Controlling feature injection improves image editing trade offs
Inject Where It Matters: Objective-Dependent Feature Control in Rectified-Flow Image Editing
Artificial Intelligence
Summary
Image editing tools often struggle to balance keeping the original image details while making desired changes. The authors introduce a way to carefully choose where and how features from the original image are applied during editing. They use adaptive masks and a selection method to pick parts of the processing pipeline that best match editing goals. This approach helps keep the image looking right while still making effective edits. The results show better overall image quality and flexible control over preserving or changing details.
What this means in practice
- •For graphic design software developers: Implement targeted feature control in image editing tools to improve balance between preserving original images and applying desired changes.$Commercial implications: Enables software products that offer users better control in automated image editing, differentiating them with advanced preservation-edit trade-offs.
- •For digital art platform engineers: Optimize machine learning pipelines to select key processing blocks automatically for improved image editing quality without runtime search.
Authors
Nina Goncharova, Alexandra Ivanova, Aibek Alanov
Abstract
Many prompt-driven FLUX editing methods rely on injecting source features into target-conditioned generation, creating an inherent trade-off between source preservation and the desired edit. We address this problem by controlling where source features are injected and which transformer blocks are used. First, we introduce adaptive masks that spatially gate feature injection using differences between source- and target-conditioned block outputs at matched inversion states. Second, we use REINFORCE to select Top-K injection blocks under different preservation-editability objectives. The selected blocks are fixed at inference and require no online search. On PIE-Bench with FLUX.1-dev, spatial masking controls the alignment-preservation trade-off. On 60 validation examples, a learned nine-block set gives lower whole-image LPIPS than the 18-block RF-Edit set within the same masked pipeline, with slightly lower alignment. Among 50 random nine-block sets, it has the lowest LPIPS but not the highest reward. Separate objective sweeps reveal different block preferences for preservation and alignment.