Yes they all have seed numbers so you can pick one and use that seed with a updated prompt.
The signatures you see are all just gibberish. The images do not come from CC. They are actually created by an AI server (Model), that you have chosen. I have had to reduce image size a lot myself. Just comes with the territory. You will also find that if you use the same prompt and a seed from that prompt you will need to update the prompt text.

For instance you ask for a cat in a top hat with glasses. You run the model and get an image you kind of like but you now want it to have a bow. You change the prompt accordingly and add the image seed where it says seed (optional).
Then rerun the same model.

Hope that explains most of what you asked.

Just to show you an example.

PROMPT- A sheep playing guitar and singing in a meadow, in the style of Bob Dylan

MODEL- ICBINP

Asked for 5 images.
The one seed i liked.



NEW PROMPT with seed from that image.
A sheep wearing a top hat playing guitar and singing in a meadow, in the style of Bob Dylan.

One image
Results



You will notice both images have the same seed. I could further add text to the prompt and get more items i would like in the image but i will not. I hope you get the idea of it all. It's just fun, and addicting.

Here is the result from the updated prompt without the seed added.
The newly added top hat.

Notice that the guitar has 7 adjustment knobs for the strings...lol
So AI does mess up. You just really need to know exactly what to ask.

playing a Gibson guitar perhaps would have given better results.

The final image i kept.

Addicting indeed and I cant believe I have fallen for this. I spent most of last night trying to get things. Haven't started today yet but probably will soon.

Decided I can do a few backgrounds with this, which are free for my photo manipulating where people ask for change of certain types of backgrounds (a few people ask for are fantasy, Christmas & beach/sunset.) - of course I will have to upscale them.
I did a couple of fantasy castle and out of 10 they all had blue roofs like the disney castle. I did ask for a gold one and got them with green roofs

Looks like you are having fun with it.

I just downloaded and installed the latest release for Gimp-3.0 here.
https://github.com/ikks/gimp-stable-diffusion/releases/

Testing now.

Let us know how it goes!
I just used the BP Model and out of 5 like these best
prompt - tirade of colourful peasantries (I meant to put pleasantries )

Sometimes mistakes lead to the best results, Sallyanne. lol

Really cool stuff, but again, I'm trying to detox from AI so stop tempting me. lol

I agree with Lyle. That image is pretty cool Sallyanne. Very Picasoish.

Oh come on Lyle you know you wanna, just once. LOL

I still have not been able to get this to work on Gimp-3.0 though. Sadly.

My only issue is the registration process; prefer not to do so (but not saying I won't, Rod).

Hi, we discovered that there is a problem with Gimp3 series in Windows that will be fixed in 3.1.4 and upwards.

In the meantime if you are on windows and have Gimp3.0.4 please go to the branch winsslfix in the directory stablehorde-gimp3 and grab gimp-stable-diffusion.py , download under plug-ins/gimp-stable-diffusion making sure you are in .config/GIMP/3.0/ .

Please look at the README.md

I'm not allowed to paste urls, sorry.

If anyone is in Mac, and uses the plugin for Gimp3, feel free to post a screenshot or submit a issue to the repo. On Linux there has been no issues reported so far.

Hi ikks, i don't see a branch winsslfix on your latest release website.
https://github.com/ikks/gimp-stable-diffusion/releases/

Hi ikks, i am pasting the direct link to the winsslfix folder for you here. You should be able to post links after 5 posts. Thanks! Testing now.

https://github.com/ikks/gimp-stable-dif ... orde-gimp3

Works great!
Is there a way to get more than 1 image at a time?

Gimp-Stable-Diffusion for Gimp-3.0.4 Revision 1

Promp- a cat in a red hat with a green bowtie
Model - majicmix realistic

Results (1 image)



Thanks ikks!

One thing i like is you do not have to open a new image before use. You just run the plugin and it creates a new image.
In 2.10 Stablehorde you have to create a new image before the plugin will run.