Saved in:
Bibliographic Details
Main Authors: Yang, Xianghui, Zuo, Yan, Ramasinghe, Sameera, Bazzani, Loris, Avraham, Gil, Hengel, Anton van den
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.18842
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Novel-view synthesis through diffusion models has demonstrated remarkable potential for generating diverse and high-quality images. Yet, the independent process of image generation in these prevailing methods leads to challenges in maintaining multiple-view consistency. To address this, we introduce ViewFusion, a novel, training-free algorithm that can be seamlessly integrated into existing pre-trained diffusion models. Our approach adopts an auto-regressive method that implicitly leverages previously generated views as context for the next view generation, ensuring robust multi-view consistency during the novel-view generation process. Through a diffusion process that fuses known-view information via interpolated denoising, our framework successfully extends single-view conditioned models to work in multiple-view conditional settings without any additional fine-tuning. Extensive experimental results demonstrate the effectiveness of ViewFusion in generating consistent and detailed novel views.