This extenstion for AUTOMATIC1111/stable-diffusion-webui adds new value of "Masked content" field in img2img -> inpaint tab. Lama is a NN model useful for removing objects from pictures
This option means how to preprocess masked content before pass it into stable diffusion. It useful when you want to remove object in photo. Use inpainting model and denoising straight +-0.4
It also supports my other extension: sd-webui-replacer
Mask:
lama cleaner:
fill:
Others
original:
latent noise:
latent nothing:
Also you can use Lama cleaner in extras tab, if you want to use it without stable diffusion:
If you have installed this extensions, they appear as models here:
-
https://github.com/light-and-ray/sd-webui-resynthesizer-masked-content
-
https://github.com/light-and-ray/sd-webui-yandere-inpaint-masked-content
-
https://github.com/light-and-ray/sd-webui-manga-inpainting
-
+ MAT - another NN inpaint model
-
+ inpainting from OpenCV as bonus
-
"Insert background" inserts image into target image. Useful for massive compositing in batch
You can adjust few settings:
Go to Settings -> Extras Inpaint:
Native lama's dataset resolution is 256p, but it shows good result for highers with little quality of content reduction. 512p is optimal