Gaussian Splatting
AI & Emerging Tech
A way of rendering a captured scene as millions of tiny coloured blobs, fast enough to explore in real time.
In practice
Gaussian splatting represents a scanned scene not as a mesh but as a cloud of soft, oriented ellipsoids (splats) each carrying colour and transparency. It renders photorealistic captures at real-time frame rates and handles the things meshes struggle with: hair, foliage, glass, glow. It has quickly become the practical successor to NeRFs for showing real spaces on the web.
Questions
What is Gaussian Splatting?
Gaussian splatting is a scene representation built from millions of small 3D Gaussians that reproduces a captured space photorealistically and renders in real time.
Why do you use Gaussian Splatting?
It gives you the realism of a photograph with the freedom of a 3D model, at speeds a browser or headset can handle.
How do you use/apply Gaussian Splatting?
Capture the scene with overlapping photos or video, train a splat model from them, then view it in a splat renderer or export for web and spatial devices.
When do you use Gaussian Splatting?
For showing real places and objects immersively (interiors, products in context, exhibitions) where a mesh would look flat or take too long to make.
Who uses Gaussian Splatting?
Spatial designers, architects, product teams and anyone bringing real environments onto a screen or into a headset.
