CVPR2023

HOLODIFFUSION: Training a 3D Diffusion Model Using 2D Images

Animesh Karnewar, Andrea Vedaldi, David Novotný, Niloy J. Mitra

摘要

Figure 1. We present HOLODIFFUSION as the first 3D-aware generative diffusion model that produces 3D-consistent images and is trained with only posed image supervision. Here we show a few different samples generated from models trained on different classes of the CO3D dataset [50].