Pixel-Aligned Multi-View Generation with Depth Guided Decoder

TL;DR

This paper introduces a novel pixel-aligned multi-view generation method that incorporates depth-guided attention in the VAE decoder, improving multi-view consistency and aiding 3D reconstruction from single images.

Contribution

It proposes a depth-truncated epipolar attention mechanism within a latent diffusion framework, enhancing pixel alignment across views and robustness to inaccurate depth during inference.

Findings

Improved pixel alignment across multi-view images.

Enhanced performance in multi-view to 3D reconstruction.

Effective handling of inaccurate depth during inference.

Abstract

The task of image-to-multi-view generation refers to generating novel views of an instance from a single image. Recent methods achieve this by extending text-to-image latent diffusion models to multi-view version, which contains an VAE image encoder and a U-Net diffusion model. Specifically, these generation methods usually fix VAE and finetune the U-Net only. However, the significant downscaling of the latent vectors computed from the input images and independent decoding leads to notable pixel-level misalignment across multiple views. To address this, we propose a novel method for pixel-level image-to-multi-view generation. Unlike prior work, we incorporate attention layers across multi-view images in the VAE decoder of a latent video diffusion model. Specifically, we introduce a depth-truncated epipolar attention, enabling the model to focus on spatially adjacent regions while remaining memory efficient. Applying depth-truncated attn is challenging during inference as the ground-truth depth is usually difficult to obtain and pre-trained depth estimation models is hard to provide accurate depth. Thus, to enhance the generalization to inaccurate depth when ground truth depth is missing, we perturb depth inputs during training. During inference, we employ a rapid multi-view to 3D reconstruction approach, NeuS, to obtain coarse depth for the depth-truncated epipolar attention. Our model enables better pixel alignment across multi-view images. Moreover, we demonstrate the efficacy of our approach in improving downstream multi-view to 3D reconstruction tasks.