AI motion graphics, rendered in code
Generative video models paint frames. Motion graphics are composed: type on a grid, timed to a beat, in a palette that has to be exactly right. supaintro generates the composition rather than the pixels, so your team can approve the words and brand decisions before rendering, then revise and re-render the same project when the launch changes.
Generated frames versus generated compositions
A diffusion or video model produces an image sequence. Everything in it, including text, is a guess about what those pixels should look like. That works well for footage of a coastline and badly for a headline, which is why generated video is full of near-words and why product UI in generated video is always a plausible fiction of an interface rather than yours.
A motion-graphics renderer produces a composition: layers, type, shapes, and timings, evaluated deterministically into frames. Ask for the word Ledgerly in Space Grotesk at 96px in #6D5CF5, entering on frame 18, and that is what renders, every time, at any resolution. supaintro is the second kind. An agent writes the composition; the renderer executes it.
| Generative video model | supaintro | |
|---|---|---|
| On-screen text | Approximated, often malformed | Exact, it is real type |
| Brand colour | Close to the reference | The hex code from your site |
| Same input twice | Two different films | The same film |
| Changing one word | Regenerate and hope | Edit and re-render |
| Resolution | Fixed at generation time | Re-rendered at any size and ratio |
| Best at | Footage, texture, imagined scenes | Typography, product evidence, brand systems |
What the system decides, and on what basis
Design decisions in a launch film are mostly about pacing, contrast, and restraint, and those are the ones most easily faked with confident language. So here is what is actually measured.
| Trait | Taken from | What it controls |
|---|---|---|
| Composition | Reference library | How each frame is laid out, and what the eye lands on first |
| Motion | Reference library | How elements enter, settle, and leave |
| Cut cadence | Reference library | How long scenes hold before the next one |
| Typography | Reference library | How type enters and how it is weighted against the frame |
| Pacing | Reference library | The shape of the film across its sixty seconds |
| Brand palette and type | Your product URL | Every colour and typeface in the composition |
What stays under your control without opening an editor
- The words. Every scene's on-screen copy is editable text in a beat sheet, not a rendered layer.
- The palette. Swap the accent and every scene that used it re-renders consistently.
- The timing. Retime a caption or trim a beat and the audio stays locked to it.
- The audio. Take the narrated cut or the music-only one; both come out of the same project.
- The scenes. Reorder, rewrite, or cut them, up to your plan's ceiling of 9.
- The ratio. The same project contains an intentional 16:9 landscape composition and a 9:16 portrait composition. Type, product proof, and hierarchy adapt instead of being cropped or uniformly scaled.
Who this suits
Founders and product marketers who need a launch film and do not have a motion designer. Designers who can see exactly what they want but do not want to spend two days keyframing it for a video that will be re-cut next month. Agencies producing launch assets at volume where the per-film economics of doing it by hand do not work.
It suits people who want live action, avatars, or a recorded product walkthrough considerably less. Those are real formats; this is not the tool that makes them.
Motion graphics questions
Can AI really make motion graphics?
It can write them. The distinction that matters is between generating frames and generating a composition. supaintro has an agent author the scene composition, which a deterministic renderer then evaluates into frames, so the output has the properties motion graphics need: exact type, exact colour, and the same result every render.
What renders the video?
A deterministic renderer that evaluates the composition into frames, rather than a generative model painting them. That is what makes the output identical on every render and re-renderable at any resolution or aspect ratio.
Can I get the project file to finish it in After Effects?
No. The output is video files, not an editable AE project, because the composition is code rather than a timeline. If your workflow requires handing an AEP to a motion designer, this is not a fit. If it requires changing text and re-rendering, that is what the tool does natively.
Does it use my fonts?
It uses the typefaces it finds on your site where they are available to render with, and picks a close pairing where they are not. Brand colours and logo are used exactly.
How is the music made?
It is generated for your film from a direction written out of your brand brief, then mastered to -14 LUFS. Because it is generated rather than licensed from a stock library, there is no track that three other launches are also using this month.
Brand extraction and the beat sheet both run before payment. If the film it plans is not the film you want, you have lost nothing.