Gaussian Point Splatting

Hacker News Top Papers

Summary

Researchers propose Gaussian Point Splatting, a stochastic rendering method using pixel-sized opaque points and 64-bit GPU atomics that scales to hundreds of millions of Gaussians in real time. The method, accepted at SIGGRAPH 2026, employs hierarchical culling and parallel programming primitives to achieve even workload distribution with only minor noise differences compared to original Gaussian splatting.

No content available
Original Article
View Cached Full Text

Cached at: 06/05/26, 02:12 AM

# Gaussian Point Splatting Source: [https://momentsingraphics.de/Siggraph2026.html](https://momentsingraphics.de/Siggraph2026.html) Moments in Graphics A blog by Christoph Peters Joris Rijsdijk, Christoph Peters, Michael Weinnman, Ricardo Marroquim\. 2026–07 in*ACM Transactions on Graphics \(Proc\. SIGGRAPH\)*45, 4\. [Official version](https://doi.org/10.1145/3811272) ## Abstract We propose Gaussian point splatting, a stochastic method to render Gaussian splats that scales extremely well to scenes with many Gaussians\. Our core idea is to sample pixel\-sized, opaque points from the Gaussians and to splat them to a framebuffer using 64\-bit atomics\. Through parallel programming primitives, we achieve an even distribution of the workload across millions of threads\. Since these threads splat points independently, multiple points may splat to the same pixel\. That makes it non\-trivial to determine how many points should be splatted for a Gaussian or how they should be distributed to achieve the desired opacity\. We successfully formalize and solve these problems, thus keeping our renders faithful to the original Gaussian splatting\. To further accelerate our method, we employ hierarchical frustum and occlusion culling\. Our method renders hundreds of millions of Gaussians in real time\. The only differences compared to the original Gaussian splatting are slight noise and differences in aliasing\. **Keywords:**novel view synthesis, gaussian splatting, large\-scale scenes, GPU atomics, point rendering, parallelism ## Images [![representative_image](https://momentsingraphics.de/Media/Siggraph2026/representative_image.webp)](https://momentsingraphics.de/Media/Siggraph2026/representative_image.webp)- [Supplemental results \(interactive online viewer\)](https://momentsingraphics.de/Media/Siggraph2026/rijsdijk2026_gps_image_viewer/index.html) ## Notes This work gets presented at SIGGRAPH 2026 on 20th of July\. The author's version has been published on 20th of May 2026\. ## Downloads and links - [Paper](https://momentsingraphics.de/Media/Siggraph2026/rijsdijk2026_gps_paper.pdf) - [Joris' project page](https://jorisar.nl/gaussian_point_splatting/) - [Supplemental video \(no temporal reprojection\)](https://momentsingraphics.de/Media/Siggraph2026/rijsdijk2026_gps_2x2_k4_1spp.mp4) - [Supplemental video \(with temporal reprojection\)](https://momentsingraphics.de/Media/Siggraph2026/rijsdijk2026_gps_2x2_k4_1spp_reprojection.mp4) - [Supplemental results \(image viewer download\)](https://momentsingraphics.de/Media/Siggraph2026/rijsdijk2026_gps_image_viewer.zip) - [Source code \(download\)](https://momentsingraphics.de/Media/Siggraph2026/rijsdijk2026_gps_code.zip) - [Source code \(github\)](https://github.com/JorisAR/gaussian-point-splatting) - [Shadertoy for opacity correction](https://www.shadertoy.com/view/WXdyWr)

Similar Articles

ZipSplat: Fewer Gaussians, Better Splats

Hugging Face Daily Papers

ZipSplat is a token-based feed-forward 3D Gaussian Splatting model that uses k-means clustering to decouple Gaussian placement from the pixel grid, achieving ~6x fewer Gaussians while setting new state-of-the-art results on DL3DV and RealEstate10K without requiring ground-truth poses or intrinsics.

GlobalSplat: Efficient Feed-Forward 3D Gaussian Splatting via Global Scene Tokens

Hugging Face Daily Papers

GlobalSplat introduces an efficient feed-forward framework for 3D Gaussian splatting that achieves compact and consistent scene reconstruction using global scene tokens, reducing computational overhead and inference time to under 78ms. The method uses a coarse-to-fine training approach to prevent representation bloat while maintaining competitive novel-view synthesis performance with significantly fewer Gaussians (16K) compared to dense baselines.

3D Gaussian Splatting in a Weekend

Hacker News Top

A tutorial on building a simplified 3D Gaussian splatting renderer from scratch in C++ and OpenGL, covering loading and rendering Gaussian splats.

Gaussian Splat of a Strawberry

Hacker News Top

A Gaussian splat of a strawberry, created from 90 perspectives with focus stacking, using slang-splat for training and SuperSplat for viewing.