[Stable Diffusion] Prompt Sharing and Learning Thread

[Mr-Fox]

Hey. I'm having this problem when inpainting, there is some weird shadows out of nowhere on boobs or on the entire body.

Guys is there a way to make pretty similiar celebrity faces?

Guys! how about you pretend to explain your issue to a really stupid friend and then write this down without missing any detail, then rewrite it again with a bit less condescension and it should be perfect..

 

[Mr-Fox]

Guys is there a way to make pretty similiar celebrity faces?

Pretty similar to what? You can make similar faces of a celebrity by using a Lora that has been trained with images of this celebrity.


What is a Stable Diffusion LoRA?

LoRAs (Low Rate Adaptation) are files that alter the results of existing Stable Diffusion models based a particular concept. This could be an art-style, a character, a real-life person, an object, or a theme.

 

[Halmes]

What is it you are trying to fix with inpaint?

im trying to deepfake a photo, take clothes off.

 

[Jimwalrus]

Guys is there a way to make pretty similiar celebrity faces?

You need to be more specific - do you already have (or have found on Civitai) a Textual Inversion / LoRA of the celeb and want to make SD produce images that look so much like them their lawyers' palms get sweaty?

Spoiler: For example

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

Or do you have a celeb in mind and want to create a TI/LoRA of them (in order to produce yada-yada-yada...)?

 

[Mr-Fox]

im trying to deepfake a photo, take clothes off.

Go to civitai and see if there is a Lora for this person. This will make it much easier. Or if you have the hardware (6Gb VRam) and the patience, train a Lora of of this person yourself. Pix2Pix is probably a better solution than inpaint.
Pix2Pix is used to change one character into something else. Meaning you can use an image of a woman and put in the prompt to change this woman into a knight with armor etc. I have not messed with pix2pix but I imagine that it would work better to turn a clothed person into naked much easier than img2img and/or inpaint.

You must be registered to see the links

You must be registered to see the links

 

[modine2021]

ugh. all of sudden SD decides it wants to act retarded. so uninstall and reinstall everything but still cant past startup message:


The system cannot find the path specified.
Cancel
venv "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\venv\Scripts\Python.exe"
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Commit hash: <none>
Traceback (most recent call last):
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 355, in <module>
prepare_environment()
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 260, in prepare_environment
run_python("import torch; assert torch.cuda.is_available(), 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'")
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 121, in run_python
return run(f'"{python}" -c "{code}"', desc, errdesc)
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 97, in run
raise RuntimeError(message)
RuntimeError: Error running command.
Command: "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\venv\Scripts\python.exe" -c "import torch; assert torch.cuda.is_available(), 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'"
Error code: 1
stdout: <empty>
stderr: The system cannot find the path specified.

 

[Mr-Fox]

ugh. all of sudden SD decides it wants to act retarded. so uninstall and reinstall everything but still cant past startup message:


The system cannot find the path specified.
Cancel
venv "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\venv\Scripts\Python.exe"
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Commit hash: <none>
Traceback (most recent call last):
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 355, in <module>
prepare_environment()
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 260, in prepare_environment
run_python("import torch; assert torch.cuda.is_available(), 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'")
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 121, in run_python
return run(f'"{python}" -c "{code}"', desc, errdesc)
File "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\launch.py", line 97, in run
raise RuntimeError(message)
RuntimeError: Error running command.
Command: "V:\AI\A1111 Web UI Autoinstaller\stable-diffusion-webui\venv\Scripts\python.exe" -c "import torch; assert torch.cuda.is_available(), 'Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check'"
Error code: 1
stdout: <empty>
stderr: The system cannot find the path specified.

I suggest to follow the installation instructions here:

You must be registered to see the links

Re install both python and git and then do the git clone.
Then type cmd in the root folder and run git pull to update.
If this doesn't work maybe it's time to open an issue post on github. Go to "issues" and press "new issue" .

 

[modine2021]

I suggest to follow the installation instructions here:

You must be registered to see the links

Re install both python and git and then do the git clone.
Then type cmd in the root folder and run git pull to update.
If this doesn't work maybe it's time to open an issue post on github. Go to "issues" and press "new issue" .

yeah. thats what i did when i first installed it the first time. don't know why now it decide to act up. was using it perfectly fine til a few hours ago. i think its that torch thingy. that it cant find(even though its installed)

 

[modine2021]

@ Mr-Fox

found the issue myself. seems miniconda was on autorun. so SD was pointing to it instead of python.exe so had to delete it in the regostry. so if anyone else get this issue do this:

1. Open the Registry Editor (press windows key, type regedit and hit Enter) .
2. Navigate to HKEY_CURRENT_USER\Software\Microsoft\Command Processor\AutoRun and clear the values.
3.check HKEY_LOCAL_MACHINE\Software\Microsoft\Command Processor\AutoRun too if its there

 

[miaouxtoo]

@ Mr-Fox

found the issue myself. seems miniconda was on autorun. so SD was pointing to it instead of python.exe so had to delete it in the regostry. so if anyone else get this issue do this:

1. Open the Registry Editor (press windows key, type regedit and hit Enter) .
2. Navigate to HKEY_CURRENT_USER\Software\Microsoft\Command Processor\AutoRun and clear the values.
3.check HKEY_LOCAL_MACHINE\Software\Microsoft\Command Processor\AutoRun too if its there

Nice, glad you fixed it!
You could potentially also just paste the error straight into chatgpt to see if it'll suggest a correction. I know gpt4 is really good at correcting code issues.

 

[Halmes]

Go to civitai and see if there is a Lora for this person. This will make it much easier. Or if you have the hardware (6Gb VRam) and the patience, train a Lora of of this person yourself. Pix2Pix is probably a better solution than inpaint.
Pix2Pix is used to change one character into something else. Meaning you can use an image of a woman and put in the prompt to change this woman into a knight with armor etc. I have not messed with pix2pix but I imagine that it would work better to turn a clothed person into naked much easier than img2img and/or inpaint.

You must be registered to see the links

You must be registered to see the links

I have 8gb vram.

 

[Mr-Fox]

I have 8gb vram.

Like I said, go to civitai first and see if there is a Lora of the person you want to make images of. Try out pix2pix. If those things did not work, then get into making your own lora.

You must be registered to see the links

You must be registered to see the links

You must be registered to see the links

You must be registered to see the links

 

[modine2021]

how to get more than oneperson together in image? posing together i mean like 3 people. it happen by accident in some pics i did before . but i want to purposely do it (multiple)? 3girl? (womana womanb womanc standing together)?

 

[Deleted member 5469217]

You must be registered to see the links

 

[Jimwalrus]

how to get mroe than oneperson together in image? posing together i mean like 3 people. it happpend by accident in some pics i did before . but i wnat to purposly do it (multiple)? 3girl? (woman a womanb womanc standing together)?

It's not easy, but "3girls close together" usually* works.

*"Usually" of course meaning that sometimes you'll get one, two or four women, and sometimes you'll get six women, a kite, a wheelbarrow and a fucking Cocker Spaniel...

 

[Sepheyer]

how to get more than oneperson together in image? posing together i mean like 3 people. it happen by accident in some pics i did before . but i want to purposely do it (multiple)? 3girl? (womana womanb womanc standing together)?

Use the specialized extension:

You must be registered to see the links

I don't have a prompt for this picture, it merely shows the capabilities of the extension:

 

[Sharinel]

I use either Latent Couple or Regional Prompter to get multiple people in an image, works well

 

[Sharinel]

I went trawling through Civ trying to get inspiration for some prompts, came across this one (all checkpoints are HyperV1)

Prompt :- official art, unity 8k wallpaper, ultra detailed, beautiful and aesthetic, masterpiece, best quality, icelandic blonde girl, extremely detailed, dynamic angle, cowboyshot, the most beautiful form of chaos, elegant, a brutalist designed, vivid colours, romanticism, by james jean, roby dwi antono, ross tran, francis bacon, michal mraz, adrian ghenie, petra cortright, gerhard richter, takato yamamoto, ashley wood, atmospheric, ecstasy of musical notes, streaming musical notes visible

Gets me this :-



Not really what I was looking for. So I amended the prompt to take out all the 'by so and so'

Prompt :- official art, unity 8k wallpaper, ultra detailed, beautiful and aesthetic, masterpiece, best quality, icelandic blonde girl, extremely detailed, dynamic angle, cowboyshot, the most beautiful form of chaos, elegant, a brutalist designed, vivid colours, romanticism, atmospheric, ecstasy of musical notes, streaming musical notes visible

Gets me this :-



or from the same batch :-




The takeaway? Using the prompt to change to the style of someone makes a helluva difference. I might try 'by Larry Flynt' in some batches

 

[modine2021]

It's not easy, but "3girls close together" usually* works.

*"Usually" of course meaning that sometimes you'll get one, two or four women, and sometimes you'll get six women, a kite, a wheelbarrow and a fucking Cocker Spaniel...

manage to get it. unfortunately ended up basically with same faces:

(change number here, 3 to whatever)-->>(3girls, multiple girls:1.3) <<--(change number at end, 3 to whatever) a photo of (rest of prompt here)

 

[VanMortis]

I went trawling through Civ trying to get inspiration for some prompts, came across this one (all checkpoints are HyperV1)

Prompt :- official art, unity 8k wallpaper, ultra detailed, beautiful and aesthetic, masterpiece, best quality, icelandic blonde girl, extremely detailed, dynamic angle, cowboyshot, the most beautiful form of chaos, elegant, a brutalist designed, vivid colours, romanticism, by james jean, roby dwi antono, ross tran, francis bacon, michal mraz, adrian ghenie, petra cortright, gerhard richter, takato yamamoto, ashley wood, atmospheric, ecstasy of musical notes, streaming musical notes visible

Gets me this :-

View attachment 2556628

Not really what I was looking for. So I amended the prompt to take out all the 'by so and so'

Prompt :- official art, unity 8k wallpaper, ultra detailed, beautiful and aesthetic, masterpiece, best quality, icelandic blonde girl, extremely detailed, dynamic angle, cowboyshot, the most beautiful form of chaos, elegant, a brutalist designed, vivid colours, romanticism, atmospheric, ecstasy of musical notes, streaming musical notes visible

Gets me this :-

View attachment 2556639

or from the same batch :-

View attachment 2556640


The takeaway? Using the prompt to change to the style of someone makes a helluva difference. I might try 'by Larry Flynt' in some batches

Sometimes just one word can change alot in an image, also the promt order. The "by Larry Flynt" got me thinking about a "funny" thing. Have you ever tried to ad your nick to the prompt. Try it, it can be hilarious. Apparently "vanMortis" can give you quite an image.



"vanMortis" also works as a negative prompt to get some filth out of your image.