A brand-new framework for learning physical character-scene interactions is coming! Some generated motions:

Without relying on any 3D interaction data, our ZeroHSI generates human-scene interaction and human-object interaction leveraging powerful video generation models! For example, "The person is sliding down the slide" and "The person is watering flowers with a watering can".

Excited to share our latest work at Meta FAIR: introducing Meta Motivo (, a behavioral foundation model for controlling virtual physics-based humanoid agents.

DiffSensei is on 🤗 Hugging Face 🤗 Models: 🤗 Dataset: 💡 An excellent way to get an overview in a glimpse: ask paper-central to summarize the paper (⚡by Llama 3.3!) Try it out for any paper:

Will be present CooHOI at Thu, 11 a.m. PST — 2 p.m. PST, at East Exhibit Hall A-C #1601. Drop by if you wanna chat!

🔥Today, we announce the MotionLCM-V2, a state-of-the-art text-to-motion model in motion generation quality, motion-text alignment capability, and inference speed. ✍️Blogpost: 💻Code:

Excited to reveal Genie 2, our most capable foundation world model that, given a single prompt image, can generate an endless variety of action-controllable, playable 3D worlds. Fantastic cross-team effort by the Open-Endedness Team and many other teams at @GoogleDeepMind! 🧞