🎨 LORE Image Editing

Official 🤗 Gradio demo
LORE: Latent Optimization for Precise Semantic Control in Rectified Flow-based Image Editing.
Here are editing steps:
1️⃣ Upload your source image.
2️⃣ Fill in your source prompt and use the brush tool to draw your mask. (on layer 1)
3️⃣ Fill in your target prompt, target object and its index in target prompt (index start from 0).
4️⃣ Adjust the hyperparameters.
5️⃣ Click the "Edit" button to generate your edited image!

🎨 [Examples] Click our examples below, draw your mask and click the "Edit" button.
🔔 [Note] Due to limited resources in spaces, this demo may only support optimization steps = 1.
🔔 [Note] Due to limited resources in spaces, you may need to resize large images <= 480.
If you need high resolution for better quality, go to https://github.com/oyly16/LORE for more usage with your own resource.

1 50
1 50
Upload an image
or
select the draw tool to start
0 30
1 10
1 10
0 1

https://github.com/oyly16/LORE

Examples (Click to load)
Brush Canvas Source Prompt Number of inversion/denoise steps Number of masked value injection steps Target Prompt Target Object Target Object Index (start index from 0 in target prompt) LORE optimization steps renoise scale Seed (-1 for random) Resize (only if input lager than this)(-1 for no resize)