When Text-to-Image Helps Editing: The Effects of Conditioning During Denoising

arXiv:2610.01681 · cs.CV · Submitted 2026-10-01 · Read on arXiv

cs.CV

Submitted: 2026-10-01

Updated: 2026-10-01

Comments: Under review as a conference paper at ICLR 2027

Code: https://github.com/pbnc5r6/conditioning-switching

License: http://creativecommons.org/licenses/by/4.0/

The gist: Unified models are trained for both instruction-based image editing and text-to-image (T2I) generation, but standard editing pipelines keep source-image conditioning throughout denoising.

Terminology

Abstract

Unified models are trained for both instruction-based image editing and text-to-image (T2I) generation, but standard editing pipelines keep source-image conditioning throughout denoising. We ask whether editing can benefit from T2I, and study how the effects of conditioning vary across edits and denoising stages. In pure editing, source attention declines for some edits over the sampling trajectory. This observation led us to task switching, which lets the model draw on its T2I capabilities. Across three unified editors and four benchmarks, switching to the T2I task for bounded intervals improves edit quality, while mean perceptual preservation remains close to pure editing across all three models. Unified editors therefore benefit from using both conditioning modes they are trained for, and the timing of the switch sets the balance between quality and preservation.

Sources

Related papers