SyncDreamer: Generating Multiview-consistent Images from a Single-view Image

TL;DR

SyncDreamer introduces a diffusion model that generates multiview-consistent images from a single view by synchronizing intermediate states across views, improving 3D consistency in generated images.

Contribution

It proposes a synchronized multiview diffusion model that models joint probability of views, ensuring consistency in geometry and colors during generation.

Findings

High view consistency in generated images

Effective for 3D tasks like novel-view synthesis and text-to-3D

Outperforms previous single-view generation methods

Abstract

In this paper, we present a novel diffusion model called that generates multiview-consistent images from a single-view image. Using pretrained large-scale 2D diffusion models, recent work Zero123 demonstrates the ability to generate plausible novel views from a single-view image of an object. However, maintaining consistency in geometry and colors for the generated images remains a challenge. To address this issue, we propose a synchronized multiview diffusion model that models the joint probability distribution of multiview images, enabling the generation of multiview-consistent images in a single reverse process. SyncDreamer synchronizes the intermediate states of all the generated images at every step of the reverse process through a 3D-aware feature attention mechanism that correlates the corresponding features across different views. Experiments show that SyncDreamer generates images with high consistency across different views, thus making it well-suited for various 3D generation tasks such as novel-view-synthesis, text-to-3D, and image-to-3D.