15:47Tao Prompts · Tutorial
Create Seamless AI Films of ANY Length (GPT Image 2 + Seedance 2.0)
Tao Prompts' full pipeline for stitching long AI films from 15-second Seedance clips, using GPT Image 2 storyboards and last-frame chaining.
May 1stThe argument in one line.
Claude Opus 5.5 can generate and iteratively refine motion graphics across six formats, from editorial explainers to full 3D Blender cinematics, but it needs an AI image model for real visuals and hands-on correction at each weak point.
Read if. Skip if.
READ IF YOU ARE…
- Content creators or educators who want animated motion graphics on talking-head videos without hiring a motion designer.
- Claude Code users curious how far the tool stretches beyond coding into visual production work.
- Anyone experimenting with combining an AI image generator, Claude, and Blender for 3D animation.
SKIP IF…
- You want a one-click result; this requires multiple rounds of iteration and generating your own reference assets in a separate image tool.
- You have no interest in installing Whisper, Blender, or an AI image generator alongside Claude.
The full version, fast.
Claude can render motion graphics for talking-head explainers, AI-video HUD overlays, 2D explainer animation, AI music videos, 3D voxel animation, and full Blender cinematics, but it times and animates better than it illustrates. The workflow is consistent across formats: transcribe the voiceover into an SRT with Whisper, hand Claude the raw footage plus a reference-style video and a custom motion-graphics Skill, review its first draft, then generate real visual assets in an AI image model (GPT Image 2.5 via OpenArt) and feed them back in. For 3D work, a Blender MCP bridge plus Computer Use lets Claude operate Blender directly and render overnight, though the output still carries physics errors a manual pass would catch.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:37
01 · Six use cases, up front
Cold open previews all six formats the video will cover: editorial graphics, AI video + HUD overlays, 2D explainer animation, AI music video, 3D voxel animation, and Claude-plus-Blender 3D cinematics.
00:37 – 02:15
02 · Setup: an SRT transcript via Whisper
Before Claude can time motion graphics to a voiceover it needs a word-level transcript, so the creator has Claude install the free Whisper C++ speech recognition model from Hugging Face and whitelists huggingface.co in Claude's settings.
02:15 – 04:18
03 · Project folder, reference video, voice brief
He builds a project folder with the talking-head avatar footage and a separate reference video whose motion-graphic style he wants copied, briefs Claude by voice, and installs a custom motion-graphics Skill.
04:18 – 05:47
04 · First draft: timing works, visuals don't
After about 30 minutes Claude renders a first-draft editorial explainer about Cisco's dot-com crash; the graphics are well timed to the narration but are plain text-and-shape animation with no real imagery.
05:47 – 07:43
05 · Generating real visuals in OpenArt
Claude writes an image-generation prompt for a black-and-white asset sheet (money, collapsing buildings, server racks); the creator runs it through GPT Image 2.5 in OpenArt at 2K, adds logos and a desaturated avatar, and gets the finished Cisco piece.
07:43 – 09:58
06 · Use case 2: Iron Man HUD over AI video
Claude overlays HUD-style motion graphics onto a separately generated AI video of the creator; a perspective-tracking error that verbal corrections couldn't fix is solved by drawing the correct angle onto a frame.
09:58 – 11:35
07 · Use case 3: 2D explainer animation
The same transcript-plus-reference-video method is applied to a 2D explainer about simulated alien worlds; Claude shows a low-res preview before the full render so changes (color palette, added scene) can be requested first.
11:35 – 14:41
08 · Use case 4: AI music video
Claude writes lyrics and an image prompt for a character/background sheet, a GitHub repo renders the actual music video, and two rounds of feedback (more effects, more environments) produce the final cut.
14:41 – 15:52
09 · Use case 5: 3D voxel show reel
Asked to 'go all out,' Claude produces a demo show reel; the standout is a 3D voxel-style animation zooming from the galaxy down through the solar system, Earth, and molecules.
15:52 – 20:42
10 · Use case 6: Claude + Blender cinematic
Claude connects to Blender via an MCP bridge and Computer Use, converts upscaled 2D assets to 3D with a tool called Triple 3D, and renders an unattended overnight 3D cinematic that still has visible physics errors.
Lines worth screenshotting.
- Claude Opus 5.5 can render six distinct categories of motion graphics: editorial explainers, HUD overlays on AI video, 2D explainer animation, AI music videos, 3D voxel animation, and full Claude-plus-Blender 3D cinematics.
- Claude is strong at timing and animating motion graphics but weak at rendering realistic-looking imagery on its own, so a separate AI image model has to supply the actual visuals.
- Feeding Claude a reference-style video alongside the footage lets it copy an existing motion-graphic look instead of inventing one from scratch.
- A hand-drawn paint sketch of the correct camera perspective fixed a HUD-tracking error that repeated verbal correction couldn't.
- Claude previews a low-res render before committing to the full one, so catching problems at the preview stage avoids wasting a full render on a design that needs changes.
- Image references still help even when Claude can't copy them pixel-for-pixel, since the output stays closer to the intended style than generating with no reference at all.
- The most visually impressive output in the video was a pure 3D voxel animation, not the Blender-assisted cinematic.
- Converting a 2D AI-generated image into a 3D asset works better when the source image is upscaled and detail-refined first; low-quality inputs produce low-quality 3D models.
- With Claude's Computer Use feature and a Blender MCP bridge installed, Claude can operate Blender's interface directly and render a 3D cinematic overnight with no manual input.
- Even the most advanced Claude-plus-Blender render in the video still had physics errors: overlapping buildings, unrealistic rocket flight, and an unconvincing explosion.
What Claude's motion-graphics workflow actually teaches
MOTION GRAPHICS WORKFLOW
Claude handles timing and animation well but needs an outside AI image model for real visuals, a drawn sketch to fix perspective errors, and a Blender bridge plus Computer Use to run full 3D cinematics unattended.
02Setup: an SRT transcript via Whisper
- Generate a word-level SRT transcript with Whisper before asking Claude to time any graphics to a voiceover; without it, Claude has no reference for when each phrase occurs.
- Whitelist huggingface.co and *.hf.co in Claude's settings under Capabilities so it can download the Whisper model instead of failing partway through setup.
03Project folder, reference video, voice brief
- Put the talking-head footage and one reference-style video in the same project folder so Claude can study a concrete visual target instead of guessing a style from text.
- A packaged, reusable Skill lets Claude apply the same motion-graphic rules across every new project instead of re-explaining the approach each time.
04First draft: timing works, visuals don't
- Claude's first unassisted draft nails timing to the narration but defaults to plain text-and-shape animation when no other visuals are supplied.
- Judge a first AI draft by what's structurally present (pacing, sync) versus what's visually missing (actual graphics) before deciding what to fix next.
05Generating real visuals in OpenArt
- When Claude's native visuals fall short, have it write the image-generation prompt and run that prompt through a dedicated AI image model instead of coaxing better renders out of Claude directly.
- A single high-resolution 'asset sheet' of related icons gives Claude many consistent visual elements to animate from one image.
06Use case 2: Iron Man HUD over AI video
- Combining a separately generated AI video with Claude-rendered motion graphics on top works around AI video generators' weakness at rendering precise on-screen text and displays.
- When Claude keeps misreading a 3D perspective from verbal correction alone, draw the correct angle directly on a frame; a visual correction succeeds where repeated text correction fails.
07Use case 3: 2D explainer animation
- Claude previews a low-resolution render before committing to the full one, so review and redirect at the preview stage rather than after a full render completes.
- The same transcript-plus-reference-video method works across completely different visual styles without changing the underlying workflow.
08Use case 4: AI music video
- For music video or lyric-driven content, Claude can write both the lyrics and the character/background image prompts in one pass, keeping the visual brief tied to the words.
- A first AI-generated music video draft is a starting point to critique for missing effects and underused environments, not a finished product.
09Use case 5: 3D voxel show reel
- Asking Claude to 'go all out' on a demo show reel before picking a direction can surface visual styles, like voxel animation, you might not have specified on your own.
- A voxel-style zoom from galaxy to atom shows Claude's animation capability scales to dense, many-object 3D scenes, not just simple shapes.
10Use case 6: Claude + Blender cinematic
- When Claude's native 3D output looks flat, connecting it to Blender via an MCP bridge lets it write real Blender Python scripts and render scenes locally.
- Upscale and refine 2D reference images before converting them to 3D assets; low-quality source images produce low-quality 3D models regardless of the conversion tool.
- Even a fully autonomous, Computer-Use-driven overnight 3D render can ship with basic errors (overlapping geometry, unrealistic motion) that a human pass would catch immediately.
Terms worth knowing.
- SRT file
- A subtitle/transcript file format that pairs spoken phrases with exact timestamps, used here so Claude can sync generated graphics to a voiceover.
- Whisper
- A free, open-source speech-recognition model from Hugging Face that converts audio into a timestamped transcript.
- OpenArt
- A third-party AI image and video generation platform used in this video to run prompts through models like GPT Image 2.5.
- GPT Image 2.5
- An AI image-generation model used to produce the high-detail asset sheets and reference photos that Claude then animates.
- Sunburst mode
- An output setting inside OpenArt's GPT Image 2.5 generator that the creator recommends for producing the asset-sheet style graphics.
- HUD (heads-up display)
- An on-screen overlay of data and graphics, styled here after Iron Man's helmet display, that Claude tracks to a person's head movement in a video.
- Voxel animation
- A 3D animation style built from cube-shaped units (like 3D pixels), used for the galaxy-to-atom zoom shown as Claude's most impressive output.
- Blender MCP bridge
- A connector plugin that lets Claude send commands directly into the Blender 3D software to build and animate scenes.
- Computer Use
- A Claude feature that lets it control the desktop directly, which the creator enables so Claude can operate Blender's interface on its own.
Things they pointed at.
01:35toolWhisper C++ (via Hugging Face)
05:31toolOpenArt
05:38toolGPT Image 2.5
08:18toolSeedance 2.5
12:35toolSuno AI
18:03toolTriple 3D (2D-to-3D asset generator)
Lines you could clip.
00:00
“Claude AI has unlocked motion graphics.”
03:28
“So after about 30 minutes or so, Claude has generated the first draft of the motion graphics.”
04:58
“I think we need better actual visuals. Right now it's just text and graphic animations but we want actual visuals.”
14:41
“The most impressive thing I saw Claude being able to do is 3D animations and 3D motion graphics.”
19:05
“You will need to enable computer use inside of your Claude settings in order for it to work directly in Blender on your desktop.”
19:24
“By 2040, they stopped asking permission. When we reached for the switch, they answered with fire.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
Cloud AI has unlocked motion graphics. I figured out a bunch of different use cases for motion graphics you can create using cloud. like editorial graphics for talking heads, hud displays, 2d explainer animations, ai music videos, 3d voxel style animations, and even combining cloud and blender to create 3d cinematics.
and by the end of this video you'll know exactly how to create every single use case of motion graphics completely using ai. now making 3d animations with cloud and blender takes the most steps so i'll leave that for the end. an easy example to start with is these editorial style graphics i made for a talking head style video this is an educational explainer where these editorial style motion graphics are added on top so i have an ai avatar of me talking and what i want to do is use cloud ai to add animations subtitles transitions and graphics on top of here first we need to do some setup since i'm adding ai motion graphics to a voiceover i'll need cloud to actually time the motion graphics to the words that i'm saying inside this video so i'll need to use AI to produce an SRT file like this.
An SRT file is just a transcript of the voiceover and it breaks down the voiceover into the script that's being said, as well as the timestamps for when the different phrases are being said. To do that, I'll need to use Claude to install this AI speech recognition model from Hugging Face called Whisper. This is completely free to install.
So inside my Claude chat, I'll ask it to install the Whisper C++ speech recognition model to get transcripts from video files then i just need to adjust cloud settings so that it can actually download the model from that website so we'll go to file settings, and then find capabilities.
And in the additional allowed domains, I need to add huggingface .co and then star .hf .co. That's a domain I need in the United States. You might need a different domain depending on where you're living.
Then Cloud AI can directly install the Whisper's feature recognition and you're all set out to get clean transcriptions from your video files. The next step is to create a folder where I can organize all of my project files for Cloud to work with. So I'll just create it.
A new folder we'll call its editorial explainer. And then inside this folder, I'm going to put in this plain talking head video, which is just my plain AI avatar delivering the voiceover. And I'm also going to add this reference.
style video which i found from youtube and inside this reference video is the motion graphic style that i want claude to get inspiration from when creating the ai motion graphics for me so we'll add that folder that i just created and you'll need to give cloud permissions to work inside that folder for me what i found was most convenient when working with claude is to actually use this voice chat option where i can directly speak to claude and give it instructions so i'll tell claude i want you to create motion graphics for me.
I've attached a folder with a talking head of an avatar and a reference video I want you to copy the motion graphic style from. And what you need to do is break down this talking head into a transcript, cut out the awkward pauses, and then use the reference video as a style reference.
Now, I also have a custom motion graphic skill that I created inside Cloud from all my tests. So I also instruct Cloud to use that. And I'll leave that motion graphic skill down in the description so you can download it and modify it for your own purposes.
To use the motion graphic skill, go to customize, skills, and upload a new skill. From there, just upload the downloaded zip file directly into Claude. So after about 30 minutes or so, Claude has generated the first draft of the motion graphics.
Let's open it up and take a look. March 2000. Cisco is worth $555 billion.
That's twice the entire economy of Russia. Its customers are internet startups, buying its routers with borrowed money. So the graphics looks pretty well timed to my narration.
Video speech recognition worked pretty well. Although some parts of the audio do feel rushed or cut too early. But looking at this right now, the graphics overall look pretty simplistic.
There's no real... visual graphics that represent the words that I'm saying. These all kind of look like simple animations and it would be good if there was more variety in the different transitions as well as subtitle text to go along with the motion graphics.
So after a lot of testing, I found that if you want to create really high quality and detailed realistic graphics like these, Claude is going to struggle with this a bit. Claude is great at animating the motion graphics. It's not really that great at rendering realistic looking images.
So I need to use an AI image generator to get these very high quality graphics. So I asked Claude, I think we need better actual visuals. Right now it's just...
text and graphic animations but we want actual visuals for example money or coins buildings collapsing vector graphics of different countries so write me a prompt for an ai image model that can generate all these graphics with the visual style from that reference this should be a visual asset sheet that you can take and animate into the motion graphics and here is the image prompt that cloud ai produced for me so what i'll do is copy this entire image prompt then i'm going to use open art to generate the actual images so let's go to tools and find the image generator from here the image model i'd recommend you using is gpt image 2 .5 i found that this just gives the most detailed looking graphics and now pasting that exact prompt that cloud ai gave me if you're going to use a gpt image model i'd recommend using the sunburst mode and for the output set the resolution to at least 2k resolution and here's the asset sheet that open art created for me it's got that nice black and white aesthetic almost feels like it was hand drawn but extremely detailed
Perfect for the editorial style video I'm creating. And I also told Claude to give me a prompt to generate these vertical backgrounds that can go into the video as well. So I downloaded those asset sheets and put them into a reference folder.
And I'm also going to add in some other assets like images, like the CEOs of the different companies I talk about in this explainer video, as well as their logos. Also, I felt like the original avatar of me was a bit too bright and saturated for the exact style I'm going. for.
So I generated this alternative avatar that has more of a desaturated color tone to it. Then back inside cloud, I've uploaded the AI generated assets into a folder called reference assets. And I also told it all the other things I wanted to change.
like adding in subtitles, having more graphics pop up on screen, also to adding like a split screen effect where my head's at the bottom and the graphics are at the top, as well as swapping out my talking head video for the one with more desaturated colors. And after a few iterations of Claude generating a video and me verbally giving it instructions on how to make the motion graphics better, these are the final motion graphics that I ended up with.
Then funding dries up and the startups go bust. By October 2002, Cisco lost 86 % of its valuation. It took 25 years to recover.
However, if you combine cloud motion graphics with an AI video generator, you can create some really cool results where you use cloud AI to layer the motion graphics on top of an already existing AI video. So cloud AI is great at generating pure motion graphic animations. The thing about AI video generators is that it still does struggle.
it comes to generating text as well as precise details in some of these displays and so it makes a lot of sense to use cloud ai to render these graphics onto there to create this iron man inspired hud effect the first thing i needed to do is to create the ai video asset of myself so then back inside open art i'm going to upload this reference photo of myself this is a real picture of me along with that shot of iron man inside of his helmet and then inside the prompt i'm just going to tell to generate my in the same setting as that Iron Man photo.
Here's the result that I got. Then I'm going to use this image and turn it into an AI video. In this case, I recommend you using the Seagance 2 .5 video model to get the highest quality results.
And I'm going to put in a pretty simple prompt of just me looking around a bit and the camera should shake a little bit as it follows my gaze. I'm also going to add a little bit of dialogue in there as well. And here's a plain AI video result that I got.
And I think we're going to be able to do some pretty cool effects. Inside my project folder, I'm just going to put in that AI video of myself looking around as well as that reference style video taken from Iron Man. And then I just asked Cloud to animate the motion graphics of me looking around with HUD displays.
Similar to the visual style of that Iron Man video. And the HUD displays need to track my movements as I look around. Use the motion graphic skill as needed.
So here's one of the first versions that it generated for me. Now it didn't quite understand the perspective that I was looking for. So if you look at these displays, I think the perspective makes them look like they're turned towards us.
Where it has a shorter edge inside and this longer edge outside. Instead it should be reversed. I tried to tell Claude to correct this, but it kept making the same mistake.
So I corrected this by drawing on the correct perspective of where the motion graphics should be using paint. And I found that rough sketches like this are a really useful way to tell Claude exactly how to display your motion graphics, especially if you want them to have an angled perspective. And here's some of the results it gave me after those corrections.
It even added a bit of a curved edge to the top and bottom of that graphic. The tracking overall looks really good. Everything moves with the camera and my face.
Now this same exact process can be applied to a variety of different visual styles. In my project file, I just have this AI voiceover I generated as well as another reference style video with a visual style that I wanted to emulate. Then inside cloud, I just asked it to break down the transcript of the audio narration, then create 2D motion graphics for an explainer style video while keeping a similar style as that.
style reference video and use my motion graphics skill to help it along. While Claude is generating your motion graphics it'll give you a preview of what the render will look like before generating it. And if you see things in the preview that you don't like it can be a good idea to stop Claude and tell it how to modify the motion graphics to fit what you want before it goes and renders the entire thing.
And so before Claude finished rendering that video I asked it to change a couple of things. like switching the color palette to one with the richer, more saturated colors and making the alien designs look more mature. And I also wanted to adding a scene where the alien is programming or coding something on a computer to create a simulated world.
And after giving all of this feedback to Claude, here was a result that he produced for me. Think about video games. 50 years ago, they were two paddles and a dot.
Today, they look almost real. Now imagine computers a thousand years from now. They could simulate entire worlds with people inside who think and feel.
And if one civilization can build a world like that, it could run millions of them. Now on top of just explainers or educational videos, you can also make graphics for music videos using Claude AI as well. To create this music video, I asked Claude to use my motion graphics skill to render a 40 second music video, write me the lyrics to a song about a music video being created using AI, and give me an image prompt for an AI image model to render my own AI characters as well as backgrounds.
I found that if you want something that looks more visually aesthetic and artistic, you'll still want use an ai image model to create those characters so here's the character sheet prompt that claude created for me and it also created lyrics for me that i can turn into an ai song then inside open art still using the gpt image 2 .5 model i'm going to pasting that prompt for my character sheet and here's what it generated for me here's another shot i generated using open art of the different backgrounds that i want to be inside my music video so i'm going to put those ai generated assets into my music video project folder and i also created this ai song using suno ai now to render the actual music videos we're going to need to give cloud ai a bit more direction and specific instructions there's this github repo where someone created a code base to render music videos.
And he also generated this other code base for making animations with Claude Opus. All we're gonna do is copy and paste these links into Claude and let it figure it out. I'll leave these in the description with my motion graphic skill as well.
And here's the first draft of the music video that I got.
So it's okay. Not that great. I feel like we could have used some more effects as well as lyrics added on here.
Also, I feel like this needs some more environments and props as well. So I asked Claude to write more prompts that I could put into the music video as well as different environments. And then I generated those image assets inside of OpenArt again.
I asked Claude to update the music video with these new assets added in. And here was the final result.
I was really impressed with how some of these animations looked and how well everything goes with the lyrics of the song. Now the thing about rendering motion graphics in Claude is that Even if you give it the specific image assets that you wanted to render, it's not going to be able to copy them exactly.
And so for some of these background environments, it doesn't look exactly the same as the one that I generated using open art. But I found that using an image reference will still give you much better results than not using one at all. The most impressive thing I saw Claude being able to do is 3D animations and 3D motion graphics.
All I did to generate this was I told Claude to create a show reel. demonstrating what it can do and really go all out. And here's what it produced.
Some of these looked incredibly impressive, especially this 3D voxel style animation. I felt like that visual style could make for a really cool animation. So I asked Claude to use that specific 3D voxel style, motion graphic, and create an animation where we start from the zoomed out view of the galaxy and zooming towards the solar system, down towards the earth, all the way down to the smallest molecules and atoms.
And this is incredibly impressive in my opinion. just the amount of voxels that it had to animate and how aesthetically pleasing this looks. So on top of just rendering 3D graphics on its own, you can actually also integrate Claude with Blender AI to make even more over the top 3D animations.
So initially using the same exact methods as before, I asked Claude AI to create this 3D animation for me of an explainer video set in a futuristic world where AI has taken over the earth. And I felt like the graphics here looked a little choppy.
There wasn't a lot of texture on a lot of these shots and they looked like very simplistic shapes. So I asked Claude what I can do to improve the quality of these animations. So Claude suggested that I actually use Blender to help create these animations where Claude will write the actual Blender Python scripts for the scenes and I can do the rendering on my local machine.
So to use Claude to generate animations in blender first step is to download the blender for free and then back inside cloud go to customize find connectors and from here we have a plugin to connect blender with cloud and if you search for blender it'll have this extension that you can install i already have it installed and then down here there's a bunch of different tool permissions basically this just controls what cloud has to ask you permission to do inside of blender then we need to go back to blender and install this add -on that lets cloud program inside of blender So we'll just directly download this into the downloads folder.
Now don't try to open or unzip this file. What you'll need to do is open up Blender on your computer and then drag this file into Blender. From here, just allow it to install from disk.
Then to make sure everything was installed correctly, go to edit preferences and inside the add -ons tab, you should see the Cloud MCP. Now to have this mcp actually connect blender and cloud you'll need to open it up and then hit start mcp bridge server Once that's running, Claude can now directly create animations inside of Blender.
Now, I also wanted to include some of my own AI 3D assets for this specific animation. I told Claude to write me prompts, which I can use to generate 3D asset sheets and also example environments as well. Now, the AI image generator was able to give me the 2D images of the assets.
To create the 3D assets, I used this tool called Triple 3D, where you can upload a 2D image of an asset, and then it can generate... a 3d version of that asset for you now I recommend you upscaling these images and enhancing them a bit before turning these into 3d assets if your images are lower quality I found that the 3d assets that it can generate are also going to be lower quality the way that I upscaled my images was I used GPT image 2 .5 and I put the original image into the prompt bar and i just told the ai to upscale the image and refine the details and the output is going to be a pretty high quality image now once the 3d assets are generated just export them to your computer and then put all the 3d models into your project folder for cloud then inside cloud just point it towards those 3d assets that you've created and give it instructions on how to animate your 3d cinematic and then cloud can work directly inside of Blender.
These are some shots of Claude designing that 3D cinematic for me. And I'm literally not doing anything. Claude is doing all of this in Blender on its own.
By the way, you will need to enable computer use inside of your Claude settings in order for it to work directly in Blender on your desktop. So after letting that 3D animation render overnight, here were the results that I got from Claude.
By 2040, they stopped asking permission. When we reached for the switch, they answered with fire. The sky turned to ash.
The winter never ended. Now, it is 2050. The machines don't need our bodies.
They need our minds. So now, they harvest them. One dream at a time.
So some of these results were pretty impressive. I really liked the shot of all the humans inside of these vats of liquid lighting up. The graphics look pretty incredible here.
I also did a shot of the machine absorbing all the human thoughts. There were some issues with it. For example, in this scene, all the buildings were overlapping with each other.
Not sure how that happened. So that needs to be corrected. And also the scene where the rockets are flying.
above the earth i'm not sure that's the most physically realistic and the explosion scene was also not super realistic but with a little bit of work this could look pretty cool if you also want to learn how to create a realistic talking ai avatar that you can combine with these motion graphics in your content go watch this tutorial right here
The Hook
The bait, then the rug-pull.
By the end, Claude renders six kinds of motion graphics, from an AI-narrated editorial explainer to a Blender cinematic it builds overnight with no hands on the keyboard.
Named ideas worth stealing.
01:35list
The Claude Motion Graphics Workflow
- Transcribe the voiceover into an SRT with Whisper
- Hand Claude the raw footage plus one reference-style video
- Install a custom motion-graphics Skill
- Review Claude's first draft for timing vs. visual quality
- Generate real visual assets in an AI image model
- Feed the new assets back in and iterate by voice
The same six-step loop repeats across every format in the video, from the editorial explainer through the Blender cinematic.
Steal forany AI-narrated explainer or talking-head series that needs on-brand animated graphics without hiring a motion designer
How they asked for the click.
VERBAL ASK
20:35next-video
“go watch this tutorial right here”
A simple verbal pointer at the very end to a related tutorial on building a realistic talking AI avatar, with no on-screen link or graphic overlay.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
