René P.

@DustinRolstad @EugeneDyabin It is Cascadeur, an animation software with auto posing and AI supported in between framing. It also has some Mocap capabilities, though the powerful AI posing is an alternative to actual motion capture.

@EugeneDyabin I do not know what that is. Is it a keyframe interpreter or something? Mocap software?

From single axel to a quad in under 10 minutes in Cascadeur

@Dadojvk @SpenserFX Pixal3D is built upon Trellis 2. Now Pixal3D is also made MIT.

@SpenserFX Wait Pixal3D + Trellis.2? Why both? Don't they do the same thing?

Synthetic splat training from a single image using the new Pixal3D + Trellis 2 backbone.. 9.4m polygons wrapped up into a 9mb splat asset.. over 200 image to splat assets made over the weekend, now just training a few data sets to see how they look

@ZGojcic @kangxue_yin @TianshiCao @jiawei6_ren Will you release thre TokenGS models too?

An Image-to-3D model and end-to-end pipeline designed and optimized for messy real-world data. Asset Harvester outperforms SOTA image-to-3D models and unlocks new levels of neural simulation for PhysicialAI. Amazing work led by @kangxue_yin @TianshiCao and @jiawei6_ren!

@kangxue_yin Is the release of the related full TokenGS still planed?

AH enables SDG using real object 3D assets harvested from driving videos. The pedestrian animation below is implemented through integration with SOMA(github.com/NVlabs/SOMA-X) and Kimodo(research.nvidia.com/labs/sil/proje…). [5/5]

🚀We just released Asset Harvester, an image-to-3D model and end-to-end pipeline that extracts real object assets from autonomous driving videos! 🌐 Website: research.nvidia.com/labs/sil/proje… 💻 Code: github.com/nvidia/asset-h… [1/5] #AssetHarvester #AVSimulation #WorldModel #AutonomousDriving

@sudoingX For now, but there is better AI specific hardware in research. In memory compute, photonic computing and analog computers with artificial neurons in hardware will change the game, when they enter production in the next 5 years… Future local artificial brains FTW!

Cuda