A declarative standard for programmable video

VideoML is a proposed XML-based standard that describes scenes, timing, and media behavior in a portable, tool-agnostic way.

Why VideoML

Video workflows are fragmented across timelines, templates, and code. VideoML proposes a shared language for describing scenes, cues, and render intent so tools can interoperate.

The goal is a stable, readable specification that supports generation, playback, and long-term archival.

What it describes

Scenes and timelines
Audio, music, and effects cues
Declarative layout and timing rules
Composable components and metadata

Example: a minimal scene

This is a tiny example of VideoML-style structure.

<vml id="intro" fps="30" width="1920" height="1080">
  <scene id="opening" title="Hello">
    <cue id="voice">
      <voice>Welcome to VideoML.</voice>
      <pause seconds="0.4s" />
      <voice>Declarative video, finally.</voice>
    </cue>
  </scene>
</vml>

Spec first, tooling second

The spec is designed to be implementation-agnostic. A reference player will demonstrate the runtime model while other tools can target the same schema.

Next step: We are drafting the core spec and shipping reference implementations you can embed today. Start with docs, browse packages, or install @videoml/player.