it says AI generated. As AI generated video and photo gets better, I would not be surprised if a lot of platforms start doing this. There's a couple of things I want to show you if you go to Google Search Central blog, you can find this article where they talk about this exact topic. Here they're basically saying that they reward high quality content however it is produced, meaning whether it's produced by a human or by AI, they will show high quality content to people on Google. Google Rewards, expertise, experience, authoritiveness, and trustworthiness. Basically, if you can achieve these things with an AI written article, and good for you. You can make money, and you can get a lot of traffic with AI generated content. You don't need to write the content yourself anymore, but you do have to write high quality prompts. The next thing I want to show you is a website called Content at scale, and they claim to be creating the most human like AI writer out there. Up here, you can click on AI detector, and you'll be taken to this page right here, which is an AI content detector. Here I'm going to paste in a story of 100 words that I had Chait write for me, and we're going to click on Check for AI content. Over here to the right, it's going to tell you that this article reads like AI. The sentences here that have been marked as orange, they are AI tagged, I guess. If we go back to Chait and we ask it explain how AI content detection works, it's going to tell us AI content detection uses algorithms to analyze text patterns and style. Then if we ask Itawa to know for sure if content is AI generated or not, it's going to tell us no guaranteed method exists to determine if content is AI generated or human written. So websites like this one content at scale.ai, which are AI content detectors. They're not a scam. They can actually detect AI content, but they can't prove it. So if you're a student at a university and you turn in an article and your teacher says he used this tool to basically prove that you made that article with AI, he's wrong. He can't prove it. But obviously, if your entire article is flagged as AI, it's pretty easy to assume that you made it with AI. This tool, content at scale.ai is free. So if you're interested in whether your content flags as AI or not, you can just come over here and put it into this box and you'll know. That's it for this video, I will see you in the next one. 35. 5.8 (ChatGPT) Language: Talk about hatipt and language. Chatupt understands basically all major languages, but it might be slightly more skilled at bigger languages. My favorite thing about this is that you can write the hatipt in several languages at once. This is something that I do frequently because I'm Swedish and I'm going to show it to you right now. This is something that I naturally do frequently. Here I have a sentence where I switch between English and Swedish three times. This first part is in Swedish, then it's English, then it's Swedish again, English, and then Swedish. What I'm saying here is write a story about how a dog ran away from home and then found its way back and made a front along the way and write it in English. You can just send that off and Chachi PT understands that perfectly. Let's do this. Let's have Cachit write a 500 words story about a dog in Spanish. Here's our dog story in Spanish, 500 words long. Let's copy that one. Move over to a new chat, paste it in here. Move to the top and now I'm going to write in Swedish. I'm going to write summarize this story in three points, one sentence per point and write it in English. Send that off to Chachi. Bam, it's going to summarize this story in three points and it's doing it in English. I think this is really cool because now we're using three different languages, giving the instructions in Swedish, giving the story in Spanish, getting an answer in English. This is something that you really have not been able to do before. If you speak two languages like me, it can be really helpful to be able to communicate in both languages at the same time. I think it's amazing. If we ask Cheti, what languages does Chaipt support, it's going to list some languages. Here, Chachi is saying that it does support Swedish and I asked it as fluently as English, and it says, no, not as fluently as English. English support is more advanced. Then I say what languages are Catbi best at, rank it in order. English, French, Spanish, German, Chinese, Japanese, Russian, Italian, Dutch, Korean, Portuguese, Arabic, Turkish, Swedish. Now my language is number 14, and my country only has 10 million people. Basically, you can expect that whatever language you speak or want to translate to, Chatb can already do a pretty good job at it. That's it for this video, S in the next one. 36. 5.9 (ChatGPT) Website to PDF: Now I just quickly want to show you a little trick that you can use that can be very helpful in a lot of different scenarios. All we're going to do is take an article online or a website, turn it into a PDF and then give that PDF to ChatBT. In the next video where we look at creating a custom GPT, this can also be useful. Just for example, here's a random article that I found about global climate highlights in 2023. If we scroll down here, we can see that this article clearly has a lot of information, a lot of statistics on climate change in 2023. Instead of copying the link and then moving over to CaT pasting it in here, Then asking CGPT, what's this about? Maybe you're having some troubles with CGPT accessing the Internet. Maybe you'll just get a better result from uploading the entire website as a PDF. What you can do is copy the link and then go to Google and just search for website to PDF. You're going to get all these results for different online tools that are free to use. Can you just click on the first one, web page to PDF. This website is called webdf.com. Then we just paste the link of the article on climate change in here. Click on Convert. That's going to turn this article into a PDF that we can then upload into CGBT. All right. It's completed. Click on Download file now. Now we can go back to Chai PT and just drag in this PDF. Then we can ask Chad PT, what is this about? Of course, Chad PT is going to read that PDF file and give us all the most important information from that PDF. The other scenario when this could be useful is if you want to let's say compare information from five different links. Here we have five different links, one, two, three, four, five. Instead of pasting all of those into the CaT GBT chat, and then saying compare the opinions from these links, which might be a little confusing to aGT because it might have some trouble accessing five links at a time. Also it's going to visit one link at a time and then give you some information and then visit another one. But if you just upload a bunch of PDFs, it will be able to access all the PDFs at the same time. So let's say you have six different PDFs like this. Then you could just pull all of those PDFs into Cha GPT. Then, for example, you could say, compare the opinions of PDF one with the opinions of PDF two through six. And this is going to be a lot clearer than if it was websites than if it had six different websites to visit. All right. That's it for this video, see in the next one. 37. 5.10 (ChatGPT) Custom GPT: Okay. Let's talk about custom GPTs. In this video, I'm going to explain to you what a custom GPT is, I'm going to show you how to make a GPT. A custom GPT is like a chat with separate settings. You could also say that a custom GPT is an AI trained on your own data. Now, this video might be a little longer than the other videos in this course. If you're not interested, you can just skip this video, but it's still good knowledge for everybody to know. A GPT is what we can see up here. Here you can see that it says Chat GPT, and then it says Default GPT and then it says Twitter Post writer. Now the first one here, chat GPT. That's the normal Cat GPT chat. If you click on that one, it's going to take us to the normal chats, where we have the normal settings. But these ones, default GPT, Twitter Post rider, these are custom GPTs that I have made, and I've named them default GPT and Twitter Post writer. If you've never made a GPT, you're not going to have any of these. It's going to be empty up here, or you could have a lot of them. You could have like ten of them. When we click on them, we just enter a chat for that GPT specifically. Now, if we go to the regular Chat GPT chat here, You remember the episode where we clicked down here and we went to customize at GPT, and here are the settings for the regular Chat GPT. These settings do not apply to your GPT, and that's because your custom GPTs, they have their own settings. That's the whole point of a GPT. A custom GPT is like a chat with separate settings. Let's go through the GPTs that I have here. If I just want to chat to Chat GPT without custom instructions, I have my default GPT, which is basically just an Mt GPT without any settings or any instructions. Then I have a Twitter post writer. This one has been programmed to write Twitter posts specifically. If we click here on Explore GPTs, we will be directed to a store for other GPTs that other people have made. Here, for example, under programming, we have Code copilot. We can click on that one and click on Start Chat, and that's going to take us to this GPT, which is specifically trained to code. Let's have a look at how to make a GPT. To make one, you're just going to come down here to your profile, click on it, and then click on my GPTs. Here you're going to see a little list of the GPTs that you have if you have any. Then you can click on Create a GPT. Now, you're going to have some different windows here. The first one that you're going to land on is Configure. This is where you add your instructions, give it a name. Then if you click on Create here, you're going to be taken to a chat, this chat is actually where you create and build and instruct your GPT. Like I said before with the custom instructions, this is a new way to have a settings page. Instead of having a bunch of buttons and settings like this, all we have is a chat and we tell it how we want it to act. Let's make a GPT that is a long format blog writer. Just for example, we're going to tell that you are an expert at writing blogs with more than 1,000 words on the topic of climate change. The next thing I'm going to do is I'm going to come right here and I'm going to click on Upload files. Now, you remember in the last video, we transformed some web pages into PDF files. Well, I have here three different PDF files on climate change. I'm just going to highlight those and drag them in here to the upload section. Upload these three PDF files. Then I'm going to come back to the instructions. I'm going to give it some additional instructions here. All the blog posts you write are 1,000 to 1,500 words. They all have five different sections with different headlines. Put H one tags on the titles. Put age two tags on the headlines. This is just going to format the size and boldness of those headlines and titles. Right with comedy and passion, even if the topic is serious and informative. Now we're going to go up to the right corner and click on Create. Here you can choose if you want only you to be able to use this custom GPT. Or if you want anyone with a link to be able to use it, or if you want to publish it to the GPT store so that anybody can use it. It's published. Click on Ve GPT. That's going to take us back to the original C A GT interface. But now up here in the left upper corner, you can see that we have long format blog writer as an option. I'm just going to tell it to write a blog on climate change statistics in 2023. It's written out this blog. First of all, I just want to say that these things are not perfect. I checked the word count on this blog and it's not quite 1,000 words. But if we want to fix that, we can just tell it, make it 1,200 words, and it's going to redo the article. It's doing the other things correctly though. It put an age one tag on this title. This is the size of an age one tag in a blog, and it also put age two tags on the headlines. I told it to be comedic even if the content is serious. Let's have a look at the first section here. Welcome to 2023, a year that's sizzled in the frying pan of climate change. Let's remember one thing. Climate change isn't just a passing trend. It's like that one guest at the party who just won't leave. It's definitely trying to be funny. I'm not saying it's successful and being funny, but it's certainly trying to be funny. Or at least writing in a comedic style. I think you get the idea, you can instruct your GPTs in specific ways and then you can access them really quickly. It's like having custom instructions on your regular chat GPT chat, but you just have a lot of them. You can have different GPTs for different purposes, of course. If we go back to my GPTs, go into the long format blog writer again, and then you click on this little pen here, edit GPT. That's going to take us back to this section where we can create our custom GPT. Right now we're on Configure. What you can do here is you can click on this plus and that's going to add an image to your custom GPT. You can upload a photo or use Dolly to create a photo. I clicked on U Di but if we move over here to the Create section, you're going to see that the GPT builder is currently generating a photo for this custom GPT. It made us a profile picture randomly, I didn't even give it any instructions. It just made it for me. Of course, you can give it specific instructions on how you want your image to look. Now as you can see up here, we have a little profile photo for this GPT. Now in this Create section is where we can update our GPT. Here I can tell it all blog articles you write should be at least 100 words. Send that off to the GPT builder, and it's going to start updating our GPT. This is what I mean. You can update your settings just by telling it what settings you want. It's going to answer as your request has been implemented. I'm all set to craft numerous detailed blog posts on climate change, ensuring each one is at least 100 words long. If you go back to configure now when we look in this box, you're going to see that it actually changed the instructions. You can either change these instructions yourself in the instructions box, or you can change the instructions in the Create tab with the GPT builder. This is where you got to watch out because if you start off in the configure tab, giving it an instruction here. Then like we just did, you come to the Create page and you give it another instruction. That instruction is going to override the old one. You can't just enter an unlimited amount of text in this instructions box. This is about the length and the size of instructions that you can have in here. I would say that there are two main custom GPT use cases that can be really helpful. One, you can create a custom GPT for your company. You can create PDF files from your website, and then you can upload those PDF files into your custom GPT so that your GPT is specifically trained on your company's data. You can give it a writing style that aligns with the writing of your business. The second one, of course, is just different writing styles. You could have one GPT that writes Twitter posts, you could have one GPT that writes blog posts, you could have one GPT that writes e mails. You get the idea. It can be very helpful to save time. When you're finished working on the instructions for your GPT, don't forget to come up here to the right corner and click on the green button that says Update. Boom, just like that, our new custom GPT is ready to be used. Hope this was valuable to you. I'll see you in the next video. 38. 6.1 Introduction to Automation: Okay. Let's talk about AI automations. For most people, this will open up a new way of thinking. Once you realize how much content is already automated, it will make you think twice when reading a blog article. Automations existed before AI. Automations and AI, they're not the same thing. But combining AI with automations is very powerful. You do not have to be a developer to create automations. I am not a technical person by nature, and I was still able to learn this. You just have to be willing to spend a little time learning. Why automate? Work a little more to create the automation. Don't work at all to complete tasks. Automating computer work is like building a machine for manual labor. You could even say that the time we are currently living in can be compared to the industrial revolution but for computer work. Few hundred years ago, people did the tasks with their hands. Now a lot of manual tasks are automated by machines. Since computers came out, the tasks for computer work have mainly been done by people's hands. People do a lot of computer tasks manually. But as time moves on, more and more computer work is being automated. Who are these videos on automation for? These videos we're going to cover the very basics of no code automations. If you're a developer or you have done automations before, this might be too beginner friendly for you. Might already know some of this stuff and you might know a lot more than me about automations. But if you're not familiar with automations and you're not a developer, this should be very exciting to you. It presents big opportunities to save time and create workflows we could not create before. No code tools. In these videos, we're not going to be talking about things that require you to code. We're going to be setting things up in no code platforms, and those platforms are make.com and Sapier. These two are the leading platforms in no code automations. They have nothing to do with AI. They existed before Chat EPT. But now that you can integrate AI and Chat PT into these platforms, you can accomplish some really cool things. First, all software applications and features were built with code by people who knew how to code. Everything in the software world still consists of code. But now there are user friendly programs that will handle the code for you. Your experience using them is like using any other platform. Like mid journey or Chat EPT. It's just another platform, and this is done within make.com or Saper. Let's get into it. I'll see you in the next video. 39. 6.2 (Automation) Zapier&Make Introduction: Let's talk about make.com and Saper. In this video, I'll just be talking about how it works and what you can do within M and Speer. Then in the next video, we're going to look at some actual real examples within M and Saper. First, there's a trigger, then there's an action, and then you can have a chain of actions. Something happens. Once that thing happen, a lot of other things happen as well. For example, you receive an e mail. It reads that e mail, then it puts the content of that e mail into a spreadsheet. I just want to say that this is not a course on no code automation. This is just an introduction to it. While no code platforms are simple, there are thousands of things you can do within them. You could learn how to set up a simple automation in a couple of hours, but you could also spend months building features in Ma or Sapier. The videos in this course on automation are supposed to open your eyes to what's possible. Let's have a look at a few more automation examples. Let's say you want to keep a list of every single e mail address that sends you an e mail. You connect your e mail address to make.com. Then you connect a Google spreadsheet to make. Then you create settings that enter the recipient's e mail address into that spreadsheet each time you receive an e mail. Here's another slightly longer example. You receive an e mail, an app within SAPR, reads that e mail. You give the text from that e mail that you got to Chat GPT within SAPR. Chat GPT reads that e mail and labels that e mail as a customer support question. This is where we can start incorporating AI. You connect a Google spreadsheet to SAPR. Sapier puts the text from your e mail into that spreadsheet. And then everything from that spreadsheet is sent off to your businesses customer support team. The next e mail that you get is labeled as a sales e mail by Chat iBT, and that e mail is sent off to the sales team. Here is another example. Let's say you have an E commerce store, you sell things online. You connect your stripe payment account to SAP. Then you connect your e mail to SAPR. Every time a sale is made, an e mail with pre written text is sent to the customer, saying, thank you for your order. Here's another example. Let's say you run a small business and receive a lot of small receipts from different purchases you make. You connect a folder within Google Drive to make. Every time you upload a file into that Google Drive folder, that file is sent to your bookkeeper. Now you can send photos of receipts and PDF invoices to your bookkeeper by just dragging them into a folder. Saves you a little bit of time. I'll see you in the next video. 40. 6.3 (Automation) Zapier&Make Examples: Okay, let's have a look at some automations within Sapier and make. This is saper.com. This is an automation that creates Instagram posts automatically for us. To start off in this automation, the first task that we have up here is connected to Google Spreadsheets. We have a spreadsheet connected to this automation, and this spreadsheet is constantly automatically updated with text content. This text content can be on different topics. The next task that we have is Dolly. As you might remember from the image section, Dolly is open AIs image generator. This task takes the text from the Google Spreadsheets and creates an image based on that text. The third task that we have in this automation is Chat EPT. This task also takes the text from the spreadsheets and rewrites it in a new format of one or two sentences. Whatever the topic is of the text in the spreadsheets, Chat EPT takes that topic and rewrites it into one or two sentences that is suitable to post on social media. Then we have another task here with another application called Publisher Kit. This fourth task takes the text output from Chachi PT and layers it on top of the image that was generated by Dolly. Then it creates a new image with the text on top of the original image from Dolly. Let's say Dolly generates an image of the ocean. Chat PT might write a sentence talking about the ocean. Then this fourth task with the publisher kit puts the text of the ocean on top of the image of the ocean. Now we have a finished image ready to be posted on Instagram. The last task of this automation is, of course, Instagram. So here I've hooked up an Instagram account to Sapier, and this last task simply just posts the image on Instagram. Here you can also type in some captions and some other information. Here's another example of another automation. Now what this automation does is that it takes a link, visits that link, summarizes that link with C hat GPT, and then places the text summary within a Google spreadsheet. The first task of this automation, the trigger for this automation is whenever a new row is added to a Google spreadsheet that I have, this spreadsheet only contains links. A new link is added to my Google spreadsheet. Then the second task here takes that link and visits it and reads all the information on that link. Now we have extracted all the texts from a certain web page that we have visited. The third task in this automation is Chat EPT. Now we're going to give all that text from the link to Chat PT to then summarize within Chat EPT and rewrite it, reformat it as one or two sentences suitable for Twitter. Once Chat HPT has done that for us, we take the text output from Chat EPT, the one or two sentences of text that is ready to be published on Twitter, and the fourth and last task in this automation adds that text into another Google spreadsheet. That's the end of this automation and then another automation can take over. Take the text from the last spreadsheet and turn it into an Instagram post or a Twitter post. Here's another automation. This one is a little longer, a little more complicated, but I'm going to go through it quickly. What this automation does is that it reads an e mail newsletter that I get every day, and then it puts all those links into the spreadsheet that we were just talking about in the other automation. Every day I get an e mail newsletter. This newsletter always contains a lot of links. These links lead to interesting articles on different topics. The first task in this automation reads the e mail that we get every day from the newsletter. The second task in this automation is Chat GPT. Here we're giving the entire e mail to Chat EPT. Chat GPT then takes that e mail and extracts all the links from that e mail. The third task takes the links from Chat EPT and reformats them as regular text. Then down here, we have another Chat EPT task. This one takes the first link and puts it into a Google spreadsheet. Here we have a task that simply makes the automation wait 2 hours before moving on to the next task. Then the automation repeats itself and it goes on and on and on, and does the same thing. Basically every 2 hours, it takes a new link from the e mail newsletter and puts that link into a Google spreadsheet. That's the end of this automation. It feeds a link to a spreadsheet every 2 hours. Then once that's done, another automation can take over and read that link and then turn it into a Twitter post or a blog post or whatever you want. Let's have a look at M. This is make.com. It's the other platform. It looks a little bit different, but it's the same concept. It works in a very similar way. Over here on the left is our first task. This is our trigger. And here on the right is our first and only action. What this automation does is that whenever a new row of text is added into this Google spreadsheet, it will take that text and post it on Twitter. As I showed you in SAPR, we have automations that take links and read them and rewrites them into Twitter posts using chat GPT. Then this automation takes that text and posts it on Twitter. Now, SAPR does not have Twitter and that's why I'm doing this in ma.com. Here's another example of a simple automation within ma.com. This is an automation that I was actually talking about earlier. What this automation does is that every time I receive an e mail, it will save that recipient's e mail within a Google spreadsheet. This way, you can keep track of everybody that sends you an e mail. It's very simple to do. There's only two tasks. First, you hook up your e mail account and then you hook up your Google account. Connect your e mail address, connect a specific Google spreadsheet, and then adjust a setting so that every time you receive an e mail, either the recipient or the contents of that e mail are placed into a Google spreadsheet. Before we end this video, make versus spar.com, which one should you choose? Personally, I think Saper is a little bit easier to use. Make is cheaper and has more features. Both of them are really good. I think whatever you want to do, you can accomplish it in either platform. However, some specific apps might only be available in one of the platforms. But almost all the major platforms like Chachi PT will be available in both Ma and Sapier. You're just starting out, Spier might be a little bit easier to understand at first. But if you have a lot of repetitive tasks, M might be a better option to save money. That's it for this video. I'll see you in the next one. 41. 6.4 (Automation) API: Quickly talk about API application programming interface. Because to set up your own automations, this is good to understand. Of course, if you know anything about development, you know what an API is. This is for the people that don't know what API is. If that's you, if you don't know and you're interested in automation, it's good to understand what an API is. But even if you're not going to make your own automations, it's also good for a general understanding of what platforms can be built upon or not built upon. API is like a connection between programs that enables them to talk to each other. When a company builds a program, first, they offer it to consumers. That's usually how it goes. Then they might or might not offer it to companies or to people who want to build their own thing with their program. That's when you need an API to access their program. So if a company like Open AI has a program like Chachi PT, they can offer API access to that program. With Open AI API for Chachi PT, you can then use Chachi PT within Sapi M, you can use it on your website or within an app that you're building. But if Open AI did not offer API access for ChatPT, then you could not use CachiPT within Sapiar M on your website or within the app that you're building. So if you see that a platform that you use has API access, you know that other people can build their own things using that platform. Why do companies offer API access to make money and grow, of course. API access can cost money. That's up to the company offering the API for their application. CAT GPT, API access costs money and you pay per usage. As a CAT GPT consumer, you pay $20 a month and you get a certain amount of usage. On the contrary, if you provide CTPT access to your own consumers through OpenAI's API, then you will have to pay Open AI for every single task that CAT GBT performs. If you build an app that uses CTBT's API, every time your consumer sends a QRE using CTPT within your app, you pay Open AI. So let's quickly go to Open AI API page. Just go to open a.com and log in, and you're going to land on this page where it will either take you to Chachi BT or to their API page. Now, this is their API page and there's a lot of other pages on this page, making it perhaps a little bit complicated if you're new. But basically here on the left, if you open up this menu, you can click on API keys. Here you can create a new key and then it's going to show up like a little snippet of code. That you can then copy and paste into SAPR or your website or wherever you want to use it in order to connect your Chat GPT account to your SAPR account or your make account or your website. Over here on the left in the menu, if we click on usage, here we can see our Cat GPT API usage. If I hover my mouse over some of these columns, it will tell me the amounts of money that I've spent on GPT four GPT 3.5 image models, which would be Dolly. Here, for example, on this day, I spent slightly more than $1 on Chat BT's API. Of course, if you build an application on this and you get a lot of customers, ChatBT is going to make a lot of money off of you, because you're not paying them $20 a month for the API axis, you're paying them every single time. Somebody uses the chat in ChatBT. It's not very expensive, but now you might understand if you see other programs online where they've built their own applications using Chat IBT, why they have to charge money for it. Because it's actually costing them money to provide you that service. Now the free version of Cat EBT, GPT 3.5, API access for that one is very cheap comparatively to the one for GPT four, which is the better version. Just as another example, if we head over to music five dot L OL, one of the platforms that we were going through in the section on audio. Appear in the menu, you can see that they have an API page. As soon as we see this API axis, we know that we can use their program to build our own program. If you want to build your own app that makes music, you don't have to develop your own system, you can just build the app and use their API access to use Music Pi system. That's it for this video. I'll see you in the next one. 42. 6.5 (Automation) Where to get information: Let's talk about where to get information for automated content. Before looking at how to make an automation for automated blogs or social media post, we have to understand where the information for those posts can come from. Because the Chat GPT API for GPT four does not include the feature to search the Internet. When you're using CaTPT in Sapi or M through OpenE API, you can't tell CAPT to search the Internet. CTEPTS API will base its answer in your automation on the information that CATIPT already have. Lucky for us, it has a gigantic amount of information. When you're using Chachi PT within Chachi PT, as a consumer, You can tell CTEPT to search for the latest information on electric cars, and it will go on the Internet and search for you. But when you're using ChatPT right here within Sapier in an automation, through OpEIes API for ChatPT, you cannot prompt Chat IPT here to search the Internet. Our process for automating content is to find information on a topic, convert that information to text, give the text the ChatPT, repurpose the text into your own brands voice, into your own format with CatiPT, and then make a blog article or social media post from that text in your own format, with your own voice. So what does this mean for us? It means we can't make an automation where ChatBT goes on the Internet and searches for the latest information every day. But we can make an automation where Chat PT creates new content about dogs every day because CatiPT already knows all the basic information about dogs. If you want to automate content which is about very specific information or content that is about the latest information, you will have to get that information somewhere else, and then give it to Chachi PT within your automation to reformat it. Our options for how to get information are. We can get information from, links that we turn into text and give to Chachi PT, or we can get information from PDF files that we give directly to ChatBT. Or we can use knowledge that Chachi PT already has. How to think if you want to use information that Chachi PT already has? Chachi PT knows most basic knowledge that has existed on the Internet for a while. Automating content about phishing, dogs, hiking, travel tips, exercise history, or another broad topic will work. Because Chachi PT already knows most information about fishing, dogs, hiking, travel tips, exercise history, or another broad topic. But automating content about the latest electrical cars will not work so well because it relies on having very specific information about the latest car models. If you're using Chachi PT as a consumer, it could go and search for information about the latest car models, but it will not know the latest information about the latest cars. GPT four does have information about companies. For example, it might be able to tell you specific information like revenue for a big company like Apple. It might even have some information about smaller companies, but there will be a lot of information in these areas that it does not have. Then we have to use other methods for retrieving information. If you want to feed your automation, specific information, then you must find other sources for that information. These sources can be newsletters, blogs, RSS feeds, which is basically like whenever a website is updated. It can be websites, podcasts, or any other source that you're creative enough to figure out on your own, converting your information into text. If your information comes from links, then you have to use an app within Sapier or make to read that link and convert everything on the links sites into text, text that you then give to CachPT in your automation. You can also convert it to a PDF that you can directly give to CachPT within your automation. Let's say your information source is a podcast. Then you can use a platform or an app to transcribe each podcast episode into text and then give that text to Chachi PT in your automation. The point here is that we can only give Chachi PT PDF files or text within our automation. I just want to repeat this. Our process for automating content, if we are not using information that Chat EPT already has is to find information sources on a topic, convert that information into text, give the text to Chat EPT, and then repurpose that text into your own brands voice or your own format, and make blog articles for social media posts from that text. I will see you in the next video 43. 6.6 (Automation) Words to avoid: Are some more tips when writing content with AI. This is specifically extra important if you're going to automate content because if you don't do this, your content is not going to be very good. But of course, this applies if you're doing it normally with Chachi PT as well, it's not just for automating content. There are words and phrases to avoid when writing content with AI. These are phrases that AI commonly use, and they sound a bit robotic. Prompt Chachi Pet to not use these words when creating content in your automation. Here's a list that I'm going to show you a bunch of words and phrases that you should avoid. Here's an example of these words and phrases being used in a sentence that is written by AI. To unlock the potential of groundbreaking technologies, we must embark on a journey to delve deep into research and innovation. In the realm of artificial intelligence, it is crucial to foster an environment where knowledge can freely flow empowering creators and innovators to discover and master new methodologies. It just sounds robotic. It doesn't sound human, it doesn't sound good. It's not a style that I want to read. The best way to get rid of these words, is to just include this in your prompt. Never use any of the following words or phrases in any text you write. And then underneath here, you can add in all of these words. And these words and phrases will be available for you to copy and paste into your prompt in the text for the video that you're currently watching. So let's go through the words and phrases. Delve, unleash, crucial, dive, discover, unlock, sure, master, ultimately, imperative, embark, Endeavor, foster, groundbreaking, enlighten, pivotal, and power, dive deep, in the realm of in the world. Breaking barriers, Aims two, continues to push the boundaries, remarkable breakthrough, unleashing the potential, advancements in the realm, elevate, explore the world of, adhere, evolve, bespoke paramount. If you start creating blog posts or any kind of contents with AI, you will soon start noticing a lot of these words in your text that you're getting from Chachi PT or whatever AI platform that you're using. In addition to including this within your prompt, never use any of the following words or phrases in any text you write. You can also give examples of how it should write, like we spoke of in the chapters on Chat GPT. You can include in your prompt an example of something that you've written yourself that you like, and then tell it in the prompt, this is how I want you to write, and this is how I don't want you to write. In your prompt, you have a list here of things to not include in any text that it writes for you. Then as you start using your prompt, you're going to keep finding words and phrases like this that you don't like. Well, then you just copy them and paste them into this list. You might have to work on your prompt for a little bit, but once you've done it and once you have removed all these AI words, it will really produce some great results. Now you might be thinking a lot of these words like master and sure discover. These are normal words. Maybe you want to use them in your articles. Yes, I understand that. But the AI often uses them way too much way too often. In scenarios where the words don't really fit, and that is what makes it sound less human when reading the text. I would highly recommend that you remove words and phrases like this. In order to get a higher quality output in your automations. I'll see you in the next video. 44. 6.7 (Automation) Automatic Blogging: Okay. Now that you have an idea of how to set up an automation within Sapier or make, we can go through a couple of examples of how to do automatic blogging. Automatic blogging, of course, you can do it in a lot of different ways, but here are two examples. One, you can base it off of a list of keywords, or two, you can base it off of the information in a newsletter. Let's start off with automatic blogging with a list of keywords. This is more suitable for broad topics that Chat PT already have enough information about. Here you would filter out keywords that are easy to rank for and also have traffic. Then you can publish 2-10 articles every day without doing anything once your automation is set up. The purpose of this could be to drive traffic to affiliate links, to an E Cmerce store, to ads, whatever you want. Or the purpose of this could be if you work at a company or maybe you run a software business or whatever, you can drive traffic to your own website. To your business by creating articles that have traffic. A great example of this is SAPR themselves. As we've talked about now, SAPR is one of the automation platforms, but they also write a ton of good articles on all kinds of different topics, topics that are not related to their service. If I just go to Google and I type in how Chat EPT works, what is the first result that I get? The first result is a sponsored post by SAPR, and this is an article, just like the ones that we're talking about now. Next up is Semrus, which is another keyword platform that we're going to talk about in a second. Then here's another article by SAPR on how Chat EPT works. Now the reason for these platforms to write articles about other companies or about topics that are not related to their own platform, their own company is that people will click on these articles and read about other topics such as how does Satip work. But as they're reading this article, they are on Sapers websites. Here, SAPR can also pitch their own products, their own platform. This is a very common way for different companies to grow. They publish a bunch of articles on random related topics. Automatic blogging with a list of keywords. You're going to need some keywords. To filter out keywords to use, you can use platforms like SEMrush or AH RFs. These are two of the major platforms on the market for filtering out relevant keywords. If you're brand new, just starting out, I would recommend SEMrush because you can use it a little bit for free every day and you can get a free trial for a week as well. Here I'm quickly going to show you how SEMRH works. This is semrah.com. Over here on the left, if we go to this one, keyword Magic tool, and we enter in phishing as a keyword. Now it's going to show us a bunch of results for different keywords related to phishing within the United States, and here you can change the country if you want to see it in another country. How to choose keywords? You should choose your keywords on two metrics. Keyword difficulty score or KD, and it should have at least a few hundred search hits per month. The keyword difficulty score goes 0-100, and you want keywords that are easy to rank for on Google that have a keyword difficulty score or KD score of 0-25. If we go back to Semruh, currently, we have a search for the keyword fishing. So what do we get? Plenty of fish, fish, Ef fisher? Here, we can see that the keyword, plenty of fish has over half 1 million search its every single month under volume, and it has a keyword difficulty score of 78. So this is a high keyword difficulty score. It's going to be very hard to show up on Google if you have an article that says plenty of fish, and somebody searches for plenty of fish. What you want to do is click here on KD and once you've clicked on KD, it's going to filter these keywords on phishing by the ones that are the easiest to rank for. So here we have a bunch of keywords that are easy to rank for, but as you can see, they don't have that much volume. Now, if you're going to spit out ten articles every month, it's okay. You could probably go with some of these. But otherwise, I would recommend coming over here to volume. And entering in may be a minimum of 400. So now we're getting all the keywords that are easy to rank for on phishing, and they also have at least 400 search hits per month. Now, what you can do if you're on a pro plan, which you can get a free trial for for a week is you can highlight all these keywords and then download a list of them, put them in a spreadsheet and automatically make blog articles for these keywords. And then if the keyword difficulty score is low enough, and you have a good blog, there will be a fairly high chance that people will find your blog articles about these keywords when they search for these keywords, and as we know, several hundred people are searching for them every single month. So that's how you choose keywords. Let's have a look at the process for automatic blogging with a list of keywords. First, you get a list of keywords for your topic. Then you need to build a blog website in WordPress. Then you build the automation to write blog articles for your blog website, and then automatically publish them on WordPress. The automation, for example, every 5 hours, a new keyword could be inserted into a Google spreadsheet. When that keyword is entered into your spreadsheet, Chachi PT writes a title and a blog article. The last step, of course, is to automatically publish the article on your blog. Once you set this up and once it works, every 5 hours, a new article will be published on your website and you don't have to do a single thing. Let's have a look at an example for automatic blogging, using a newsletter for retrieving the information and the process for doing this. First off, you would subscribe to an e mail newsletter in your Niche. Then in your e mail settings, you would forward all newsletters to a receiving address that is connected to your automation within Sapier or M. Then in your automation, you would summarize and rebrand the text from the newsletter with Cha chi PT, or like I showed you earlier, you can make SAPI or make visit all the links in the newsletter and write blog articles based on the information in those links. Then you automatically publish each article on your blog. That's two examples of how to do it. Again, this is not a course on M and Saper. If you're interested in actually doing this, there's a lot of other courses on M and Saper and automations. You can find free information on YouTube. Takes a little bit of learning, but once you learn it, it's pretty simple. Of course, there's a bunch of other ways that you can do automatic blogging. These are just two of them. But hopefully, now you get an idea of how it can be done. I will see you in the next video. 45. 6.8 (Automation) Automating Social Media: Okay, let's talk about automating social media posts. Example, you can have one automation that posts the same texts on Twitter, Linked in, Facebook, and Instagram. Ways to do it. You choose a way to retrieve information, like we just talked about, rebrand it into your own voice format and length. Now Twitter and LinkedIn are easy, since they're mainly based around text. Facebook also works pretty well for posts that are just text without video or photo. And on Instagram, like I showed you earlier, you can layer the text on top of an auto generated image. I know there's a lot of Instagram accounts that already do this, and some of them, they have millions of followers. All you really have to do is create the automations, hook up your social media accounts to the automation, and then focus on writing a good prompt that writes good text output to be posted on these platforms that is able to sufficiently summarize an article or a website or something like that that has a lot of information on your topic and rewrite it in a short format in a couple of sentences. So that it's suitable for Twitter or Linked in or Facebook Instagram. You could have one automation that posts the exact same text on all these platforms. You could also have different prompts for the posts that you write for Twitter or LinkedIn. Maybe you want to write longer posts for LinkedIn. But for Twitter, you want your posts to be just one or two sentences. I just want to show you an example real quick of how it works here. Here's the automation for Instagram posts that I showed you earlier. Right now we're in Sapier. And if we click on the last task here, which publishes this photo to Instagram, here you can see the step details. Here it's listing the Instagram account that we're posting on. Here's the image URL for the photo that we're going to post. This is the photo being posted. Not only can you automate the photo that is being posted on Instagram. You can also automate the process of writing different captions for each post or including different hash tags and different posts. What about automating video? Instagram reels, tik tok, YouTube shorts? Obviously, video is harder to automate because it's harder to make video, but it's still doable. Here's an example of how you could use an automation to make video content. Let's say you're a musician. You create an automation that downloads short format vertical videos that are trending, and then you automate the process of replacing the audio for those videos with your own music. Then you automate uploading these videos with your own music to TikTok, Instagram, YouTube. Once you've built this automation, you don't have to do anything if the videos do well, which they should because they're already trending, so they have a track record for success already. Then you'll get thousands of views with your own music. Now, I would not say we're quite at the place yet where you can fully automate a well edited video, but you can automate parts of the process of making a video. You could automate the process of writing a video script, generating audio for that script with 11 labs, or creating suitable photos for your video. For example, every day you could wake up with an auto generated text and audio draft for a new YouTube video. You just do the final touches. So automating part of the process is also a very good way to save a lot of time. It's a good way to be able to publish a lot of content without putting in that much work. Let's say you want to build a brand as a business advisor, and you want to post on Instagram, Twitter, and LinkedIn. But you don't want to risk low quality content being pushed through. Well, you could automate the process of summarizing four different newsletters and blogs about business advice. You can create an automation where every day you wake up and Catt has summarized 20 articles and made ten suggestions for you for posts that you could make based on the best advice in those articles that it's summarized. Then you can pick your favorites and revise them, maybe rewrite them in your own words a little bit. Once you've done that, once you've chosen the final output, you place that text into a spreadsheet and then it's automatically posted on all three platforms. Without you having to spend 30 minutes every day doing it manually. Hopefully, this is giving you some ideas of what's possible and what you can do. Maybe you can automate the entire process or maybe you're just going to automate a part of the process. Anyways, there's a lot of possibilities. That's it. I'll see you in the next video. 46. 6.9 (Automation) Automation Ideas: Okay. More automation ideas. Most automations that can help you are not centered around content. Most of the automation ideas that we've talked about so far. Is that to do with content, but here are a few additional ones that I think could be helpful to a lot of people. If you don't want to automate creating the content, you can automate posting the same video to several platforms. If you're making a vertical short form video, you can have an automation that posts it with the same captions to TikTok, Instagram, and YouTube all at the same time, and that's going to save you a bunch of time. Same thing if you have a photo, you can post it to Instagram, Twitter, Linkn, Facebook, all at the same time with an automation. These automations are a lot easier to create than the ones that create the content for you. So this is really a no brainer if you're making content for several platforms. You can track engagement from different platforms in a spreadsheet. So if you're creating content on several platforms, instead of checking the engagement separately on TikTok, Instagram, Twitter, whatever it might be, set up an automation that feeds in your engagement from different platforms into a spreadsheet. It will be automatically updated, and it will be way easier for you to keep track of your engagement across platforms and see what platforms are performing better than others. Can automate e mail responses. For example, you could set up if you receive an e mail that contains a certain thing that is recognizable by Chat EPT. Chat EPT can write an e mail back and send that off that includes that person's name or some other information in that e mail. This is super use case specific, of course. But it's good to know that there's a lot of things that you can do here with e mail. You could make your own newsletter. Based on the methods that we've talked about for retrieving information. Instead of using that information for blog posts or social media, you could set up an automation that writes a we letter every week. Maybe this automation collects information from several other news letters on the same topic over the course of a week. Maybe you have other information sources as well. Then GIPT can, of course, summarize all this information into your own newsletter in your own style. You can identify companies based on their e mail. This is a feature in Saper. You can identify companies based on their e mail. You can get more information about the sender of each e mail based on their domain in their e mail. This is in Saper and Saper has a feature where it can read all the sender addresses for all the e mails that you receive and then look up the domain for that sender and give you back some information like, what's the company? What do they do? How many employees do they have? If you get an e mail from Matt at apple.com, Sapier could look up Apple and tell you, Yeah, you just got an e mail from Apple. It's a big company they make phones and computers. You could create an automation that uploads pictures or PDFs to a folder and then extracts data from that picture or that PDF with C chat GPT. The most obvious use case here would be just uploading receipts and invoices, and then Cat GPT would extract the date, the price, the company that it's from, Then it could insert all this information into a spreadsheet for you so you don't have to do it manually. Here's one that I like. You can make personalized pitch e mails. Let's say you have a list of e mails and names and company names and websites. Maybe you have 1,000 companies with 1,000 people that you want to e mail, but you don't want to send the same e mail to everyone. Well, you could set up an automation where just like in the ones that I've already showed you, Sapier or Make could go through each company's website, extract data about the company for that website, and then give that text that data to Chat EPT, and Chat EPT can write a little summary. For a personalized pitched e mail that you then send off to that company. You don't even have to send off the e mails automatically, but you could write out that one sentence of personalized text to that company using Chachi PT using an automation. I hope you enjoyed these automation ideas. Hopefully, you came up with some of your own for your own use cases, for your own purpose that could benefit you. I really believe that everybody that works with computers can save a bunch of time with automations if they just put a little bit of effort into building.