While a 3D map would provide more information, a 2D map is much easier to process. "It's very efficient and very compact because you're just using a two-dimensional histogram," Ramani said.

Step 2: Generate a synthetic 3D motion dataset by optimization using the learned line-codnitioned 2D diffusion model cd aist_scripts/ sh step2_gen_syn3d_data.sh Step 3: Train a multi-view 2D motion ...