Gaussian splatting for real estate promotion
Gaussian splatting tours offer better fidelity than Matterport, but platform is early.
Metal-accelerated 3D Gaussian Splatting for Apple Silicon
3DGS training in 90 seconds on M4 via fused Metal kernels, no PyTorch overhead—unprecedented Apple Silicon story.
ML engineers on Apple Silicon, researchers needing fast 3DGS training, iOS/macOS app developers
gsplat · taichi-3dgs · threestudio
So I wrote the whole training pipeline from scratch as Metal shaders: projection, tile-based rasterization, SSIM loss, backward pass, Adam, and densification. Everything runs on the GPU
msplat trains 7k iterations of full-resolution Mip-NeRF 360 scenes in ~90s on my M4 Max. In the README I compare against gsplat's published numbers, which were measured on a TITAN RTX. Ofc these are different hardware classes, so take the wall-time comparisons with a grain of salt
Python bindings are on PyPI (pip install msplat), and there are Swift bindings if you want to embed this in a native app. Happy to answer questions about any of the internals
Repo: https://github.com/rayanht/msplat (Apache 2.0)
Gaussian splatting tours offer better fidelity than Matterport, but platform is early.
Browser-only 3D splat generation beats cloud APIs with WebGPU and WASM fallback.
Custom rhêgma software sculpts Gaussian splat clouds into vapor-like artistic distortions.
Using an SVO to voxelize Gaussian splats is a sensible way to prune overlap checks — hierarchical voxels fit the problem and should cut costly pairwise collisions. Can't judge the execution: the Reddit thread is blocked with no visible code, benchmarks, or demos, so this currently reads like an intriguing sketch rather than a drop-in tool.
CPU only Gaussian Splat rendering over SSH beats launching a desktop viewer.
CPU Gaussian splats in the terminal at 10-25 FPS—six render modes, full 3D nav, ships today.