lllyasviel / lllyasviel/stable-diffusion-webui-forge

[img2img] Soft Inpainting + USDU + ADetailer

Open
#2,959 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
13k
Forks
1.7k
PR merge metrics
No merged PRs in 30d

Description

Hey,

I'm using an img2img payload to upscale/refine my pics.
I recently switched my bot from A1111 to Forge.
But now comes more issues...

Issue 1:
If using Ultimate SD Upscale + Adetailer, Adetailer works fine in every tile.
If using Ultimate SD Upscale + ADetailer + Soft Inpainting, the first tile will be processed fine, then replaced in a black image.
Then it will process every black tile after that. The final image is black except the first tile.
This was working fine in a1111 v<1.8 or so, broken for a while now. Very sad.

![Image](https://github.com/user-attachments/assets/a4500f17-0b8c-49bb-bfac-03d7dbdea6a7)

Issue 2:
Soft Inpainting just does nothing ?! My tiles seams are completly visible (9 tiles here). Works fine in a1111 :

![Image](https://github.com/user-attachments/assets/3d7cc40c-d3bf-41ce-ac00-529301a5a98e)

I already lowered my sharpening, color correction and denoise in the image above to make the seams less visible, but it's not even close from a solution.
Not sure if it's relevant to post my huge payload.

Thanks for any help.

Cheers ! 🥂

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing the reported img2img combinations using Ultimate SD Upscale, ADetailer, and Soft Inpainting, comparing Forge with the referenced A1111 behavior. Confirm that Soft Inpainting produces visible seams and that combining it with ADetailer does not turn later tiles black; the issue does not name files, tests, or a specific entry point.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.