How to Make Motion Graphics with AI (No After Effects)
Describe the animation and an AI agent writes and renders it - logo reveals, animated titles, social loops. The tested workflow, prompts and the one-word skill.
Two different things get called "AI motion graphics"
One is AI video generation - typing a prompt and getting dreamed-up footage (Veo, Kling, Sora-style tools). The other is rendered motion graphics - your logo, your text, your assets moving on screen the way After Effects would make them move. This post is about the second kind, which is what most businesses actually need: an animated logo, a title card, a subscribe loop, a stat counter.
The workflow that works
1. Attach your assets (optional)
A logo PNG, a product shot, brand colors as a screenshot - or nothing. If you attach nothing, the agent builds the visuals from scratch; if you attach a logo, it recreates the mark cleanly and animates it.
2. Describe the finished animation, not the steps
This is the whole skill. Good: "a 4 second logo reveal, dark background, the mark builds then the wordmark fades in". Bad: "use easing and keyframes". The agent knows the tools - your job is the direction.
3. Say "do it yourself"
If you just ask for an animation, a chat agent will offer to help you plan one. The magic words are "do the animation and render yourself, then give me the finished file" - that is what turns advice into output.
4. Iterate like you would with a human
"Make the build slower", "add a subtle glow", "same thing as a GIF". The agent revises its own render and hands back a new file. You never touch a timeline.
Copy-paste prompts that work
All of these assume you are in a Krater chat (or any agent that can actually run tools). The last sentence is the important one:
"Animate this logo into a short logo reveal video (about 4 seconds, mp4). Do the animation and render yourself, then give me the finished file."
"Create an animated title card: dark background, bold white text that says LAUNCH DAY, letters slide in one by one, 3 seconds, mp4. Do it yourself and give me the file."
"Make a 5-second looping Subscribe animation for YouTube - the button pulses and a bell icon rings - mp4. Render it yourself and give me the file."
"Animate a counter going from 0 to 10,000 users with a green progress bar filling up, 4 seconds, dark theme, mp4. Do the render yourself."
"Animate this product image with a slow zoom-in while a soft shadow appears under it, 5 seconds, mp4. Render it yourself and give me the file."
Make it a skill, run it forever
If you animate regularly, save the workflow as a Krater skill - a reusable assistant you call by name. The prompt that creates it:
"Create a skill called motion-designer: when I describe an animation or attach a logo or image, write and render the animation yourself - logo reveals, animated titles, social loops - and return the finished video file. Never just describe what it would look like."
From then on it is: attach logo, invoke skill, get reveal. No prompt needed twice. (How skills work)
What it is genuinely good at vs. not
Good: logo reveals, animated titles and lower thirds, social loops, stat counters, simple explainers, web banners, GIF loops, product spotlights. Basically: flat 2D motion you would otherwise pay a designer or fight After Effects for.
Not good: 3D character animation, complex compositing, broadcast title packages. Those still belong to a motion designer - be suspicious of any tool that claims otherwise.
What it costs
Animations are billed in credits. Our tested 4 second logo reveal ran about 2 credits; multi-scene work costs more. Krater plans start at $9/month with 500 credits - that is a lot of logo reveals for the price of one template license elsewhere.
We build Krater, so we are obviously biased. The numbers above are from real runs on our own test account - try the logo reveal prompt yourself and compare.
FAQ
Can ChatGPT or Claude make motion graphics?
They can describe an animation or write code for you to run, but they cannot render and return a video file inside the chat. An agent with a sandbox (like Krater's) can - it runs the render on its own computer and hands back the file.
Do I need my own assets?
No. Attaching your logo or product shot makes it on-brand, but the agent can build visuals from scratch if you describe what you want.
MP4 or GIF?
Both - ask for either, and ask for the aspect ratio you need (square, vertical, wide).
How is this different from Canva or Renderforest?
Those are template tools: your content dropped into someone else's animation. The agent writes the animation for your material - custom timing, your brand, no template watermark.