ingra14m/Deformable-3D-Gaussians

Health v2 · maintenance only

Adoption not part of the score

1255 stars · 84 forks observed · 2026-08-28

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

Official PyTorch implementation of the CVPR 2024 paper 'Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction'. It reconstructs dynamic scenes from monocular video using deformable 3D Gaussian Splatting with a deformation field, enabling real-time rendering and novel-view/time synthesis.

Use cases

• reconstruct dynamic scenes from monocular video
• novel view synthesis of dynamic scenes
• real-time rendering of 4D scenes
• compare against D-NeRF and HyperNeRF baselines
• research deformable 3D Gaussian splatting
• render time-interpolated frames of a moving scene

When to choose

• you need a high-fidelity, real-time alternative to NeRF-based dynamic scene methods
• you want the official reference implementation of the Deformable-3D-GS paper
• you work with D-NeRF, NeRF-DS, or HyperNeRF datasets
• you need explicit 3D representations for dynamic scenes

When to avoid

• you need a production-ready application with a polished UI
• you lack a CUDA-capable GPU
• you need multi-view or static-scene-only reconstruction with minimal setup
• you need long-term maintenance or commercial support

Facets

library · maturity stable

machine-learning deep-learning graphics image-processing data-visualization computer-vision graphics deep-learning machine-learning python 3d-gaussian-splatting dynamic-scene-reconstruction novel-view-synthesis monocular-video cvpr-2024 neural-rendering research-code linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ingra14m/Deformable-3D-Gaussians")