playcanvas/supersplat-viewer

SuperSplat Viewer – Quick Overview

What it is – A web‑based viewer for SuperSplat 3‑D scenes (Gaussian‑splatting point clouds). It runs as a single static site, using PlayCanvas’ engine and either WebGPU or WebGL for rendering.

Why it matters – Gaussian‑splatting is a modern way to visualise ultra‑dense 3‑D reconstructions (e.g., photogrammetry, NeRF‑style captures). This viewer lets developers and artists load, explore, and debug those scenes directly in the browser, with optional AR/VR support via WebGL.

Core Features (as described in the README)

Feature	Details
Rendering back‑ends	Uses WebGPU when available, falls back to WebGL. WebGL is required for AR/VR.
Stochastic renderer	Experimental unsorted splat renderer (WebGPU only) with developer knobs (`splatSource`, `variant`).
UI controls via URL	Load custom settings, scene files, skyboxes, collision meshes, and tweak UI visibility (`noui`, `reticle`, `ministats`, language).
Performance tweaks	Antialiasing (`aa`), splat budget, high‑precision rendering toggle, full‑load LOD streaming, post‑effect toggle (`nofx`).
Debug & diagnostics	Debug panel (`debug`), heat‑map overlay for voxel collisions, runtime CPU/GPU graphs.
Embedding API	Packaged as an NPM module (`@playcanvas/supersplat‑viewer`) with an API reference for programmatic control.
Local development	Simple Node‑based workflow (`npm install`, `npm run develop`). Can link against PlayCanvas engine debug build via `ENGINE=debug`.

How to Use It

Run the demo – Open the hosted static site (or npm run develop locally) and pass URL parameters to point at your own .ply, .sog, or compressed scene files.
Customize – Supply a settings.json file, skybox image, or collision mesh to enrich the scene.
Embed – Install the NPM package and use the documented API (docs/api.md) to integrate the viewer into other web apps.

Who Might Want This?

3‑D content creators working with Gaussian‑splatting pipelines who need a quick way to preview results.
Developers building web‑based AR/VR experiences that require high‑density point‑cloud rendering.
Researchers testing new rendering tricks (stochastic splats, LOD budgets) without writing a custom engine.

Getting Started Quickly

# Clone & run locally
git clone https://github.com/playcanvas/supersplat-viewer.git
cd supersplat-viewer
npm install
npm run develop   # opens http://localhost:3000

Then append URL params, e.g.:

http://localhost:3000?content=./myScene.compressed.ply&skybox=./sky.hdr&webgl

Bottom line – SuperSplat Viewer is a focused, open‑source tool for visualising cutting‑edge Gaussian‑splat 3‑D data in browsers, with flexible rendering options and a developer‑friendly API.