Transcripts
1. Introduction: AI videos look magical until you start
making one yourself. And honestly, AI videos
have come a long way, but creating consistent scenes is still harder than it looks. Characters change
between the shots, voices don't match, and even
the small details shift. Hi, I'm Nikhil, a marketer and a creator who started
experimenting with AI videos because I
wanted a faster way to make small stories without any heavy equipment
and a lot of courses. After trying a
bunch of AI tunes, I realized each one of them
is strong at a specific part, and the real magic
happens when you put them together in a order. In this class, I will walk you through the exact
workflow that I follow, creating a character,
making a starting frame, turning it into a video, and keeping the
voice consistent. And finally, combine everything
into a one final video. This class will make your
video creation process simple and controllable. So if you want an easy
and reliable entry to the AI filmmaking,
join me in this class. Let's build your multi
scene AI video together.
2. Scripting with ChatGPT: Most important part of
any story is scripting. So we can do our own scripting, but we are going to give
prompts to AI tools. I mean, definitely, we
can do that ourselves, but it's always better to make things faster
with the help of AI. Let AI do all the work.
Let's get into scripting. Okay, so let's start where everything has started
in terms of AI. We are at CharPT
and we are going to generate our script over
here with all the scenes, all the characters, as well as all the prompts that
we are going to require. To create the videos. I will show you the platform that we'll be using its flow. We will be using
frames to video. The idea is to generate
the start frame as well as end frame to maintain the consistency in
the transition. So let me show you what
is the prompt over here. I could have directly given you the prompt, but over here, I want you to look how I think
whenever I write a prompt. I'm just going to write I'm creating a 30 seconds
kids story video. In three D digital
rendering using the tools Google Wisk and Google
flow with view model, give me a script for the story using either
animals or humans. This story should have only two characters to
maintain consistency. So you can see that I'm giving all our pointers to had JBT so that we can get
a perfect output out of it. Then for each scene, I will generate a
start frame image as well as end frame image. Okay, give me prompts for to generate character to
generate start frame, as well as end frame
image for each scene, then detailed PO three
prompt with dialogs, as well as background music, and other things. Please note, I will be creating multiple videos and then stitch
them together at the end. Hence, the transition
must be smooth. Here, I would also like to add with consistent background. I think I have most of the
things ready over here. Now let's see the output. Loomi and Bobo's
glowing leaf adventure. Okay, this is our story. Lumi is a curious little fox and Bobo is a cheerful bear cub. Okay. Now part one,
character generation. C we have character proms. Then scene wise, frame
and frame proms, okay? And then background
for all proms. Okay? All right. I think we
have everything over here. Now, let's head over to is
3. Creating Characters with Google Whisk: That we have our
script scene frame, all of them are ready,
we have to create our characters and we have
to maintain consistency. For that, we are going
to use Google whisk. Let's tag into it. Either of two links would work. This will take you
to labs, Google, and we are going
to launch Wisk to the tool over here we have
subject scene styling, and we have few
other options here, which I'll explain
while working on it. Let's create our
first character, which is Lumi the fox. And I want it in a
white background. Usually, I prefer
full body image so that we can have a proper consistency
throughout our journey. So we have this fox over here, but something that I don't like is the name
written over here or I can just add it
here, no name written. I tell you this should work. Awesome. This one is utter, so I'm going to download it. Um, let's look at it, perfect. Now I'm going to
drag this over here, which is our subject. Subject basically
means characters or whatever you want
to keep consistent. Then let's generate our
second character, Bobo. I would need wide
background and no names, so I'm just going to replace
anything that was before it. Awesome. We have Bobo over here. You can see it's
in good quality, good shape and everything. Let's just drag this as well because these are
two main characters. Now, let's head over back to Cha JPT and create a background, which is usually for
all the scenes, right. So you can see something is checked over
here. Let's uncheck it. This is something
we will be using later to maintain the
consistency. Let's see. Awesome. This background is good enough, so I'm going to download it. It's not really important
to download every time, but I just do it
just in case if I miss the image or if I
find any error over here. You can see the background, I have added it to the scene so that our scene
remains consistent. Now, let's go scene
by scene wise. We'll be creating start frame. Start frame is something while creating video which will
be uploaded over here. I've copied frame one, and you can see it's about Lumi. I'll go ahead and upload and select Lumi over here because we want this
character to be consistent. Now, other thing
that you need to check is precise reference. Click on this button over here, and this should be turned on. This is what maintains the consistency
throughout the project. And you may see some numbers over here,
but it's completely fine. Don't touch it and go
ahead with the generation. This looks good for
our first frame. Let me download it and
add it to the scene. This is what I was
talking about. Sometimes we may see
some mirrors over here. In that case, we can click on the plus icon and upload image. Yeah. Now let me copy the end frame. Usually, start frame and end frame are related
with each other, so we are going to uncheck this and keep
this one as checked. Since the end frame is
also about loomi we'll keep mi checked over
here and then generate. Both of them look identical, but anything works in our case. I'm just going to
download it and you can see it's very related
to each other. All right. So our first thing is completed where Loomis
holding a leaf. Now second scene says
three D render Bobo, the baby bear cub walking
into sunny forest, clearly noticing Loomi afar, joyful expression, same
forest background, and lighting toy story style. Okay. So you can see we have both the characters
in the same scene. So what I'm going to do is
add the prompt over here. Then check on both of them because we want
both the characters. This was our last scene, so I would ideally want
to continue with that. Let's see if we get
the perfect image. Looks good enough. But the problem is in
the previous image, Lumi was looking at that side and now he's here
on the left side. If we change the scene in the video, it will
start from her. In that case, it can create some kind of inconsistency
in the video. We can say towards
Lumi curiously. We can just change the prom
bending from the left. Let's see if we can
get the perfect image. Yeah, this is good. Let's download this one. Yeah, and add it to
one of our seals. Now the second end frame is render Bobo bending curiosity towards it's ideally the sale. But let's see if we get
something different out of it. Now this scene is selected. Ideally, um, I think he should kind of bend more towards it with
this prompt over here. Let's see. Or we will go ahead with this one and download it as a end frame. Not really, all three
of them are identical, let's go ahead and download this one because
there is a difference between these two images where
he completely bends down. The third scene says
it's more about mi Lumi releasing the glowing leaf
upward, watching it rise. Okay. Now in this scene, we don't have Bobo at all, so we can focus only on Lumi. Go back over here where
we just have Lumi and we can uncheck Bobo
and enter the prompt. So you can see it has
maintained the background, but they have removed
Bobo out of it. So think about it that if we connect these
things in the video, it will appear like
Bobo just vanished. So we need a close
up shot of it. I'm just writing here
close up shot of Loomi. Let's see if we get
a perfect image. Let me gloom each face so that
we can focus more on this. This would have also worked because the camera angle
has changed a bit. You can see the background
is a little different. Yeah, this one works. We can consider that Bobo is
standing right over here. And the end frame is this, let me just the So looking upwards, Is
glowing leaf float, let's add a distance
over here one, two, 2 meters above, and again, close up shot of Lomi's
face looking upwards. We are getting the same shot, but I think this one works better because if the
previous shot is this, we can consider that
camera angle has changed. Now, the fourth scene is three D render Bobo clapping excited
while watching this. Let's make them back
to this other scene. And let's add Bobo, select check Bobo
here, and generate. This is good. The leaf is above
and it looks like a consistent thing.
Let's download it. The next the end frame over here is that leaf is
forming in hard shape. Whatever we have downloaded,
I think this one, let's add as a scene see all both of them
are selected and Gin. O. I know. It's not the same. Looking up at the
glowing leaf in the sky, forming a red heart
shape in the air. Let's see if we get
something out of it. And ideally, the bear
should be on the left. See. Perfect. This one is good enough that
leaf color has changed. I will give us story. We have downloaded this. You can see in the forces, all of our frames are ready. In the next video, we will create videos out
of these frames, and we'll have a
complete, I think, 30 seconds, 28 seconds video.
4. Video Creation with Google Flow: Now we have created our script, our characters, our scenes. Now we'll be using Google's app, which is flow to
create our videos. It uses the same VO three tool
and it is same as Gemini, that a little better with a little bit more
features. Let's die in. Now in this video, we are going
to create the videos from the prompt and add
frames as references. I'm just going to search for flow and you will see
a couple of links. You can just click on
any one of these create WFlow new project and maybe just rename this
as Lumi and Bobo. Okay. I will select
frames to video, and there are other settings
which says the aspect issue, I would go ahead with landscape because all of our images
were in landscape. Outputs per prompt, you can get up to four
outputs in a single prompt, but I would usually go with
one because we select four, the cost over here, so
all the free accounts or all the pro accounts get certain amount of
limit over here. So if I use four, my 80 credits will be exhausted. Hence, let's go with one. If something is wrong, we'll
generate another video. There are a few models that
we have to select from. Two of them are no audio, so this is of no use to us. But the other one is
quality video 3.1 quality. This will perfectly creative video just that it's going
to use hundred credits. This is something
that I don't want. Um, I will go ahead with fast because it is only
using printing credits. There won't be much difference, but quality is usually, uh, you won't see
disappointments here. So let me just copy
prong from scene one. Add it over here. Let's select our frames, upload. Yeah, I think this was the
image for the first frame, and second frame would be
let's wait for it to upload. And here, the second that he takes
it in the hats, right? In the first image,
he sees the lease, then takes it in the hand and
we have prompt ready here. Let's just generate the video. So it's ready, and it took around I think 50 seconds to 1 minute. Let
me just play it. Wow. What a sparkly leaf. So you can see that something
has changed, but it's okay. We can start the video here because at the end,
we have to cut it. So this seems perfect, and let's add it to the scene. Once we do that, a
new window will come up where you can see we have
a timeline kind of thing. Let's go back and copy
our second scene. Then select frames to video. Basically, I'm just selecting all the frames and all
the videos over here. I think this was
the first frame. And then this is the
end frame of it. He just sits down over here. We are waiting for
it to rod generate. This is ready as well. Let me just play Loomi, what did you find? I think it's magic. This is good tool. Let's
add it to the scene. You can see it has been added. So if you switch
to Scene Builder, now we have two scenes that are stitched
together. So if I play. Wow. What a sparkly leaf. Louie, what did you find? I think it's magic. Wow. What a sparkly. So you see, there's
inconsistency over here. So what I would do is I would
go back to the project, paste the same prompt, whatever was there and
change the frames. Basically, I'm going to play
around with the frames. In my mind, I think
if I just add again, oi holding one of the leaves, this one, and in the end
frame, let it just load. Here we can have
this as a end frame. There will be a continuity out of you can see our
previous frame ended here, and we can say that a seaman transition from
start frame to end frame Bobo bear cub entering from the left
because it's in the left. Cool. Now just run it. As this scene is ready. Let's see if we have bobo
entering from the left. Loomi. What did you find? I think it's magic. Loomi. Perfect. Idally he
shouldn't have said Loomi, but that is something we'll take care of in the next video. If we switch to scene builder, we can remove this
one over here. So now you can see we have
three videos over here. The first one that we
generated this one where the bear appeared
almost automatically. And this one over here where we had this is
the correct format. If you can just
rearrange by clicking on that button and
move it over here, let's see if we have consistency
since the beginning. Wow. We can just do this like this. Wow. What a sparkly leaf. H oui. What did you find? I think it's magic. Loomi, what did you find? I think it's magic. Wow. So these two videos, the dialogue is same over here, but in our next lesson, I think we can just give
them additional dialogue. I'll keep it here. But if the video is
satisfactory enough, I'll just delete it
this one at the end. So now moving on to
the third scene, and add it to the text box here
frame by frame. Let's select. And so it will show that
camera is moving. You can see slow upward tilt. So you can imagine that there
is a jump cut in the video. This is ready and see. Look, Bobo, it's flying. That's fc, let's add
it to the scene. Move on to the next prompt. And our final scene is
ready. Let me play. Magic is everywhere
when we're together. So perfect. Let
me just add it to the scene and switch
to Scene Builder. So this is our complete movie. And the third one over here
was something unnecessary, if you want to just remove this. So this is our four
scene video over here. Let me play it from
the beginning. Wow. What a sparkly leaf. Loui. What did you find? I think it's magic. Look, Bobo, it's flying. Magic is everywhere when
we're together. Wow. It's a perfect 29 second video of something that we aimed for, and I'm just going
to download it. So videos exported,
just click on Download and you will
have download over here. It's a small 27 MB clip with
our old movie over here. But you can see if you are not satisfied with the
dialogs and everything, we can change the
voices which leave. Something which we will
look into the next video.
5. Backround Music and Voices: Our videos are ready. We have a good 20 to 30
seconds of video, but you will notice it
has a very generic sound, which is not that great. Now we are going to try out some new sounds
using unlevel laps. Alright, so this was our
video that we recorded, which is good enough. Let me just play it and let's see how does the video look? Oh what a sparkly Lee. Loui. What did you find? I think it's magic. Look, Bobo, it's lying. Magic is everywhere
when we're together. So this is good
enough to publish anywhere that a
few small changes, and it might happen
that you may not receive the exact I mean, a video good enough, right? In that case, for example, over here, instead of
Bobo, he says, Loomi. Loomi. But oumi is her name, right? So we have to change that. And also, if you notice the background music is
different for each scene. However, I really like when
he walks on the grass, we can hear the grass
sound. Let me just play it. What did you buy? And then we have some
magical sounds over here. What? Something like a
chime or something, right? And when he claps, we have a clap sound as
well. This works for you. I mean, um, the inconsistency in background voices
and a few dialogues here and there. You can
go ahead with that. But now I'm going
to teach you how to make sure that the
background music is consistent for
which we obviously need a little bit of
video editing skills, which I also show you how
you can do it for free in IvingVF where you can get
the background music, the dialogues, and everything. Let's head over to snot AI, which is so sorry, suno.com. This is an AI tool which gives you all the
AI generated music, songs, or this kind of stuff, anything related to music. Now over here, let's click on Create so I was already trying to come up
with a few sounds over here, but I want to show
you how I did it. So this was our last
answer by the Char GPT. I asked it to give me
a prom for Sno AI. So Cha JBT would have a context. What is Sono to create background music
for the whole story. Another har JBT has already
all the context of the story, and it's better that we continue the conversation
from the same chat. So you can see that we have a
complete proper background. But the problem over
here is this is too big. If I just copy and
paste it here, you will see the
description is too long. So I have asked it,
give me a short prompt. So we have a very short
prompt over here. I'm just going to copy it
and then paste it, right? And let's click on Instrumental because we don't have any lyrics or we don't have any
song whatsoever. So now you can see the
Create button is enabled. Let me click on this. And
we can ignore the V five. This would work perfectly well. The music is being generated. Let me see if I can
play this one. Cool. Good enough. I really
like this one. Currently, now you have
to click on this button over here and you will see that download is not available. For that, you will have to go
ahead and publish it first. So you just have to click on three buttons over here
and click on Publish. Once you publish,
then only you will be able to download this music. Now this is published. Now if I go to Download button, we can download it
in MP three format. Click on Download anyway. Now we have our own
original forstrolic music. Now, other thing is we have
to work on the dialogs. Uh for that, we will
go to level Labs. Now it's a paid tool, but all of us get
10,000 credit for free. You can see it over here. We have 10,000
credits remaining. Now, go to the voices and you will see we have
a library of voices. Just search for the keywords
like cartoon so that we can see we have a couple of options over
here and some are paid. Basically, all the
whoever is the creator of this particular voice,
they will be paid. But we have to select where
we don't have $1 sign. So you can just I think
you can click on the. Hi. My name is Jerry
Twinkle, and I am so, very excited to I
think this goes well for our gloomy
character, right? Now, let me just play some
other voices as well. Blasted video games when
Is your age, I Fun fact. Did you know that sound
travels four times faster? Hi. Welcome to the M. So I think we can go
ahead with this one and BB Bloom over here. Sorry, Nathan. Fun
fact. Did you know? But the problem is
this one is paid. Let me search for
something else. I'm only looking for
English American accent. There are many forms of
strength, and I know them all. Now, see my full power. Yeah, this works well for
our Bobo, the character. So let's go ahead with
Loumi's character first time. I will just go ahead and sorry, so I will just go ahead, go to text to speech. Invoice, you can search. Remember the name,
whatever it was there. So it was Cherry Twinkle. Select this and so you can see, I have written a prompt. Sorry, give me all the
dialogues of Lumi. Now we have all the
dialogues of Lumi, right? So I'm just going
to copy this one. We have this over here. Then I think it's magic. I'm just going to
enter each dialogue in the new line so that there
will be a little bit of pause in between and we
can cut out the dialog. I know I ending is
either of these two, but I think this one is fine. Awesome, let me
just generate it. Wow. What a sparkly leap?
I think it's magic. Look, Bobo, it's flying. Magic feels even brighter
when we share it. Magic is everywhere
when we're together. Awesome. So you can click on
Download button over here. And now we have Loomi's voice. Now I'm going to ask Cha JBT
give me Bobo's dialogues. And I can also search for
Khartoum directly here. I have forgotten the name. I can just close It was Noko Cool. By now, I think, we have all
the bo dialogues over here. Zomi what did you find? Then Fords really flying This is the best magic ever. And then I remember that dialogues were
a little different. I want the dialogues. I mean, obviously, Chad
GBD can make mistakes. You can look this. So I want the dialogues which were mentioned in the frames. Wow, what a spark relief. Bobo says, Lome.
What did you find? Then Okay. So I think Bobo has
only two dialogues. In the previous response, it gave so many dialogues
over here, right? So yeah, I'm just gonna let's just also cross
check with our video. Wow. What a sparkly leaf. Loomi. What did you find? I think it's magic. Look, Bobo, it's flying. Magic is everywhere
when we're together. So these two were
the Dialogs by Bobo, I'm going to congenerate
speech. You can see some. What did you fight? Magic is everywhere when we're together. You can see our creates
are being used. We can slow down the
speed a little bit. I found it to be a little fast. Maybe let's try 8.5. Sorry maybe 0.90. Regenerate. Lumi, what did you fight? Magic is everywhere,
when we're together. I still find it to
be a little faster. Let's make it more slow. Let's try 0.80. Loomi, what did you find? Magic is everywhere
when we're together. I think it's Merton. Let
me click on Download. Now we have three
files over here, which is one of them is
our background music. The second one is Loomis voice, and then third one is
Bobo's voice, right? So now we are going to
go to the video editor. I will be using I
Movie because I think that's the best
option for everyone, um, because not everyone
wants to invest in the high end video
editing tools, right. So like I promised, everything is free for now, and, uh even in Sono, we get around 5,200
credits, I think, which is good enough to generate maybe five to ten voices, sorry, five to ten
background music. In the next video, we will look how we can edit and
bring this all together.
6. Editing Final Video with iMovie: All right, now we are
in the final video and now we are going to stitch
everything together, and our video would be ready. Let's just die in.
Hello, and welcome. In the last video, we
saw how we can generate the voices as well as
the background music. Now we can move over to
our iMovie over here. You can see I was also using it to edit a couple of things. Now I'm just going to put everything together.
This is our movie. And the recent download one
was this background music or then I let me just edit it
so that it's easier for us. The first one was Lumi's voice. And this was Pobo voice. All right. I'm going to add Lumi's voice. I'll cut it later. Sorry, this is Lumi
and this is Pobo. Wow, Lumi talks a lot, I guess. In the background
music, first of all, let's get it till what we have just
right click over here, click on Spit Click
and enter delete. Also, we can reduce
this voice to maybe 1%. Okay. Just Alex mute this voice we're going
to select this clip. We are already in the voice tab, click on this and our original
audio voice is muted. Now, we have to cut these dialogues and
place it the ride places. For that, we'll be
following the same thing. We'll just split it
dialog by dialog split. Split. It's sliding. Split. For. I think one of them was, I will have to delete one
of these because both of them were an alternate
given from the ChatGPT. But anyways, I know it's irritating when I just move
the C plays the audio. But anyways, let's put it aside, Bobo for now, and we'll put Bobo at the end
when Loomis completed. So let me see let me just put them farther away so that we can focus on
one voice at a time. It. Wow. So this was wow, and this
was what a sparkly leaf. Wow. What a sparkly leaf. So I think it will start
a little bit over here. So in the preview, we have to see when he
starts, you know, talking. Wow. What a sparkly leave? And so you can see there
is a small jump cut. So we can also zoom it a little bit so that we have more
frames to look into. M. And I think this is where
we have a jump cut. Yeah, this one. So you can see, since we did everything
frame by frame, we had end frames
and original frames. So we can just, you know, eliminate this
transition as well, which looks very unnatural. So this was our end
frame. Let me just split. And right after death's
transition over here, split and delete it. Let's see how does it look. So now you see the
continuity over here, right? He says something. Maybe we can just duplicate and Is loomi but let's paste
the vow again over here. Okay. And I did Command
V for pasting it. Now Bobo comes in. I think it's magic. He asks, What is it, but we are not going to edit
Bobo's voice as of now. Let me just also split and remove all of
the blank voices. Let me just pull them farther away so that now we can focus
on this one particular. Bob Bobo is coming.
He asked what is it? And We have a little bit of discontinuity
over here as well. Let's split over here and right in the next
frame, let's split again. Delete. So just also ensuring the smoothness over here. Yeah. I think it's magic. And he says this here. No, I think it's
magic it's here. I think it's magic. Yes. And what's the next dine ok? Look, Bogle. It's flying. So more. Look. I'm just also trying to match it with his lip, however
it is moving. Look, Bobo. It's Bobo. Look, Bobo, it's flying. Oh, I think there's a gap
in between hone a bit. Look, Bobo. It's flying. Okay. He claps. Magic feels even brighter
when we share it. Magic is everywhere
when we're together. I think I can delete this one because I remember the dialogues
are a little bit mixed, but let me see. So right where Lumi
opens his mouth, I'm just gonna add
the voice over here. Magic is everywhere,
but we're together. Okay, we are done with Lumi. And let me now I
dare Popos voice. Fight. He says, Loomi
what did you find? So let me just add it here. Fine. Magic is everywhere
when we're together. Let. I'm just gonna split magic is everywhere Dalogu and put
it at the last for now. Oomy Wh did you find? So let's put it at
the right place. Loomi, what did you fight? I think it's magic. And this allo. Magic is everywhere.
What we're together. It could have been
better if I had chosen the right phones, but it's fine. Magic is everywhere.
What we're together? We can change the speed as well. You can see show speed Editor and drag it a
little further uns. Messages everywhere.
We put together. No. The previous one was good, so I'm just going
to do Command Z. Because they acted
at the same time. Now, if you remember, we had certain voices where
he was walking, we had grass walking in a sound, and then few chimes over here. If I want, I can just Google
some of the voice effects. Let me say chime voice
effect, free download. You can go to Pixabay. It's I think free tool. Not this one. Magic. Let me just write
Magic O. That was bad. I think it's the perfect
when the leaf starts flying, I'm just going to doload
it and add it to editor. And reduce the sound a little bit because they want
dialogues to be clear enough. Maybe 20% to do it is fine. Look, Bogo. But leaves start flying here, so you're gonna start it here. Look, Bogo. It's flying. No. I'll just copy and create a duplicate and let's
put it towards the end as well somewhere here. Magic is everywhere,
we're together. Or what happens if
I do it over here. You have to experiment guys, so that's how it is. Magic is everywhere
when we're together. Yeah. Well, it's good. So you can see
something is exceeding. Let's split this, delete it. Otherwise, we'll have a blank
screen when our clip ends. Remove all the extras, and now just let me also try grass sound When the bobo
enters, grass footsteps. This was good. The second one. Let me put somewhere when
he was walking. Yeah. I mean, the middle I like the middle part of this
grass walking sound. Detected its. Wow. Now, he's not walking anymore, so I'm just gonna
split and this clip. Now, you would have
noticed we are not listening to any
background voice, so let me just increase
the sound here. It was zero, so
let's drive at 10%. Now, let's play our movie. Wow. What a spark. Wow.
Louis, put a dpi. I think it's magic. Look, Bobo. It's blind. Oh, wait. So we need a clap, as well. Let me just add it. Just small change and we are
good to go. Clapping sound. Yeah, this one is good. And hurry to the editor where it starts clapping when it's flying o. And he stops here, so let me split and this one. It's flying. Okay. Nat is everywhere,
but we're together. Awesome. Our whole clip is
ready and this was it, guys. You can click on Export button
over here, Export file, make sure it is in 720 P because our original video is in
720 P. Quality is high, compresses faster, and
this is 28 seconds video with 43 MB as estimate five. Click on next, give
the movie title, which is Loomi and
Bobo final edit. Now, see, it is exported. Now let me just open
it and play it. So this is the fine
version of our video. Wow. What a sparkly leaf. Wow. Lully, what did you find? I think it's magic. Look, Bobo, it's flying. Magic is everywhere
when we're together. Awesome. So this was
everything, guys, and I hope you create
more and more videos. You can follow me,
and I have added all the resource link in the description.
Thank you very much.