Skip to content
@IGL-HKUST

IGL-HKUST

Intelligent Graphics Lab at Hong Kong University of Science and Technology
README.md

IGL-HKUST

We are the Intelligent Graphics Lab at Hong Kong University of Science and Technology (IGL-HKUST). We focus on developing new technologies for 3D reconstruction/generation, motion capture/generation, and 3D-aware video-based world models. We focus on the applications of these technologies in video game making, robotics simulators, autonomous driving simulators, and large-scale mapping.

Works

Feedforward 3D Reconstruction

GitHub stars (arXiv'26) Track4World: Feedforward World-centric Dense 3D Tracking of All Pixels

GitHub stars (CVPR'26) UniSH: Unifying Scene and Human Reconstruction in a Feed-Forward Pass

GitHub stars (CVPR'26) MVInverse: Feed-forward Multi-view Inverse Rendering in Seconds

GitHub stars (NeurIPS'25) TrackingWorld: World-centric Monocular 3D Tracking of Almost All Pixels

GitHub stars (CVPR'25 Highlight) Align3R: Aligned Monocular Depth Estimation for Dynamic Videos

3D-aware Video Generation

GitHub stars (arXiv'25) CoMoVi: Co-Generation of 3D Human Motions and Realistic Videos

GitHub stars (SIGGRAPH'25) Diffusion as Shader: 3D-aware Video Diffusion for Versatile Video Generation Control

3D Generation

GitHub stars (ICLR'26) PartSAM: A Scalable Promptable Part Segmentation Model Trained on Native 3D Data

GitHub stars (CVPR'26) MeshMosaic: Scaling Artist Mesh Generation via Local-to-Global Assembly

GitHub stars (NeurIPS'25) Synchronizing 2D and 3D Generative Models for Single-view Human Reconstruction

GitHub stars (SIGGRAPH'25) PDT: Point Distribution Transformation with Diffusion Models

GitHub stars (CVPR'25) PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing

GitHub stars (SIGGRAPH'24) DreamMat: High-quality PBR Material Generation with Geometry- and Light-aware Diffusion Models

GitHub stars (NeurIPS'24) Era3D: High-Resolution Multiview Diffusion using Efficient Row-wise Attention

)GitHub stars (ECCV'24) Disentangled Clothed Avatar Generation from Text Descriptions

GitHub stars (ICLR'24) SyncDreamer: Generating Multiview-consistent Images from a Single-view Image

GitHub stars (CVPR'24) Wonder3D: Single Image to 3D using Cross-Domain Diffusion

Neural Rendering & 3DGS

GitHub stars(ICLR'26) MEGS2: Memory-Efficient Gaussian Splatting via Spherical Gaussians and Unified Pruning

GitHub stars (AAAI'26) NeuralGS: Bridging Neural Fields and 3D Gaussian Splatting for Compact 3D Representations

GitHub stars (SIGGRAPH'23 ToG) NeRO: Neural Geometry and BRDF Reconstruction of Reflective Objects from Multiview Images

GitHub stars (CVPR'23) F2NeRF: Fast Neural Radiance Field Training with Free Camera Trajectories

GitHub stars (CVPR'22) Neural Rays for Occlusion-aware Image-based Rendering

GitHub stars (NeurIPS'21 Spotlight) NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction

6DoF Object Pose

GitHub stars (ECCV'22)Gen6D: Generalizable Model-Free 6-DoF Object Pose Estimation from RGB Images

3D LLM Agent

GitHub stars (JAG'26)SpatialLLM：Enhancing Large Language Models for Urban Spatial Intelligence

GitHub stars (ICLR'25)CityAnchor: City-scale 3D Visual Grounding with Multi-modality LLMs

Popular repositories Loading

  1. DiffusionAsShader DiffusionAsShader Public

    [SIGGRAPH 2025] Diffusion as Shader: 3D-aware Video Diffusion for Versatile Video Generation Control

    Python 833 39

  2. TrackingWorld TrackingWorld Public

    [NeurIPS 25] TrackingWorld: World-centric Monocular 3D Tracking of Almost All Pixels

    Jupyter Notebook 194 16

  3. CoMoVi CoMoVi Public

    Official repository of paper "CoMoVi: Co-Generation of 3D Human Motions and Realistic Videos".

    Python 110 1

  4. SyncHuman SyncHuman Public

    [NeurIPS 2025]SyncHuman: Synchronizing 2D and 3D Generative Models for Single-view Human Reconstruction.

    Jupyter Notebook 77 2

  5. MEGS-2 MEGS-2 Public

    MEGS²: Memory-Efficient Gaussian Splatting via Spherical Gaussians and Unified Pruning

    Cuda 60 6

  6. GO-Renderer GO-Renderer Public

    GO-Renderer: Generative Object Rendering with 3D-aware Controllable Video Diffusion Models

    Python 42 1

Repositories

Showing 10 of 12 repositories
View all repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…