RegionDrag: Fast Region-Based Image Editing with Diffusion Models (ECCV 2024)

Jingyi Lu†, Xinghui Li‡, Kai Han†
Visual AI Lab, The University of Hong Kong†; Active Vision Lab, University of Oxford‡

Overview

RegionDrag supports a variety of inputs. You can input regions or points to drag image contents from 🔴 RED to 🔵 BLUE. Below is a overview of different components in our UI. For detailed instructions to install RegionDrag, check out our README.

Tips

• Increasing the Handle Noise Scale can remove handle content. If it does not work, you can drag 🔴 some other contents to cover 🔵 the contents you would like to remove.
• The image displayed in the Results column is a preview obtained from your inputs before using Run Drag. A better preview generally implies a better editing result.
• If you find the preview image satisfactory, you can try changing the Method from Encode then CP to CP then Encode.

Input pairs of regions

• Step 1: Upload one image on the left, and click Fit Canvas to adjust size of image
• Step 2: Add Regions (Draw mask on the left, and then click Add Region)
• Step 3: Click Run Drag

Input pairs of polygons

• Step 1: Upload one image on the left, and click Fit Canvas to adjust size of image
• Step 2: Click points on the middle image (You can select to input triangles or quadrilaterals)
• Step 3: Click Run Drag

Input regions and manipulate them by points

• Step 1: Upload one image on the left, and click Fit Canvas to adjust size of image
• Step 2: Draw masks to represent regions on the left
• Step 3: Click points to control these regions in the middle
• Step 4: Click Run Drag