Consistent View Synthesis with Pose-Guided Diffusion Models

https://poseguided-diffusion.github.io/ Outputs Input Figure 1. Consistent view synthesis via pose-guided diffusion model. (top) Given an input image and a sequence of camera poses, we present a pose-guided diffusion model to generate a sequence of frames that are photorealistic and 3D consistent. (bottom) Our proposed method can synthesize diverse sequences from the same set of inputs.