zju3dv/4K4D

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1807 stars · 78 forks observed · 2026-08-28

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

4K4D is a research implementation of a 4D point cloud representation for real-time dynamic view synthesis at up to 4K resolution, built on the EasyVolcap volumetric video framework. It renders dynamic 3D scenes at hundreds of FPS using hardware rasterization and includes training and inference code with pretrained models.

Use cases

• render dynamic 3D scenes in real time from multi-camera video
• train a 4D view synthesis model from RGB video captures
• render free-viewpoint volumetric video at 4K resolution
• reconstruct dynamic humans from multi-view camera rigs
• experiment with differentiable point cloud rasterization

When to choose

• you need real-time free-viewpoint rendering of dynamic scenes at high resolution
• you have multi-view video captures and want to build a volumetric video
• you want a research baseline for dynamic view synthesis

When to avoid

• you need a production-ready application rather than research code
• you lack a high-end NVIDIA GPU
• you need simple single-image or static-scene 3D reconstruction

Facets

library · maturity active

graphics machine-learning gpu-computing image-processing computer-vision graphics deep-learning machine-learning python view-synthesis neural-rendering 4d-reconstruction volumetric-video point-cloud cvpr-2024 real-time-rendering cuda-rasterization gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="zju3dv/4K4D")