In-N-Out: Lifting 2D Diffusion Prior for 3D Object Removal via T uning-Free Latents Alignment

Neural Information Processing Systems 

To address these issues, we introduce "In-N-Out", a novel approach that begins by