Stable Diffusion Models

Already have some ideas on how to set this up to be user friendly, but it may need a few more updates.

GRisk

I don't have a tutorial for now, but the alpha is the places in your image that the GUI will paint new stuff

GRisk

Thanks

alpha channel area of the image is what will be "painted" while existing rgb pixels will remain unchanged. There's no paint editor like dalle2 or others. Instead Open an image in photoshop, gimp/ etc, erase parts of it leaving an alpha channel, save the png, use it as your inpainting image.

mgfxer

Hi all is there a tutorial for the inpainting don’t get what alpha has to do with it

can you include text help file if you are going to separate it? Thanks. I don’t mind one big file.

Rai T00n

In the next update, to select other models. (Waifu diffusion)

GRisk

Yep, it will appear on the next update

GRisk

My goal is to set all information on the main GUI and have the CMD with just a copy of informations, but speed is the key for now, so I'm focusing in one thing at the time.

GRisk

Yeah, but there is some limitations. I will create a post about this later on.

GRisk

it would be good to somehow edit img2img masks etc+. that one would write something like #keep style from image,face of woman/ prompt woman in forrest in fantasy cloath

download accelerator work i use it,because same problem

Would it be possible to train Stable Diffusion with my own images so that my own style is adopted? So new images should be created in the style of my own art. Is it possible to build in a function?

You can use jdownloader2 (jdownloader.org) or another free download manager to resume downloads from google drive and other hosters.

Exactly my thoughts too. I came here because of simplicity of installation, that's key 1. (BecauseI'm not that tech savy either).

Luis Salas Lastra

I did have problems getting it downloaded as it is, failed once about ten minutes from the end, super frustrating. Even with good internet, personally I'm in a house with sometimes six people hammering the connection at once, so 3gb can still be pretty hefty in 2022 with no resuming. Whatever you can do to keep things manageable is greatly appreciated.

I don’t mind if the models are in a separate zip file, but I’m pretty computer illiterate, so instructions on how to add them to the gui would be great. Unless it’s truly as simple as just unzipping and adding them to the folder??

This would be a great idea !

And whats the difference between this models?

Hi i have read somewhere its not allowed to pack models with the sd file. The user must register on their page to download the models. Not that you get licence problems.

Hi, in my case I think that can be good have a better user experience, auto update pressing a button, progress bar without need another windows with the terminal, save seeds as shortcuts, the select folder for the photos turn to default every time that you open the app... In my case I choose this version because is way easier, just double click and is running, so for me is have a nice GUI experience is a good deal, after that the facial restoration or more functions are welcome 😁

Gastón Darío de la Rosa Bandi

Hi, a little unrelated but I’m hoping you can add facial restoration like you did the upscaler. NMKD’s stable diffusion GUI has it and it honestly makes a huge difference.

Is this what the "image model" dropdown section is going to be used for?