Official code for "LagerNVS Latent Geometry for Fully Neural Real-time Novel View Synthesis" (CVPR 2026)

Our method reconstructs 3D worlds from video diffusion models using non-rigid alignment to resolve inherent 3D inconsistencies in the generated sequences.

Need for Speed: Zero-Shot Depth Completion with Single-Step Diffusion

[CVPR 2026] "LaS-Comp: Zero-shot 3D Completion with Latent–Spatial Consistency"

Holi-Spatial: Evolving Video Streams into Holistic 3D Spatial Intelligence

LATTICE: Democratize High-Fidelity 3D Generation at Scale (CVPR'26)

Global LiDAR-Visual Bundle Adjustment

使用 AI Agent 在单 GPU 上自动训练并调优大语言模型

AI agents running research on single-GPU nanochat training automatically

[CVPR'26] SimRecon: SimReady Compositional Scene Reconstruction from Real Videos

RelaxFlow: Text-Driven Amodal 3D Generation

TexSpot: A diffusion-based 3D texture enhancement framework with spatially-uniform point latent representation for boosting visual fidelity, geometric consistency and robustness of 3D texture gener…

Collection of forcing related autoregressive video Gen

Scalable View Synthesis Model [CVPR 2026]

Elevate your AI research writing, no more tedious polishing ✨

[CVPR '26] SceneTok: A Compressed, Diffusable Token Space for 3D Scenes

Native and Compact Structured Latents for 3D Generation

Paper list of agent for science

Code for the ShapeR research paper

Depth Anything 3

D3-Human: Dynamic Disentangled Digital Human from Monocular Video

Train, inspect, edit, automate, and export 3D Gaussian Splatting scenes from a single native application.

[CVPR 2025 Oral & Best Paper Finalist] Difix3D+: Improving 3D Reconstructions with Single-Step Diffusion Models

YAAC: Another Awesome CV is a template using Font Awesome and Adobe Source Font.

[CVPR 2025 Highlight] Matrix3D: Large Photogrammetry Model All-in-One

[3DV 2025 Oral]: A Large-scale Dataset of Gaussian Splats and Their Self-Supervised Pretraining

Original implementation of "Radiant Foam: Real-Time Differentiable Ray Tracing"