Introducing 👀Stereo4D👀 A method for mining 4D from internet stereo videos. It enables large-scale, high-quality, dynamic, *metric* 3D reconstructions, with camera poses and long-term 3D motion trajectories. We used Stereo4D to make a dataset of over 100k real-world 4D scenes.

I can’t attend the #NeurIPS conference this year, but Yuanbo Xiangli will present Neural Gaffer in person. Drop by our poster at West Ballroom A-D #7001 if you are interested! Time: Fri 13 Dec 4:30 p.m. — 7:30 p.m.

I hope this attitude is an outlier in research community.

Ever wondered how roses grow and wither in your backyard?🌹 Our latest work on generating 4D temporal object intrinsics lets you explore a rose's entire lifecycle—from birth to death—under any environment light, from any viewpoint, at any moment. Project page:

💥 Think more real data is needed for scene reconstruction? Think again! Meet MegaSynth: scaling up feed-forward 3D scene reconstruction with synthesized scenes. In 3 days, it generates 700K scenes for training—70x larger than real data! ✨ The secret? Reconstruction is mostly

Very cool (and almost *frustrating*) optical illusion. If you pause this video at any time, the shapes disappear. They exist *only* temporally, not visually. This is from Branta Games on YouTube (full video linked below)

🤔Can Generative Video Models Help Pose Estimation? ✅Yes! We find that generative video models can hallucinate plausible intermediate frames that provide useful context for pose estimators (e.g. DUSt3R), especially for images with little to no overlap. 🔗 inter-pose.github.io

Introducing CUT3R! An online 3D reasoning framework for many 3D tasks directly from just RGB. For static or dynamic scenes. Video or image collections, all in one!

🐅 Want to rig your favorite meme character? Try “RigAnything: Template-Free Autoregressive Rigging for Diverse 3D Assets”! ✨RigAnything is a transformer-based model that sequentially generates skeletons without predefined templates. It creates high-quality skeletons for

Our paper LVSM has been accepted as an oral presentation at #ICLR2025! See you in Singapore! We’ve just released the code and checkpoints—check it out here: github.com/haian-jin/LVSM.🚀

Excited to share our work: Gaussian Mixture Flow Matching Models (GMFlow) github.com/lakonik/gmflow GMFlow generalizes diffusion models by predicting Gaussian mixture denoising distributions, enabling precise few-step sampling and high-quality generation.

Supervised learning has held 3D Vision back for too long. Meet RayZer — a self-supervised 3D model trained with zero 3D labels: ❌ No supervision of camera & geometry ✅ Just RGB images And the wild part? RayZer outperforms supervised methods (as 3D labels from COLMAP is noisy)

Bored of linear recurrent memories (e.g., linear attention) and want a scalable, nonlinear alternative? Our new paper “Test-Time Training Done Right” propose LaCT (Large Chunk Test-Time Training) — a highly efficient, massively scalable nonlinear memory with: 💡 Pure PyTorch

Check out Tianyuan’s latest exciting work on long-context memory with efficient test-time training! This simple and scalable technique has broad applicability across various tasks, including large-scale scene NVS, AR video models, LLMs, …

Check out our new work, Self-Forcing! By addressing the training/inference mismatch, Self-Forcing enables real-time streaming video generation on a single GPU while acheiving competitive or superior performance compared to SOTA video models that run significantly slower.

I'll be presenting our work with Kai Zhang at #cvpr2025. We finetune video models to be 3d consistent without any 3d supervision!  Feel free to stop by our poster or reach out to chat: Sunday, Jun 15, 4-6pm ExHall D, poster #168 cvpr.thecvf.com/virtual/2025/p…

I’ll be presenting CAT4D this Sunday at 1:45pm and our poster session will start afterwards at 4pm. Feel free to come and say hi! cvpr.thecvf.com/virtual/2025/o…