“"Write HTML. Render video. Built for agents." — Hyperframes README/docs”
You know that feeling when you want generated videos, but the tool asks you to think like a video editor or rewrite web code into React first? Hyperframes addresses that by making the source file plain HTML with timing data attributes. The before state is translating web layouts and GSAP animations into a video framework. The after state is writing HTML, previewing it in a browser, and rendering it to MP4.
Think of it like stop-motion for a web page. You define a composition in HTML, add timing with data attributes, and preview it in the browser. During render, Hyperframes computes the exact time for each frame, asks a frame adapter such as GSAP what the screen should look like, captures that frame through Chrome's BeginFrame API, and sends the frames to FFmpeg. The key idea is that the animation clock comes from `frame / fps`, not wall time.
If you build agent workflows, video automation, or code-driven marketing clips, Hyperframes is worth a hands-on spike. It fits you if you prefer HTML, CSS, GSAP, and FFmpeg over React component video code. It is not for you if you need mature distributed rendering today or a stable adapter API.
Worth exploring as an experimental repo with strong activity: 9,428 stars, 49 releases, 423 commits, and a release on 2026-04-23. Do not treat it as a drop-in Remotion Lambda replacement because the docs say Hyperframes runs on a single machine today. Start with a local proof of concept and test Docker rendering before you tie it to production output.
Deep-dive insight, Easy and Pro modes, plus action playbooks — the full breakdown is one tap away.