ShenhanQian/GaussianAvatars

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1050 stars · 145 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Official research code for GaussianAvatars, a CVPR 2024 Highlight method that creates photorealistic, fully controllable head avatars by rigging 3D Gaussian splats to a parametric morphable face model (FLAME). It supports training from multi-view video, expression/pose reenactment from a driving video, and local rendering via a viewer script.

Use cases

• create photorealistic 3D head avatars from video
• animate a head avatar with expressions from a driving video
• render novel views of a rigged 3D Gaussian avatar
• do research on 3D Gaussian splatting for faces
• reconstruct a controllable FLAME-based avatar from multi-view captures

When to choose

• you need research-grade photorealistic head avatar reconstruction and reenactment
• you want a rigged, controllable 3D Gaussian representation tied to a morphable face model
• you have multi-view video data and a GPU to train on

When to avoid

• you need a production or commercial product - the license is CC-BY-NC-SA and forbids commercial use
• you want a turnkey consumer app rather than a Python research pipeline
• you lack multi-view capture data or GPU resources for training

Facets

library · maturity active

machine-learning computer-vision graphics image-processing simulation computer-vision graphics artificial-intelligence deep-learning python gaussian-splatting head-avatars 3d-reconstruction face-reenactment research-code cvpr-2024 neural-rendering flame linux gpu cuda

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ShenhanQian/GaussianAvatars")