Cora: Correspondence-aware image editing using few step diffusion

TL;DR

Cora is a novel diffusion-based image editing framework that uses correspondence-aware techniques to improve structural accuracy and texture transfer, enabling high-quality edits involving significant changes like pose, object, and texture modifications.

Contribution

Cora introduces a new editing approach that incorporates semantic correspondence and attention interpolation to enhance structural preservation and texture accuracy in diffusion-based image editing.

Findings

Cora outperforms existing methods in structure and texture preservation.

It effectively handles pose changes and object modifications.

User studies favor Cora's editing quality over alternatives.

Abstract

Image editing is an important task in computer graphics, vision, and VFX, with recent diffusion-based methods achieving fast and high-quality results. However, edits requiring significant structural changes, such as non-rigid deformations, object modifications, or content generation, remain challenging. Existing few step editing approaches produce artifacts such as irrelevant texture or struggle to preserve key attributes of the source image (e.g., pose). We introduce Cora, a novel editing framework that addresses these limitations by introducing correspondence-aware noise correction and interpolated attention maps. Our method aligns textures and structures between the source and target images through semantic correspondence, enabling accurate texture transfer while generating new content when necessary. Cora offers control over the balance between content generation and preservation. Extensive experiments demonstrate that, quantitatively and qualitatively, Cora excels in maintaining structure, textures, and identity across diverse edits, including pose changes, object addition, and texture refinements. User studies confirm that Cora delivers superior results, outperforming alternatives.