“"Think of SuperSplat as the photoshop of gaussian splats?" — slimbuck (SuperSplat developer), Hacker News item #42060856 (https://news.ycombinator.com/item?id=42060856)”
You know that feeling when you capture a Gaussian Splat scan — a room, an artifact, a product — and the raw PLY file is full of floating artifacts, off-axis Gaussians, and background debris that make the result look unusable? Cleanup used to mean re-capturing (time-consuming), writing custom Python scripts (slow), or installing SplatForge in Blender (steep setup curve). Every existing GS editor required a local GPU, a desktop OS, or a paid license — which blocked the handoff-to-client workflow entirely.
You drag a .ply Gaussian Splat file into superspl.at/editor, and the PlayCanvas WebGPU engine renders every Gaussian — each a tiny, oriented, colored ellipsoid — as an interactive scene in your browser. You draw a selection box around noisy regions and delete them, adjust colors, and set camera keyframes on a timeline for animated fly-throughs. Clicking Publish packages your edits into a .ssproj file (a ZIP of JSON metadata plus PLY data), uploads it, and returns a shareable superspl.at URL viewable by anyone in a browser. For large scenes, the SOGS (Streamable Occupancy Grid Splats) format streams chunks progressively so 10M+ Gaussian scenes appear incrementally rather than blocking on a full download.
If you work with Gaussian Splat captures — photogrammetry artists, real estate visualization teams, cultural heritage digitizers, or 3D content creators using Luma AI or Polycam — SuperSplat is your cleaning and delivery step before handing scenes to clients or embedding them in web pages. If you build frontends and want an MIT-licensed WebGL/WebGPU GS renderer to fork, the TypeScript codebase is the most active open-source option in the space. Not the right fit if you need to reconstruct splats from raw photos, composite GS with polygon meshes in a DCC tool like Blender, or run a multi-user ...
Worth your time: the 2–3 week release cadence, company backing from PlayCanvas, and 7,050 stars show this is not abandonware. The zero-install browser path means you can evaluate it in 5 minutes at superspl.at/editor. Caveat: an open WebGL/WebGPU crash bug on the SOGS format (issue #864) and the Separate command failure on large selections (issue #868) mean you should test your specific scene sizes before relying on it in a production pipeline.
Deep-dive insight, Easy and Pro modes, plus action playbooks — the full breakdown is one tap away.