I took one picture of my coloring book cover and turned it into a 30-second ad with three shots, using Seedance 2.5 on Higgsfield AI. I didn't film any of it. I wrote the prompt by the second with timestamp control, and you'll see what each generation cost in credits.
Seedance 2.5 can start from text, an image or your own footage, runs up to 30 seconds in one generation, and can put several shots inside that 30 seconds.
⏱ Chapters 0:00 A 30-second ad from one photo 0:35 The only thing I started with 1:09 Higgsfield AI and Seedance 2.5 1:44 The settings: 30 seconds, 16:9, multi-shot 2:42 Writing the prompt by the second 4:34 Generating it (480 credits) 5:14 Did it keep my cover? 5:27 Changing only the camera with Edit Video 6:35 The funny part 7:07 What I'd do with this ad
📝 The exact prompt I used (steal it): A 30 second product ad for a printed coloring book, three shots. Use the uploaded image as the book's front cover. 0-5s: Close-up of colored pencils rolling slowly across a wooden kitchen table in soft morning light. Slow push in. No people yet. 6-12s: An adult in a sweater sits down at the table after a long day, sets down a mug, and lets out a breath. Warm, quiet, natural light. 13-20s: Their hand slides the coloring book into the middle of the table. The front cover is clearly visible for the first time. Medium shot. 21-30s: Hero close-up of the front cover, sharp and centered, colored pencils beside it. The camera moves slowly around the book. Hold completely still for the last three seconds. Keep the cover art and title exactly the same as the uploaded image in every shot. No words, captions or logos on screen. Room sound only, no music, no voices.
Seedance 2.5 · Higgsfield AI · AI video generation · multi-shot AI video · text to video AI · image to video
🙏 Faith + Money: "Whatever you do, work heartily, as for the Lord and not for men." (Colossians 3:23 ESV)
⚠️ Transparency: Some links may be affiliates; I only recommend tools I'd use myself.
That is a 30 second ad that I made for my coloring book.
I didn't film any of it.
I started with one picture of the cover.
Seed Dance 2.5 on Higgs Field made every shot.
I'm gonna show you how I directed it second by second
and what it actually cost to make.
This is the only thing that I'm starting with.
It's just one picture of the cover,
the same kind of image that you'd put
on like a product listing.
I put this on my actual coloring book that I put on Amazon.
If you sell anything on the side, a book,
a template, a candle, you've probably run
into this.
You need a video to sell it,
and a video means a product shoot, a camera, lights, a table
that doesn't look like your kitchen, or paying somebody to do all that.
So today I'm going to take this one picture,
I'm going to make a real ad out of it, 30
seconds, three shots, and I'm going to write it
the way a director would by the second.
And we're going to do all of it with AI.
So this is Higgs Field AI.
It's a website where you make AI video and images.
The model I'm using today is C-Dance 2.5.
And quick thing before we start,
I'd love for you to follow along with me on this.
Higgs Field works best on a desktop computer.
So if you're following along,
it works better on your computer rather than on your phone.
And I'm going to write down my credits before I make anything.
So at the end, you'll see exactly how much it costs.
I have $1,975 to start with.
We'll see what I have when I'm done.
So we're going to go to the video tab here.
And there's a whole bunch of things that we could do.
We're just going to go to create video.
We're going to keep it pretty simple here.
And the main thing that we want
is to make sure the model is Seed Dance 2.5.
We're going to make the duration 30 seconds long.
The aspect ratio, we're going to keep it 16 by 9,
so that's a horizontal video like a YouTube video
like you're watching here. And C-Dance 2.5 can do up to 30 seconds.
That's twice as long as the model
before it, and it doesn't have to just be one long shot.
You can put several shots inside of that 30
seconds. So this is a multi-shot AI video in a single generation,
so it's going to save a lot
of time. Rather than generating six five-second shots,
we're just going to do one 30-second shot.
And there's three ways you can start one of these.
You can just do text to video AI,
where you just describe it, you know, by words.
You can also start from an image,
or you can start from your own footage.
So I'm going to start from the image
because I need the book to look like my book.
So I'm going to add the reference image
of this coloring book cover and that's all that we're
going to need to get started here.
So then we have quite the prompt that we're putting here
and I did not come up with this prompt myself.
I actually worked with Claude and I kind of said
what I was looking for and I let Claude put together
the prompt for Higgs field but a 30
second product ad for a printed coloring book.
This is the Amazon KDP book I did. Three shots
use the uploaded image as the book's front cover.
So again we're just saying in normal English
What we're looking for.
And then I give it timestamps.
So zero to five seconds, I want a close-up of colored pencils
rolling slowly across a wooden kitchen table in soft morning light.
Slow push-in, no people yet.
Then six to 12 seconds,
an adult in a sweater sits down at the table after a long day,
sets down a mug, and lets out a breath of warm and quiet natural light.
Hopefully you're getting a picture in your head.
What usually happens is I have in my head an idea of like,
okay, here's how I want this to look.
I start working with Claude to say,
here's kind of my general idea.
And then Claude puts it into a prompt like this
with you know what seconds should show what.
13 to 20 seconds, their hand slides the coloring book
into the middle of the table.
The front cover is clearly visible
for the first time, medium shot.
So that's where it will show the actual cover
for my coloring book.
Then 21 to 30 seconds, hero close-up of the front cover,
sharp and centered, colored pencils beside it.
The camera moves slowly around the book,
hold completely still for the last three seconds.
So that is the video.
And then a few qualifications that I give it.
Keep the cover art and title exactly the same
as the upload imagery in every shot.
So make sure you're showing my coloring book.
No words, no captions, no logos, no on-screen.
Room sound only, no music, no voices.
So this is called timestamp control.
So instead of writing one paragraph
and then you know hoping that Hicks Field gets
what I had in mind correctly,
I say what I want for each timestamp in the video.
All right, this is gonna cost 480 credits.
Let's go ahead and generate.
We'll see what it comes up with.
All right, here's what it came up with.
Okay. So first thing I really care about, is this the same cover?
And this is, this is the same image
that I had as my cover. So it followed that really well.
This is what it looks like as a coloring
book. And it did, it took the 480 credits that it said it would.
Now I want to test something with
this. I'm going to keep everything that happens in the video,
the reveal, the person, the table,
the book, everything the same,
but I want to change the camera. So it's like you have two
different cameras recording the same thing from the same video,
but different angles.
So I'm going to go to edit video here,
and this is the beauty of AI on a real shoot.
If you didn't have two cameras recording it,
you're going to have to redo it and it might
not be quite the same.
So I'm going to do edit video here.
Go to video generations, we're gonna do the one
that we just did, confirm.
We're gonna take this video and the prompt we're gonna do
is keep the people, the action, the lighting,
and the look exactly the same,
but change only the camera in 13 to 20 seconds.
Shoot it from directly overhead,
looking straight down at the table
so the book slides into the center of the frame.
So we're gonna do the same video,
but the camera should be up above now.
599 credits, let's go.
So the same video at the beginning here.
So here's the funny part. The first version already shot
the reveal from straight above.
The timestamp prompt got it right the first time,
so the edit didn't actually need to change
anything. I realized as I was watching it,
so that was a waste on my part, but that's best case.
Honestly, I didn't need to fix anything.
It was the way I intended the first time,
but it did take my 599 credits.
So the moral of that story is you want to be cautious
and make sure you're doing something that needs
to be done before you spend the credits.
But I have a working ad here.
These took about three minutes to generate. In total,
this took me however long this video is,
maybe like 10 minutes to get this full ad done. In the
good old days, it would have taken a camera,
lights, and a whole day at least. I've made a
lot of videos like this in the past for previous businesses.
Maybe you pay $5,000 for a video like
this and just AI has made it so much easier.
AI video generation has come a long ways.
So what I actually could do with this for my KDP coloring book
is definitely put it on the product
page. This is a good video. I'd cut the hook into a short.
I could generate maybe a similar video
at nine by 16 so that I can put it at, you know,
as a YouTube short or an Instagram reel.
And feel free to reuse my prompt.
So I'll put my prompt in the description.
You can steal it, use whatever you want,
but I just made it with Claude.
You know, just talk back and forth with Claude
till you get the prompt of what you're looking for.
But that's Seed Dance 2.5 on Higgs Field AI.
I gave it one picture, directed it by the second,
got back a 30 second multi-shot ad in one prompt.
Link is in the description.
Go check it out.
Follow along what I did today.
Again, works better on desktop rather than your phone.
But thank you so much for watching.
I love reviewing AI tools.
I love AI video generation,
a lot of fun stuff that you can do with it,
and it's getting better and easier and cheaper day by day.
Thank you guys so much for watching.
God bless you all, and have a great rest of your day.