Inpainting doesn't work with A1111

[ZJDavis]

I'm not able to get inpainting to work correctly. Is there a guide for configuration that can be added to the readme or is it possible the feature is broken?

Using a normal checkpoint: it generates an entirely separate image with the selection area becoming a mask using an inpainting checkpoint: it generates flat gray, every time.

I expect it's user error but I can't find any information on this for the plugin.

[nextlevelbros]

I too am unable to get inpainting or outpainting to work. I have tried all sorts of settings and read everything written about the topic in the old discord and here in the issues, but nothing has helped to get me and closer to being able to inpaint and outpaint with this plugin.

I have recorded a video that shows exactly what I see on my end when I try to inpaint and outpaint. Here is my screenrecording, I hope someone who has the knowledge to shine a light on this will take a look and advise what is going on and how to get it working: https://videoshare.nextlevelbros.com/conversations/17c2c7d0-777b-542b-9528-8f93f91c5ab2

The gist of it is that I never get anything close to usable, so the whole endeavour is just not working for me. Like many others I am confused as to how the masking is supposed to work, about various settings and of course about the fact that I cannot get anything usable out of this plugin.

My stable diffusion Automatic 1111 is totally up to date, as are all my extensions. Everything works flawlessly when using the Automatic 1111 web ui. I have installed this plugin into photoshop and also the appropriate extension into Automatic 1111. My Photoshop is also totally up to date.

[kamerosoul]

I have been using this plugin recently and it is very good, many improvements have been implemented: Here you have some my settings:

• use latest plugin version V1.41. Photoshop 2023
• I used a 2000x3000 px picture 72ppi
• My selection is about half of image size and covers total person body and head.
• my prompt is very simple, positive and negative
• mask is about 15% of selection
• sampling =100, CFG scale = 7, Mask blur = 40, mask expansion = 50, mask content = original
• Denoise strengh = 0.7 It is very critical. Too much will result in deformed, low value will result in little change
• Use mask content original if picture has requested feature and just need to modify.
• In controlnet use openpose, control_sd15_openpose
• Set CtrlImg and preview annotator. Check preview is acceptable.
• Model: Ciberrealistic V40. Some models does not work fine for inpainting. Try different models Hope it helps