×

[/sddg/ - Stable Diffusion Development General, PyGUI edition]

[Features]

- fully functional tkinter-based UI with two alternative themes

- negative prompts, AKA unprompts

- wide range of supported hardware, Intel CPU of 8th gen and up is ideal

- supports both Linux and Windows

- supports model choice, SD or WD

[Installation & Usage]

1. Install the OpenVINO repo normally, as per the CPU tard guide (https://rentry.org/cputard).

2. Download the modified rpyth files and replace demo.py and stable_diffusion_engine.py with them. Get from https://www.litechan.org/liteshare/pygui-1-3-0.zip

3. Download RealESRGAN portable executable from https://github.com/xinntao/Real-ESRGAN and unzip it somewhere.

4. Activate your conda prompt or environment.

5. Run "pip install Pillow pyyaml sv-ttk".

6. Run "python pygui.py"

7. Go to Settings -> Configure, link the upscaler RealESRGAN executable, link the demo.py file from the openvino repo, link the python executable you'll be using for running demo.py ("C:\ProgramData\Miniconda3\python.exe" in my case, or you can go with just "python" in the env), hit Save.

8. Run "python pygui.py".

9. Open Queue -> Add Item and do the needful.

Enjoy! Feel free to report issues.

dickbutt

#70

nice kek

how to use waifu data?

#73 probably no way for now unless they compile a model for OpenVINO. There's another CPU-only fork somewhere. Do you need it?

I was able to follow the guide here rentry.org/cputard

but now I'm getting this output hastebin.com/uvibefiwev.swift

when running pygui.py

I'm using an i5 haswell cpu on Windows 7.

To the person who asked about using the wafu diffusion model, someone converted it to open vino here: https://huggingface.co/ShadowPower/waifu-diffusion.openvino

You might need to change some files to use it, but it uses all the same files as the normal model.

#75 try "pip3 install numpy==1.23.3", looks like you have an incompatible version. The one I specified is the one I have installed and everything runs just fine.

If that won't help, try another numpy version (they can be listed using "pip3 install numpy=="), but in the miniconda3 shell. There might be incompatibility warnings as well, pay attention to those.

#76 I might make a switch in the files if the model works at all, I'll be testing it.

[News]

New version released over at https://www.litechan.org/liteshare/pygui-1-2-4.zip with a loading bar! It's a minor change, but it makes proompting a bit more comfortable. And yes, I can update the OP as I wish, check it from time to time to see if an update has been released. The widget is very thin, so you might not notice it at first.

Where the fuck do I get data.yml

#77 hey not the same anon but I'm getting the same issue #75 is getting. I've tried installing different versions but nothing has thus worked.

#81

FIRST TIME SETUP

Go to Settings -> Configure in the GUI

-Hit [?] to open file browser and and link the RealESRGAN executable by hitting 'open'

-Link the your demo.py file from the openvino folder the same way

-Add the path to your Python executable, it should be C:\ProgramData\Miniconda3\python.exe

-Hit save

That generates the data.yml.

#82 What's your python? Is it the one in vin? If so, then

[The path might not be just C:\ProgramData\Miniconda3\python.exe]

To figure out the proper python path, enter the following:

python

import sys

print(sys.executable)

THAT is the python path. Before that all, run "conda activate vin".:

How do I prompt to get handsome Hitler AI images?

"C:/Users/********/Desktop/cpu ai/stable_diffusion.openvino/demo.py" --prompt "****** with a rose on her long black hair" --output "C:/Users/********/Desktop/***** ai.jpg" --num-inference-steps 42 --model bes-dev/stable-diffusion-v1-4-openvino --init-image C:/Users/********/Desktop/lagf/photo_2020-08-01_22-32-27.jpg

Exception in Tkinter callback

Traceback (most recent call last):

File "C:\Users\********\.conda\envs\vin\lib\tkinter\__init__.py", line 1892, in __call__

return self.func(*args)

File "C:\Users\********\Desktop\cpu ai\stable_diffusion.openvino\pygui.py", line 427, in view

self.image = self.image_pil.copy()#Image.open(self.data.upscaled)

AttributeError: 'str' object has no attribute 'copy'

Is this a normal output before rendering?:

[This post was removed by the Litechan administration]

nice try anon

> Is this a normal output before rendering?

I also have the same issue

>I also have the same issue

this

When I run this, this uses two of my four cpu cores. Is there a way to change the number of cores it uses?

#91 It uses an instruction set that can't be virtualised, therefore the limitation is the number of physical cores.

#91 #92 forcing all threads does seem to have a bit of an effect;

edit stable_diffusion_engine.py, add a new line under

self.core = Core()

and add this

self.core.set_property("CPU", {"INFERENCE_NUM_THREADS": 12})

replace 12 by your number of cores (physical+virtual)

on my 3600x i gain about 0.75s/it compared to the default 6 threads it uses.

Make sure the new line is indented the same way as the line above, it should be by 8 spaces.

#76 I am rather new to this program. Is there a way to select this particular model? Or do I just dump the files into the model directory and it will pull from whatever it can?

#100 There's an option for that in pygui now, whenever you use the script there's a model selector.

executing C:/Users/USER/Desktop/dot/stable_diffusion.openvino/realesrgan-ncnn-vulkan.exe -i "a-field-of-flowers.png" -o "a-field-of-flowersu.png" -n realesrgan-x4plus

[0 Intel(R) UHD Graphics 630] queueC=0[1] queueG=0[1] queueT=0[1]

[0 Intel(R) UHD Graphics 630] bugsbn1=0 bugbilz=0 bugcopc=0 bugihfa=0

[0 Intel(R) UHD Graphics 630] fp16-p/s/a=1/1/1 int8-p/s/a=1/1/1

[0 Intel(R) UHD Graphics 630] subgroup=32 basic=1 vote=1 ballot=1 shuffle=1

decode image a-field-of-flowers.png failed

Exception in Tkinter callback

Traceback (most recent call last):

File "C:\Users\USER\.conda\envs\vin\lib\tkinter\__init__.py", line 1892, in __call__

return self.func(*args)

File "C:\Users\USER\.conda\envs\vin\lib\tkinter\__init__.py", line 814, in callit

func(*args)

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\pygui.py", line 80, in manager

self.check_queue()

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\pygui.py", line 112, in check_queue

self.queue_items[ind].check()

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\pygui.py", line 413, in check

self.image = Image.open(self.data.upscaled)

File "C:\Users\USER\.conda\envs\vin\lib\site-packages\PIL\Image.py", line 3131, in open

fp = builtins.open(filename, "rb")

FileNotFoundError: [Errno 2] No such file or directory: 'a-field-of-flowersu.png'

I got that, where did i screwed up?

#106

Where's your image output? Did you try opening it before upscaling was complete?

#107

I tried to generate with and without an output folder/name, but I still get the same result.

Another thing to mention is this:

Traceback (most recent call last):

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\demo.py", line 5, in

from stable_diffusion_engine import StableDiffusionEngine

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\stable_diffusion_engine.py", line 4, in

from openvino.runtime import Core

ModuleNotFoundError: No module named 'openvino':

#108

Bruh install openvino. Activate your conda environment, then "pip install openvino".

How'd you miss that?

#108

Well, i did that but it installed Numpy?

(vin) C:\Users\USER\Desktop\dot\stable_diffusion.openvino>pip install openvino

Requirement already satisfied: openvino in c:\users\user\.conda\envs\vin\lib\site-packages (2022.1.0)

Collecting numpy<1.20,>=1.16.6

Using cached numpy-1.19.5-cp39-cp39-win_amd64.whl (13.3 MB)

Installing collected packages: numpy

Attempting uninstall: numpy

Found existing installation: numpy 1.23.4

Uninstalling numpy-1.23.4:

Successfully uninstalled numpy-1.23.4

Successfully installed numpy-1.19.5

And then when i tried again to generate something, this is the traceback:

Traceback (most recent call last):

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\demo.py", line 5, in

from stable_diffusion_engine import StableDiffusionEngine

File "C:\Users\USER\Desktop\dot\stable_diffusion.openvino\stable_diffusion_engine.py", line 2, in

import numpy as np

ModuleNotFoundError: No module named 'numpy'

The adventure just begun! I'm the same guy who forgot to install openvino, i tried writing this to see what could happen:

(vin) C:\Users\USER\Desktop\dot\stable_diffusion.openvino>python demo.py --prompt "Street-art painting of Emilia Clarke in style of Banksy, photorealism"

And it just started downloading random shit:

Downloading: 100%|█████████████████████████████████████████████████████████████████████| 905/905 [00:00<00:00, 990kB/s]

Downloading: 100%|███████████████████████████████████████████████████████████████████| 939k/939k [00:02<00:00, 394kB/s]

Downloading: 100%|███████████████████████████████████████████████████████████████████| 512k/512k [00:01<00:00, 417kB/s]

Downloading: 100%|█████████████████████████████████████████████████████████████████████| 389/389 [00:00<00:00, 398kB/s]

Downloading: 100%|█████████████████████████████████████████████████████████████████| 2.12M/2.12M [00:02<00:00, 815kB/s]

Downloading: 100%|███████████████████████████████████████████████████████████████████| 464k/464k [00:00<00:00, 783kB/s]

Downloading: 100%|██████████████████████████████████████████████████████████████████| 492M/492M [01:43<00:00, 4.74MB/s]

Downloading: 100%|████████████████████████████████████████████████████████████████| 3.02M/3.02M [00:01<00:00, 2.54MB/s]

Downloading: 21%|█████████████▊ | 733M/3.44G [02:36<08:21, 5.39MB/s]

Wth is that 3.44G thing?! Should i be scared?

#111 that's the model, man, you'd think something that competes with human artists would be 11Mb or something?

×

Mine's just stuck on initializing, is that normal?

#113 Got it working with #83's advice.

What are the steps to convert custom model.ckpt files to openvino optimized files? I want to add dreambooth models and any other SD models I find.

#117 ONNX is accepted, though there are prob better tutorials than this one.

https://docs.openvino.ai/latest/openvino_docs_MO_DG_prepare_model_convert_model_Convert_Model_From_PyTorch.html

Is there a way to have the UI display the seed of the generated image when you don't input it manually?

#169 not yet

×

Previews added! Seems like bes-dev abandoned the project entirely, we'd need to convert ONNX to OpenVINO to get model 2.1

If you have problems with script unable to find 'numpy' replace Python directory with

"C:\Users\\.conda\envs\vin\python.exe"

It worked. For me at least.

#245 and be sure type your username after C:\Users\

×

Woooo, litechan is back and so is the inspiration for the project! I'm working on the docs for picrel and PyGUI-GPU itself.

Exception in Tkinter callback

Traceback (most recent call last):

File "/usr/lib/python3.12/tkinter/__init__.py", line 1948, in __call__

return self.func(*args)

^^^^^^^^^^^^^^^^

File "/home/anon/Swap/PYGUIGPU/pyguigpu.py", line 235, in refresh_settings

s_dict = get_dict(self, params)

^^^^^^^^^^^^^^^^^^^^^^

File "/home/anon/Swap/PYGUIGPU/pyguidialog.py", line 58, in get_dict

return_dict[keys[n]] = smart_eval(inp.input[n])

^^^^^^^^^

AttributeError: 'Input' object has no attribute 'input'

Primary command: /home/anon/Swap/Stable-Diffusion-ONNX-FP16/sd_env/bin/python "/home/anon/Swap/Stable-Diffusion-ONNX-FP16/test-txt2img.py" --model /home/anon/Swap/stable-diffusion-2-1-base/tokenizer --prompt "bit titty goth gf" --negprompt "None" --steps 10 --size 448 --height 448 --cpuclip --cpuvae --seed 17568608147512182587 --output "/home/anon/Documents/bit-titty-goth-gf-tokenizer_seed17568608147512182587448x448.png"

Secondary command:

executing /home/anon/Swap/Stable-Diffusion-ONNX-FP16/sd_env/bin/python "/home/anon/Swap/Stable-Diffusion-ONNX-FP16/test-txt2img.py" --model /home/anon/Swap/stable-diffusion-2-1-base/tokenizer --prompt "bit titty goth gf" --negprompt "None" --steps 10 --size 448 --height 448 --cpuclip --cpuvae --seed 17568608147512182587 --output "/home/anon/Documents/bit-titty-goth-gf-tokenizer_seed17568608147512182587448x448.png"

usage: test-txt2img.py [-h] --model MODEL [--size SIZE] [--steps STEPS] [--scale SCALE] [--prompt PROMPT]

[--negprompt NEGPROMPT] [--seed SEED] [--cpu-textenc] [--cpuvae]

test-txt2img.py: error: unrecognized arguments: --height 448 --output /home/anon/Documents/bit-titty-goth-gf-tokenizer_seed17568608147512182587448x448.png

Exception in Tkinter callback

Traceback (most recent call last):

File "/usr/lib/python3.12/tkinter/__init__.py", line 1948, in __call__

return self.func(*args)

^^^^^^^^^^^^^^^^

File "/usr/lib/python3.:

#253 you'd have to use the modified script as the original has no --height param, also the model folder is one step higher, no need to specify one folder with a model, specify the folder where several models are located as folders, if that makes sense.

#254

ok, but now im getting a different error...

Primary command: /home/anon/Swap/Stable-Diffusion-ONNX-FP16/sd_env/bin/python "/home/anon/Swap/PYGUIGPU/based-txt2img.py" --model /home/anon/Swap/stable-diffusion-2-1-base/ --prompt "bit titty goth gf" --negprompt "None" --steps 10 --size 448 --height 448 --cpuclip --cpuvae --seed 16445083366228544265 --output "b"

Secondary command:

executing /home/anon/Swap/Stable-Diffusion-ONNX-FP16/sd_env/bin/python "/home/anon/Swap/PYGUIGPU/based-txt2img.py" --model /home/anon/Swap/stable-diffusion-2-1-base/ --prompt "bit titty goth gf" --negprompt "None" --steps 10 --size 448 --height 448 --cpuclip --cpuvae --seed 16445083366228544265 --output "b"

Traceback (most recent call last):

File "/home/anon/Swap/PYGUIGPU/based-txt2img.py", line 142, in

generator.seed(args.seed)

File "mtrand.pyx", line 4789, in numpy.random.mtrand.seed

File "mtrand.pyx", line 250, in numpy.random.mtrand.RandomState.seed

File "_mt19937.pyx", line 166, in numpy.random._mt19937.MT19937._legacy_seeding

File "_mt19937.pyx", line 180, in numpy.random._mt19937.MT19937._legacy_seeding

ValueError: Seed must be between 0 and 2**32 - 1

Exception in Tkinter callback

Traceback (most recent call last):

File "/usr/lib/python3.12/tkinter/__init__.py", line 1948, in __call__

return self.func(*args)

^^^^^^^^^^^^^^^^

File "/usr/lib/python3.12/tkinter/__init__.py", line 861, in callit

func(*args)

File "/home/anon/Swap/PYGUIGPU/pyguigpu.py", line 112, in manager

self.tabs[n](self.busy)

File "/home/anon/Swap/PYGUIGPU/pyguigpu.py", line 279, in __call__

self.image = Image.open(target).convert("RGB")

^^^^^^^^^^^^^^^^^^

File "/home/anon/.local/lib/python3.12/site-packages/PIL/Image.py", line 3227, in open

fp = builtins.open(filename, "rb")

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

FileNotFoundError: [Errno 2] No such file or directory: 'b':

#258

This is the issue

ValueError: Seed must be between 0 and 2**32 - 1

You must input a seed between 0 and 4294967295 (an unsigned 32 bits integer), the seed you're trying to input is far above that.

Is there a way to make the models at https://huggingface.co/models?pipeline_tag=text-to-image&library=openvino&search=helenai work with this?

#270 I know of #118 and I contacted the original dev who got OpenVINO to work with SD, but sadly he did not reply. Tried doing the conversion with 2.1 myself, but ran into issues using the models afterwards.

New script finally!

https://www.litechan.org/liteshare/pyguigpu81.zip

#249 :