Consolidating Attention Features for Multi-view Image Editing

Or Patashnik, Rinon Gal, Daniel Cohen-Or, Jun Yan Zhu, Fernando de la Torre

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

Large-scale text-to-image models enable a wide range of image editing techniques, using text prompts or even spatial controls. However, applying these editing methods to multi-view images depicting a single scene leads to 3D-inconsistent results. In this work, we focus on spatial control-based geometric manipulations and introduce a method to consolidate the editing process across various views. We build on two insights: (1) maintaining consistent features throughout the generative process helps attain consistency in multi-view editing, and (2) the queries in self-attention layers significantly influence the image structure. Hence, we propose to improve the geometric consistency of the edited images by enforcing the consistency of the queries. To do so, we introduce QNeRF, a neural radiance field trained on the internal query features of the edited images. Once trained, QNeRF can render 3D-consistent queries, which are then softly injected back into the self-attention layers during generation, greatly improving multi-view consistency. We refine the process through a progressive, iterative method that better consolidates queries across the diffusion timesteps. We compare our method to a range of existing techniques and demonstrate that it can achieve better multi-view consistency and higher fidelity to the input scene. These advantages allow us to train NeRFs with fewer visual artifacts, that are better aligned with the target geometry.

Original languageEnglish
Title of host publicationProceedings - SIGGRAPH Asia 2024 Conference Papers, SA 2024
EditorsStephen N. Spencer
PublisherAssociation for Computing Machinery, Inc
ISBN (Electronic)9798400711312
DOIs
StatePublished - 3 Dec 2024
Event2024 SIGGRAPH Asia 2024 Conference Papers, SA 2024 - Tokyo, Japan
Duration: 3 Dec 20246 Dec 2024

Publication series

NameProceedings - SIGGRAPH Asia 2024 Conference Papers, SA 2024

Conference

Conference2024 SIGGRAPH Asia 2024 Conference Papers, SA 2024
Country/TerritoryJapan
CityTokyo
Period3/12/246/12/24

Keywords

  • Diffusion Models
  • Image Editing
  • Multi-view

Fingerprint

Dive into the research topics of 'Consolidating Attention Features for Multi-view Image Editing'. Together they form a unique fingerprint.

Cite this