DragonDiffusion: Enabling Drag-style Manipulation on Diffusion Models

TL;DR

DragonDiffusion introduces a novel image editing technique that enables precise, drag-style manipulations on diffusion models without fine-tuning, allowing for diverse edits like object movement and resizing using internal feature guidance.

Contribution

It presents a new method for image editing on diffusion models using feature correspondence guidance, without requiring fine-tuning or extra modules.

Findings

Enables various editing modes such as object moving and resizing.

Does not require fine-tuning or additional modules.

Achieves content preservation and semantic alignment in edits.

Abstract

Despite the ability of existing large-scale text-to-image (T2I) models to generate high-quality images from detailed textual descriptions, they often lack the ability to precisely edit the generated or real images. In this paper, we propose a novel image editing method, DragonDiffusion, enabling Drag-style manipulation on Diffusion models. Specifically, we construct classifier guidance based on the strong correspondence of intermediate features in the diffusion model. It can transform the editing signals into gradients via feature correspondence loss to modify the intermediate representation of the diffusion model. Based on this guidance strategy, we also build a multi-scale guidance to consider both semantic and geometric alignment. Moreover, a cross-branch self-attention is added to maintain the consistency between the original image and the editing result. Our method, through an efficient design, achieves various editing modes for the generated or real images, such as object moving, object resizing, object appearance replacement, and content dragging. It is worth noting that all editing and content preservation signals come from the image itself, and the model does not require fine-tuning or additional modules. Our source code will be available at https://github.com/MC-E/DragonDiffusion.