SyncDreamer: Generating Multiview- consistent Images from a Single-view Image
Yuan Liu1,2 Cheng Lin2∗∗\ast∗ Zijiao Zeng2 Xiaoxiao Long1 Lingjie Liu3 Taku Komura1 Wenping Wang4††\dagger† 1The university of Hong Kong 2Tencent Games 3University of Pennsylvania 4Texas A、M University {yuanly, chlin, Equal contribution.Corresponding Authors.
Abstract
In this paper, we present a novel diffusion model called SyncDreamer that generates multiview-consistent images from a single-view image. Using pretrained large-scale 2D diffusion models, recent work Zero123 (Liu et al., 2023b) 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. Project page: https://liuyuan-pal.github.io/SyncDreamer/.
中文速览
从单张图片重建三维物体时，最棘手的难题是：用扩散模型分别生成各个视角的图像，各视角之间的几何形状和颜色往往对不上。SyncDreamer 的核心思路是把 N 个目标视角的图像当作一个整体来建模——在扩散模型（diffusion model）的每一步去噪过程中，通过一个具有三维空间感知能力的特征注意力机制（3D-aware feature attention），让所有视角的中间状态实时同步、相互参照，从而在一次反向扩散过程中同时生成几何和外观都高度一致的多视角图像。模型权重从在大规模数据集上微调过的 Zero123 初始化，因此对真实照片、手绘草图、水墨画等各类风格的输入都有良好泛化能力；生成的多视图直接喂给标准 NeRF 或 NeuS 即可重建三维形状，省去了耗时的蒸馏优化和文本反演。实验表明，SyncDreamer 在多视角一致性和三维重建质量上均优于基线方法，为单图转三维、文本转三维等任务提供了一条更快捷、更可控的新路径。
原文 arXiv:2309.03453;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2309.03453v2