Spark LOD Stream

A minimal example project demonstrating Gaussian splat streaming with SparkJS. Features progressive LOD (Level of Detail) loading, VR support, and mobile touch controls.

Live Demo

https://sparklodstream.netlify.app/

Features

• Streaming LOD: Progressively loads splat chunks (-lod-0.spz, -lod-1.spz, etc.) based on camera distance
• VR Support: WebXR with hand tracking via SparkXr
• Mobile Controls: Virtual joystick for touch devices
• Desktop Controls: WASD/arrow keys + mouse look via SparkControls

Setup

npm install
npm run dev

Build and Deploy

npm run build
npx netlify-cli deploy --prod --dir=dist

Usage

To enable streaming for your own splats, use the paged: true option with a -lod-0.spz URL:

const splat = new SplatMesh({ 
  url: "https://your-cdn.com/scene-lod-0.spz",
  paged: true,
});

The system will automatically fetch -lod-1.spz, -lod-2.spz, etc. as needed.

Project Structure

index.html          - Entry point
main.js             - Scene setup, controls, animation loop
mobile-controls.js  - Virtual joystick for touch devices
lib/                - Local Spark build
  spark.module.js
  spark.module.js.map
  spark_internal_rs_bg.wasm

Updating Spark

To update the local Spark build from ~/projects/spark/:

./update-spark.sh

License

MIT