Single-Stage Diffusion NeRF: A Unified Approach to 3D Generation and Reconstruction
This work addresses the problem of fragmented approaches in 3D generation and reconstruction for computer vision researchers, offering a more integrated solution.
The paper tackles the challenge of creating a unified model for 3D-aware image synthesis, such as scene generation and novel view synthesis, by introducing SSDNeRF, which uses a single-stage diffusion model to learn a generalizable prior from multi-view images, achieving results comparable to or better than leading task-specific methods.
3D-aware image synthesis encompasses a variety of tasks, such as scene generation and novel view synthesis from images. Despite numerous task-specific methods, developing a comprehensive model remains challenging. In this paper, we present SSDNeRF, a unified approach that employs an expressive diffusion model to learn a generalizable prior of neural radiance fields (NeRF) from multi-view images of diverse objects. Previous studies have used two-stage approaches that rely on pretrained NeRFs as real data to train diffusion models. In contrast, we propose a new single-stage training paradigm with an end-to-end objective that jointly optimizes a NeRF auto-decoder and a latent diffusion model, enabling simultaneous 3D reconstruction and prior learning, even from sparsely available views. At test time, we can directly sample the diffusion prior for unconditional generation, or combine it with arbitrary observations of unseen objects for NeRF reconstruction. SSDNeRF demonstrates robust results comparable to or better than leading task-specific methods in unconditional generation and single/sparse-view 3D reconstruction.