[Stable Diffusion] Prompt Sharing and Learning Thread

Kaseijin

Kaseijin

Active Member
Jul 9, 2022
611
1,062
HI!
I haven't do anything yet because my PC dies. In the meantime I get a new one, I have a question to you guys.
There are a page that "transform" images of HS2 into a near close real ones.
Take a look:
You don't have permission to view the spoiler content. Log in or register now.

My question is:
Is there a way to do the opposite? I mean, input a real portrait image and get an HS/AIS look-like image?
 
  • Like
Reactions: Mr-Fox and Sepheyer
Jimwalrus

Jimwalrus

Well-Known Member
Sep 15, 2021
1,053
4,036
Kaseijin said:
HI!
I haven't do anything yet because my PC dies. In the meantime I get a new one, I have a question to you guys.
There are a page that "transform" images of HS2 into a near close real ones.
Take a look:
You don't have permission to view the spoiler content. Log in or register now.

My question is:
Is there a way to do the opposite? I mean, input a real portrait image and get an HS/AIS look-like image?
Click to expand...
Sure, in fact if you look at the sample images on that page many of them are taking a real photo and converting it.
I don't actually know what "HS/AIS" means, but I'm assuming it's an art style similar to that of the first image above - in which case there should be quite a few Checkpoints you can use as the base.
Put the original into Img2Img, set a medium level of 'Denoising strength' and the right Checkpoint, add a prompt and away you go.
Once your PC is up and running you can give it a try - the experimentation's part of the fun of this.
 
  • Like
Reactions: Kaseijin, Mr-Fox and Sepheyer
J

jarman Kell

New Member
Oct 11, 2020
9
8
does anyone know where upscalers go in the stable diffusion folder? so there appear in the drop down menu for upscaler?
 
K

Kalpod42

New Member
Jul 28, 2020
5
1
I'm so damn confused by SD. I'm just getting started with it, and I've been able to make some pretty good images, or at least one's I'm reasonably happy with. That said, sometimes it seems to get "stuck" in a pose or idea, and no matter what I change in the prompt or what image I use in img2img, it can't let go.

For example. I'm trying to get an anal missionary view. I'm using this image as the start
You don't have permission to view the spoiler content. Log in or register now.

These are the sorts of images I'm getting.
You don't have permission to view the spoiler content. Log in or register now.
The first is straight up unholy, and while the second one isn't nearly as bad, its not what I'm trying to generate, which is very explicit in the prompt. Is there something I'm missing here?

FWIW, I've tried messing with the seed, though my understanding is that shouldn't matter as much in img2img?
 
J

jarman Kell

New Member
Oct 11, 2020
9
8
devilkkw said:
Not only there, if upscaler is ESRGRAN based goes there, but there are other format, like SWINR, and they go in SWINR folder.
Click to expand...
Is there a way to tell what they are based off of? Most of the one I download from civitai are .safetensor (probably spelled wrong):
 
Mr-Fox

Mr-Fox

Well-Known Member
Jan 24, 2020
1,401
3,804
Jimwalrus said:
Sure, in fact if you look at the sample images on that page many of them are taking a real photo and converting it.
I don't actually know what "HS/AIS" means, but I'm assuming it's an art style similar to that of the first image above - in which case there should be quite a few Checkpoints you can use as the base.
Put the original into Img2Img, set a medium level of 'Denoising strength' and the right Checkpoint, add a prompt and away you go.
Once your PC is up and running you can give it a try - the experimentation's part of the fun of this.
Click to expand...
HS (Honey select) AIS (AI Girl).
 
  • Like
Reactions: Kaseijin and Jimwalrus
Mr-Fox

Mr-Fox

Well-Known Member
Jan 24, 2020
1,401
3,804
jarman Kell said:
I did put them in that folder and restarted the webui. But they’re not appearing it’s kind of weird. I will have to do some more digging that’s again.
Click to expand...
Did you do a full restart or only a webui reload? You need to close the cmd window and restart webui-user.bat .
 
  • Like
Reactions: Jimwalrus
Sharinel

Sharinel

Active Member
Dec 23, 2018
611
2,568
Kalpod42 said:
I'm so damn confused by SD. I'm just getting started with it, and I've been able to make some pretty good images, or at least one's I'm reasonably happy with. That said, sometimes it seems to get "stuck" in a pose or idea, and no matter what I change in the prompt or what image I use in img2img, it can't let go.

The first is straight up unholy, and while the second one isn't nearly as bad, its not what I'm trying to generate, which is very explicit in the prompt. Is there something I'm missing here?

FWIW, I've tried messing with the seed, though my understanding is that shouldn't matter as much in img2img?
Click to expand...
From a very quick look at your png's it's absolutely positively definately (maybe) your denoising strength.

You have yours at 0.75 which is starting to get into "Fuck this prompt, imma do my own thing" territory, anything over 0.2 starts to melt the prompt.

Here's a plot I made of the original photo (I didn't download the Loras so used the jpg as the basis)
xyz_grid-0001-216977857.jpg

If you look at 0.5 things are already starting to go wonky, then at .75 it's nothing like the original. Try toning it down a bit, or doing your own x/y/z/ prompt to see how it would look if you have a pic you think you could use as a base.
And apologies if you are already aware of this but you find this in the script area, x/y/z is one of the dropdown option - set it up as so. Just try not to have 4 batches like I did as I'm a muppet :)

1683569608732.png
 
  • Like
Reactions: sharlotte, Jimwalrus, Kalpod42 and 2 others
Mr-Fox

Mr-Fox

Well-Known Member
Jan 24, 2020
1,401
3,804
Kalpod42 said:
I'm so damn confused by SD. I'm just getting started with it, and I've been able to make some pretty good images, or at least one's I'm reasonably happy with. That said, sometimes it seems to get "stuck" in a pose or idea, and no matter what I change in the prompt or what image I use in img2img, it can't let go.

For example. I'm trying to get an anal missionary view. I'm using this image as the start
You don't have permission to view the spoiler content. Log in or register now.

These are the sorts of images I'm getting.
You don't have permission to view the spoiler content. Log in or register now.
The first is straight up unholy, and while the second one isn't nearly as bad, its not what I'm trying to generate, which is very explicit in the prompt. Is there something I'm missing here?

FWIW, I've tried messing with the seed, though my understanding is that shouldn't matter as much in img2img?
Click to expand...
It will only get "stuck" if you are using the same seed over and over or if you are using a dominant embedding or hypernetwork or Lora etc. Did you try lowering the denoising strength? If you don't get the desired result despite the prompt, increase the cfg scale. Keep in mind that the quality of the source image can have an effect on the image quality of the generated image.

I reworked the prompt a little
00038-1583820884.png
 
  • Like
Reactions: sharlotte, Sepheyer, Jimwalrus and 3 others
Mr-Fox

Mr-Fox

Well-Known Member
Jan 24, 2020
1,401
3,804
jarman Kell said:
You must be registered to see the links



And yes I did a full restart closing cmd and Firefox. Then once back inside of stable I restarted it using the ui
Click to expand...
I would recommend going with a different upscaler from the
You must be registered to see the links
. I'm particularly fond of NMKD Superscale that you can find
You must be registered to see the links
, NMKD have many different upscalers with different purposes, I would also recommend getting NMKD Faces. NMKD is ESRGAN based.
 
  • Like
Reactions: CBTWizard, devilkkw and jarman Kell
devilkkw

devilkkw

Member
Mar 17, 2021
328
1,115
In search of some extension, i discovered
You must be registered to see the links
UI mod extension for a1111.
Tested and work like a charm.
Do you use some different UI? what's your favorite?
 
  • Like
Reactions: Mr-Fox
J

jarman Kell

New Member
Oct 11, 2020
9
8
You don't have permission to view the spoiler content. Log in or register now.

You don't have permission to view the spoiler content. Log in or register now.

Mr-Fox said:
I would recommend going with a different upscaler from the
You must be registered to see the links
. I'm particularly fond of NMKD Superscale that you can find
You must be registered to see the links
, NMKD have many different upscalers with different purposes, I would also recommend getting NMKD Faces. NMKD is ESRGAN based.
Click to expand...
thanks i will have to give that upscaler a try.

on a off note. ive started messing with everything v4 and cant seem to replicate the results of this photo dab5a1047d0f97dc0b4bae0c3f5d351c.png

even when i copy the image straight from the meta data mine comes out gray out like this. 00021-3184075657.png what could i be doing wrong besides not having the exact same upscaler, i wouldnt think that would matter. everything i have generated with v4 comes out looking grayed out like this. Thanks again Mr fox!
 
  • Hey there
Reactions: Mr-Fox
J

jarman Kell

New Member
Oct 11, 2020
9
8
jarman Kell said:
You don't have permission to view the spoiler content. Log in or register now.

You don't have permission to view the spoiler content. Log in or register now.


thanks i will have to give that upscaler a try.

on a off note. ive started messing with everything v4 and cant seem to replicate the results of this photo View attachment 2607141

even when i copy the image straight from the meta data mine comes out gray out like this. View attachment 2607142 what could i be doing wrong besides not having the exact same upscaler, i wouldnt think that would matter. everything i have generated with v4 comes out looking grayed out like this. Thanks again Mr fox!
Click to expand...

hey everyone i figured it out. i had to manually select the .vae file for the everything v4. thanks again 00022-2478039571.png
 
  • Like
  • Yay, update!
Reactions: VanMortis, devilkkw, Jimwalrus and 1 other person
Jimwalrus

Jimwalrus

Well-Known Member
Sep 15, 2021
1,053
4,036
jarman Kell said:
hey everyone i figured it out. i had to manually select the .vae file for the everything v4. thanks again View attachment 2607232
Click to expand...
I was about to suggest the issue was not having a VAE enabled.
Tip: I'd recommend to tick the box to always use a default VAE (84000 as recommended by Mr-Fox above seems the best general one) which it will then do unless the model comes with one already baked in
 
Last edited:
  • Like
Reactions: Mr-Fox
You must log in or register to reply here.
Top Bottom