INTERACTION-AWARE VIDEO OBJECT REMOVAL

Summary from the TLDR AI newsletter

Netflix's VOID is a video inpainting model that removes objects along with their physical and visual interactions, using interaction-aware mask conditioning built on CogVideoX.