Jeff Tan

CS PhD @Stanford. Previously Robotics @CarnegieMellon

[Zihan Wang @Z1hanW]

Introduce CRISP, a real-to-sim pipeline that recovers human motion and simulatable scene geometry from monocular video!

CRISP builds contact-faithful 3D scene for simulation - 8× fewer sim failures, +43% faster sim, and improves human motion!

Interactive demos👉: https://t.co/locrdrxO16

Exciting collaboration w/ @JiashunWang @jefftan969 @_Tsukasane @ Jessica Hodgins @shubhtuls @RamananDeva

[Hadi AlZayer @HadiZayer]

Our new work, coupled diffusion sampling, allows fast and diverse multi-view editing!
Not by training — but by having one diffusion model guide another. 🤝
As a bonus: it can construct video-editing datasets
and can make video models generate longer videos.
🧵👇

[Jeff Tan @jefftan969]

DressRecon is joint work with @DonglaiXiang, @shubhtuls, @RamananDeva, and @gengshanY. (6/6)

Project page: https://t.co/DtIjOYE0pU
Paper: https://t.co/3iIbG4XtHY
Code: https://t.co/PRc3EJIBe4 (planned in Dec)

[Jeff Tan @jefftan969]

Excited to share DressRecon, which builds 3D human body models with loose clothing and handheld objects from a single video!

DressRecon produces high-quality results by optimizing video-specific body shape and motion (1/6)

https://t.co/IiBWk2PEr2

[Jeff Tan @jefftan969]

The reconstructed body models can be rendered from any view. Below, we show 45-deg and 90-deg novel views. More results on our project page! (5/6)