Chuanxia Zheng

Training on Syn4D could unlock breakthroughs in dynamic scene understanding, where current datasets fall short in providing dense, complete, and accurate geometric annotations.

Forget representational differences - the secret to better feed-forward 3D scene modeling lies in tackling five core design problems.

Ditch the per-ray prediction bottleneck: NOVA3R's global scene representation and diffusion-based decoder unlock more complete and accurate 3D reconstructions from unposed images.

Text-to-video models can now generate videos that actually respect spatial relationships, thanks to a new geometry-based reward function that beats VLM-based alternatives.