mrnerf/awesome-3d-gaussian-splatting

This repository is a curated directory of papers, tools, and code projects related to 3D Gaussian Splatting, a technique for turning photos or video frames into interactive 3D scenes. Instead of storing a 3D scene as solid geometry, the approach represents it as millions of small colored blobs (called Gaussians) that together form a realistic-looking image when rendered. The result is something like a photorealistic 3D snapshot you can walk around inside. The maintainer is actively tracking new research papers on this topic, which has been growing rapidly. There is a searchable web interface at mrnerf.github.io that lists all catalogued papers with filters, so you can browse the field without reading through the raw repository files. Beyond papers, the repo links to a wide range of working code. There are several independent software implementations in different programming languages (Python, C++, and others), each with different speed and portability trade-offs. Game engine plugins for Unity and Unreal are listed, so developers building interactive experiences can drop 3D Gaussian scenes into their existing projects. Web-based viewers running in a browser are also linked, along with desktop viewers for Linux and native apps for iOS and Blender. For people who want to work with the data format, there are conversion tools that translate between different file types the various implementations use. There are also utilities for editing point clouds and preparing camera data. The learning resources section includes blog posts explaining the math and technique at different levels of detail, YouTube tutorials ranging from two-minute overviews to longer walkthroughs, and links to recorded talks from standards-body meetings about where the format might go next. This repo does not contain original research or its own software. It is a living index maintained by one person and open to contributions. Its value is as a single organized starting point for anyone trying to understand the current state of 3D Gaussian Splatting work across the research and developer community.