Step into the World from a Single Image
Jia-Bin Huang · 4:02
This video is a research paper presentation showing how to generate a consistent 3D-scene video (walking around a room) from a single input image, using a diffusion model whose cross-view attention is constrained by e...