Abstract: We present a diffusion-based model for 3D-aware generative novel view synthesis from as few as a single input image. Our model samples from the distribution of possible renderings consistent ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results