Image To Image Tutorial In 13 Minutes – Stable Diffusion (Automatic1111)

Nice guide. A couple things I'd highlight:

Inpaint Area: Using "Whole picture" won't actually touch any part of the image outside of your mask. It will essentially use the context of the entire image to make changes to the masked region, whereas "Only masked" will only look at the masked region to generate the change. There's some behind the scenes upscaling and downscaling for the work A1111 is doing for "Only masked" but the final image won't have any changes outside the masked region.

I'm surprised you didn't talk about Denoising, as that's a pretty critical parameter for inpainting. The higher you set the denoising, the more the masked area will be adjusted. If you're totally changing the area, you'll pretty much want a 1 for denoising and you'd reduce it further and further the closer the existing masked area is to what you want.


Just some things I thought were worth mentioning, I enjoy your videos.

Was going to ask about denoising but it wasnt covered

If I want to fill an empty living room (without touching the position of the walls and the pattern of the floor), what if I want to arrange the furniture freely (in a brainstorming session)?

Thank you so much for the great in-depth tutorial! I've been using image2image for months and even I learned a couple of new things! Excellent tutorial!

Loved your explanation, very helpful thank you soo much!

thank you!

Bro this is all gold as per usual 👏