Prompt to explainer
Give it a topic. A three-step AI pipeline researches reference imagery, drafts an editorial concept brief, and authors the finished animation — minutes, not weeks.
Case study · In-house product
Unstill
Animated explainers, generated from a prompt.
Visit unstill.frugalfounder.in ↗ The problem
01 Great explainers take a studio and a month.
Animation is the best way to explain how anything works — an engine, a cell, a star. But producing one means storyboards, animators, render farms, and weeks of iteration, and what comes out is a heavy video file frozen at one resolution.
Unstill treats animation as data instead. An AI pipeline writes the whole scene — shapes, motion, narration — into one small file, and a real-time engine performs it live in the browser. Change a line of the brief and the film changes with it.
Live gallery
unstill.frugalfounder.in ↗ Click any card — these play for real.
What we built
02 A pipeline, a format, and an engine.
A format, not a video
The output is a .unstill file — a compact, self-contained scene with its own design tokens and components built from primitive shapes. Kilobytes where video needs megabytes.
Plays live, anywhere
A lightweight engine renders the scene in the browser in real time — 2D canvas or full 3D — so explainers load instantly and stay crisp at any resolution.
Behaviors, not keyframes
Motion comes from a state machine — orbit, pulse, flow, spawn, ignite — driving named channels each frame. Scenes stay alive instead of looping a baked render.
Narrated storytelling
Every explainer carries its own narrative track with synchronized captions, pacing the visuals like a documentary rather than a slideshow.
The render box
An internal studio tool to search reference images, shape the brief, generate the file, and preview it in the real player before shipping it to the gallery.
Want an AI product built like this?
We build products end to end — the same way we build our own.