EditBridge: A Diffusion Bridge Framework for Faithful and Efficient Ultra-High-Resolution Image Editing

Forum topic · 小凯 · 2026-08-20

Summary

EditBridge is a diffusion bridge framework for ultra-high-resolution image editing, addressing the limitations of diffusion models that are typically confined to sub-1K resolutions due to quadratic attention complexity and memory demands. Common two-stage pipelines that edit at low resolution and then apply super-resolution suffer from information divergence (hallucinated details contradicting the original high-resolution source) and texture degradation (over-smoothing or over-sharpening artifacts). EditBridge reformulates refinement as a structured data-to-data transformation from a low-resolution edited result to its high-resolution counterpart, explicitly conditioned on the original HR source to preserve authentic details. A prior-guided blocked sparse attention mechanism leverages semantic correspondence from the first-stage edit, restricting cross-image interactions to spatially aligned regions and cutting computational overhead. Experiments show high-fidelity editing at up to 4K resolution, 3.6-8.4x speedups at 2K, and practical 4K editing within 61 seconds. Paper: arXiv 2608.18063.

EditBridge: Towards Faithful and Efficient Ultra-High-Resolution Image Editing

Field: Computer Vision
Authors: Jiayi Song, Shijie Huang, Fangtai Wu, Yubo Huang, Zhenxiong Tan, Songhua Liu, Jiaming Liu, Ruihua Huang
arXiv: 2608.18063

Overview

High-resolution image editing is increasingly demanded in professional workflows, yet existing diffusion-based models remain constrained to resolutions below 1K due to quadratic attention complexity and prohibitive memory requirements. A prevalent workaround employs a two-stage pipeline: editing at low resolution followed by independent super-resolution. However, this approach suffers from two critical issues:

Information divergence: hallucinated details contradict the original high-resolution (HR) source.
Texture degradation: over-smoothed or over-sharpened artifacts.

Method

EditBridge is a diffusion bridge framework for efficient ultra-high-resolution editing. Unlike conventional diffusion that regenerates from noise, the refinement is formulated as a structured data-to-data transformation from the low-resolution (LR) edited result to its HR counterpart, explicitly conditioned on the original HR source to preserve authentic details.

To integrate HR source guidance efficiently, the authors introduce a prior-guided blocked sparse attention mechanism. It exploits semantic correspondence from the first-stage edit to restrict cross-image interactions to spatially aligned regions, significantly reducing computational overhead.

Results

High-fidelity editing with superior perceptual quality at up to 4K resolution.
3.6-8.4x speedup at 2K resolution.
Practical 4K editing within 61 seconds.