High-Fidelity 3D Shape Generation via Scalable Geometric Refinement

Pixio: a capable vision encoder dedicated to dense prediction, simply by pixel reconstruction

JoyCaption is an image captioning Visual Language Model (VLM) being built from the ground up as a free, open, and uncensored model for the community to use in training Diffusion models.

Taming large-scale few-step training with self-adversarial flows! 👏🏻

Glance: Accelerating Diffusion Models with 1 Sample

PyTorch implementation of NEPA

Towards Scalable Pre-training of Visual Tokenizers for Generation

A Cross-Platform Backend for High-Performance Sparse Convolutions

[NeurIPS 2025] Code for Low-Rank Head Avatar Personalization with Registers

Native and Compact Structured Latents for 3D Generation

Sharp Monocular View Synthesis in Less Than a Second

"E-RayZer: Self-supervised 3D Reconstruction as Spatial Visual Pre-training" official implementation.

Ovis-Image is a 7B text-to-image model specifically optimized for high-quality text rendering, designed to operate efficiently under stringent computational constraints.

Light-X: Generative 4D Video Rendering with Camera and Illumination Control

DGGT: Feedforward 4D Reconstruction of Dynamic Driving Scenes using Unposed Images

[ICLR 2025 Oral] Official code for "LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias"

Offical code for "FastGS: Training 3D Gaussian Splatting in 100 Seconds"

Agility Meets Stability: Versatile Humanoid Control with Heterogeneous Data

HunyuanVideo-1.5: A leading lightweight video generation model

The repository provides code for running inference with the SAM 3D Body Model (3DB), links for downloading the trained model checkpoints and datasets, and example notebooks that show how to use the…

DiffusionNFT: Online Diffusion Reinforcement with Forward Process

Official implementation of HYPIR: Harnessing Diffusion-Yielded Score Priors for Image Restoration (SIGGRAPH 2025)

Official implementation of the CVPR'25 Paper "Towards Lossless Implicit Neural Representation via Bit Plane Decomposition"

Seed1.5-VL, a vision-language foundation model designed to advance general-purpose multimodal understanding and reasoning, achieving state-of-the-art performance on 38 out of 60 public benchmarks.

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

🚀🚀 「大模型」2小时完全从0训练26M的小参数GPT！🌏 Train a 26M-parameter GPT from scratch in just 2h!