3
5 Comments

I launched an open-source toolkit for generating videos from JSON

https://www.producthunt.com/products/videoflow?launch=videoflow

Hey everyone,

I just launched VideoFlow, an open-source TypeScript toolkit for creating videos from code.

The simplest comparison is probably Remotion, but the core idea is different: instead of treating the video as a React component tree, VideoFlow treats the video as portable data.

A video is described as JSON:

scenes, layers, timing, transitions, effects, keyframes, render settings, etc.

That JSON can then be rendered in different places:

in the browser, on a server, in a React preview/player, or eventually inside a visual editor.

I built it because I think a lot more video will become generated rather than manually edited: product demos, personalized marketing clips, AI-generated explainers, social templates, reports, and so on.

Current version includes:

  • Apache-2.0 open source core
  • TypeScript builder API
  • portable VideoJSON format
  • browser/server/live-preview render paths
  • 27 transitions
  • 42 GLSL effects
  • layer groups
  • React video editor component

I’m still early, so I’d really love feedback on the positioning.

Does “video as JSON” feel clear and interesting, or should I frame it more directly as an open-source video generation toolkit?

Link: https://videoflow.dev

submitted this linkon May 11, 2026
  1. 1

    This is a sharp direction.

    Treating video as portable structured data instead of a component tree is a much more interesting angle than just “create videos from code.”

    One thing I’d watch is the name. VideoFlow explains the category, but it also sounds close to a lot of generic video workflow tools. The product itself feels more like a developer engine for programmable video generation.

    If this grows beyond an open-source toolkit into serious video infrastructure, a more ownable .com like Exirra.com would probably age better than staying inside the descriptive “video workflow” frame.

  2. 1

    that's great toolkit, may be i will try it.

    1. 1

      did u tried it yet?

      1. 1

        not now, may be later