[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.

Native and Compact Structured Latents for 3D Generation

Enjoy the magic of Diffusion models!

Source code for ICCV 2025 paper "FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases"

Official Release of ICCV 2025 paper -- DiscretizedSDF

[SIGGRAPH'24] 2D Gaussian Splatting for Geometrically Accurate Radiance Fields

Wan: Open and Advanced Large-Scale Video Generative Models

[ICLR 2025] OmniPhysGS: 3D Constitutive Gaussians for General Physics-based Dynamics Generation

Official implementation for WorldScore: A Unified Evaluation Benchmark for World Generation

Mamba SSM architecture

Code for PhysDreamer

Original reference implementation of "3D Gaussian Splatting for Real-Time Radiance Field Rendering"

3DGS submodel: simple-knn

[ICCV 2023] Official code for NeuS2

Lightning fast C++/CUDA neural network framework

Code release for NeuS

Instant neural graphics primitives: lightning fast NeRF and more

[ICCV 2025 Oral] MVTracker: Multi-view 3D Point Tracking

Code for PhysDreamer

Official codebase for "Self Forcing: Bridging Training and Inference in Autoregressive Video Diffusion" (NeurIPS 2025 Spotlight)

[CVPR 2024] Official implementation of "Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction"

[ICLR 2024] Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting

[ECCV 2024 - Oral] ACE0 is a learning-based structure-from-motion approach that estimates camera parameters of sets of images by learning a multi-view consistent, implicit scene representation.

ViPE: Video Pose Engine for Geometric 3D Perception