EpiDiff: Enhancing Multi-View Synthesis via Localized Epipolar-Constrained Diffusion
This addresses the challenge of slow and inconsistent multiview synthesis for 3D reconstruction applications, though it appears incremental as it builds on existing diffusion models with localized modifications.
The paper tackles the problem of generating consistent multiview images from a single view for 3D mesh reconstruction, proposing EpiDiff which uses a lightweight epipolar attention block to enhance cross-view interaction, resulting in generation of 16 multiview images in 12 seconds and surpassing previous methods in quality metrics like PSNR, SSIM, and LPIPS.
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/.