EpiDiff: Enhancing Multi-View Synthesis via Localized Epipolar-Constrained Diffusion

TL;DR

EpiDiff introduces a localized diffusion model with epipolar constraints for rapid, high-quality multiview image synthesis from a single view, improving speed, diversity, and consistency.

Contribution

It proposes a lightweight epipolar attention block integrated into diffusion models, enhancing multiview synthesis without sacrificing speed or generalizability.

Findings

Generates 16 multiview images in 12 seconds

Outperforms previous methods in PSNR, SSIM, LPIPS

Produces more diverse and higher quality views

Abstract

Generating multiview images from a single view facilitates the rapid generation of a 3D mesh conditioned on a single image. Recent methods that introduce 3D global representation into diffusion models have shown the potential to generate consistent multiviews, but they have reduced generation speed and face challenges in maintaining generalizability and quality. To address this issue, we propose EpiDiff, a localized interactive multiview diffusion model. At the core of the proposed approach is to insert a lightweight epipolar attention block into the frozen diffusion model, leveraging epipolar constraints to enable cross-view interaction among feature maps of neighboring views. The newly initialized 3D modeling module preserves the original feature distribution of the diffusion model, exhibiting compatibility with a variety of base diffusion models. Experiments show that EpiDiff generates 16 multiview images in just 12 seconds, and it surpasses previous methods in quality evaluation metrics, including PSNR, SSIM and LPIPS. Additionally, EpiDiff can generate a more diverse distribution of views, improving the reconstruction quality from generated multiviews. Please see our project page at https://huanngzh.github.io/EpiDiff/.