Search papers, labs, and topics across Lattice.
1
0
3
Current egocentric video generation models fail to execute multi-step manipulation tasks realistically, often skipping crucial actions and exhibiting poor physical dynamics.
