Introduction

This is an inpainting model of the excellent Animagine XL by @CagliostroLab -- thank you for your work, mate!

The whole purpose of inpainting models is to provide at least somewhat better results when inpainting or outpainting an image. When e.g. outpainting an image, with the normal checkpoint models you tend to get a very visible seam between the original part of the image and the newly extended part -- this model helps eliminate that seam! Inpainting, on the other hand, can remove unwanted details or fix distorted details in an image, or you can even add entirely new details to it -- just change your prompt accordingly and draw a mask over the area, where you want the new detail to appear and off you go!

Here is an example of a rather visible seam after outpainting:

The original model on the left, the inpainting model on the right. (Yes, I cherrypicked one of the worst examples just to demonstrate the point)

And here is a less drastic example showing a similar effect, but now using inpainting to change a detail in the image instead of outpainting to expand the image's dimensions. Notice also how the inpainting model stuck better to the color tone of the rest of the image.

Disclaimer: you definitely can get good results even without, but it's typically easier with an inpainting model.

How to use:

See the original model for any recommendations for general settings and prompts, but when inpainting/outpainting, I tend to set CFG scale between 2-7 and denoising strength I usually begin with at 0.8, though I may bring it down 0.6, if I feel the model being a little too creative in the stuff it generates or doesn't stick well to the background color around the mask. May also need to bring it all the way up to 0.9, if you have a detail you really want to get rid of -- denoising strength has a large effect on the outcome and it may require a lot of playing around with, so if you're not happy with your results, keep everything else the same and just give this setting a go.

Other stuff:

Created in Automatic1111 by merging the base SDXL inpainting model (by wangqyqq), Juggernaut XL (the non-lightning model), SDXL base and the fp16 fix VAE.

描述:

Matches the AnimagineXL v3.1 model.

训练词语: