hnmr293 / sd-webui-llul Goto Github PK
View Code? Open in Web Editor NEWLLuL - Local Latent upscaLer
License: Other
LLuL - Local Latent upscaLer
License: Other
Show error on M2 pro, pytorch, I use --upcast-sampling --opt-sub-quad-attention --use-cpu interrogate
and torch==2.0.0.dev20230131
Error completing request████████████████████▋ | 4/12 [00:01<00:03, 2.29it/s]
Arguments: ('task(cmmoguzralnnpel)', '', '', ['OL'], 12, 2, False, False, 1, 1, 7, -1.0, -1.0, 0, 0, 0, False, 512, 512, False, 0.45, 2, 'R-ESRGAN 4x+', 8, 0, 0, [], 0, False, 'none', 'None', 1, None, False, 'Scale to Fit (Inner Fit)', False, False, 64, 64, 64, 1, False, False, 'none', 'None', 1, None, False, 'Scale to Fit (Inner Fit)', False, False, 64, 64, 64, 1, False, True, 1, 0.15, False, 'OUT', ['OUT'], 5, 0, 'Bilinear', False, 'Pooling Max', False, 'Lerp', '123.7265625', '246.2578125', False, False, 'positive', 'comma', 0, False, False, '', 1, '', 0, '', 0, '', True, False, False, False, 0) {}
...
File "~/Projects/stable-diffusion-webui/extensions/sd-webui-llul/scripts/llul_hooker.py", line 82, in lerp
return torch.lerp(v0, v1, t)
RuntimeError: "lerp_kernel_scalar" not implemented for 'Half'
I am getting these errors in latest SD WebUI
*** Error running postprocess: D:\Programas\stable-diffusion-webui\extensions\sd-webui-llul\scripts\llul.py, 1.14it/s]
Traceback (most recent call last):
File "D:\Programas\stable-diffusion-webui\modules\scripts.py", line 750, in postprocess
script.postprocess(p, processed, *script_args)
File "D:\Programas\stable-diffusion-webui\extensions\sd-webui-llul\scripts\llul.py", line 236, in postprocess
xi0 = int(x)
ValueError: invalid literal for int() with base 10: '133.1999969482422'
ValueError value changes when the gray box is moved to other place in the canvas. Tried a few things, like disabling Hypertile, AgentAttention, changing resolutions, etc. but nothing worked.
Error running process: G:\stable-webui\extensions\sd-webui-llul\scripts\llul.py
Traceback (most recent call last):
File "G:\stable-webui\modules\scripts.py", line 386, in process
script.process(p, *script_args)
File "G:\stable-webui\extensions\sd-webui-llul\scripts\llul.py", line 146, in process
self.last_hooker = Hooker(
TypeError: Hooker.init() got an unexpected keyword argument 'force_float'
As title says, can you make it so that we can drag image into the canvas, like controlnet do? For better visual reference.
When trying to use this in img2img I get below error. The generated file is completely black. Any tips?
Here's my versions:
python: 3.10.6 • torch: 2.0.0+cpu • xformers: N/A • gradio: 3.28.1 • commit: [3284ccc0](https://github.com/AUTOMATIC1111/stable-diffusion-webui/commit/3284ccc091f09146997fd93bb88a2ecd27ab3a1b) • checkpoint: [9aba26abdf](https://google.com/search?q=9aba26abdfcd46073e0a1d42027a3a3bcc969f562d58a03637bf0a0ded6586c9)
[aa47b3ee (Thu May 4 16:14:34 2023)](https://github.com/hnmr293/sd-webui-llul.git/commit/aa47b3eeb45c53f0d6ccaae59abf36e8ed6731f5)
Stacktrace:
Error running postprocess: C:\Users\steve\projects\stable-diffusion-webui-directml\extensions\sd-webui-llul\scripts\llul.py
Traceback (most recent call last):
File "C:\Users\steve\projects\stable-diffusion-webui-directml\modules\scripts.py", line 444, in postprocess
script.postprocess(p, processed, *script_args)
File "C:\Users\steve\projects\stable-diffusion-webui-directml\extensions\sd-webui-llul\scripts\llul.py", line 264, in postprocess
area_image = tensor_to_pil(area_tensor, mode='RGB')
File "C:\Users\steve\projects\stable-diffusion-webui-directml\venv\lib\site-packages\torchvision\transforms\functional.py", line 323, in to_pil_image
raise ValueError(f"Only modes {permitted_4_channel_modes} are supported for 4D inputs")
ValueError: Only modes ['RGBA', 'CMYK', 'RGBX'] are supported for 4D inputs
edit: seems to be a controlnet version issue, other plugins also say the same
Can you not add a function to simplify the details, and change the frame to a mask
Trying to figure this thing out. As a side note some docs might be helpful for all the advanced knobs.
That said, on main thing stopping me from using this more is the size of the box is often either too big or too small for what I want to do. Would it be possible to allow both moving AND scaling the box?
Hello!
Is it possible to run it in img2img mode? Can't see the difference.
Work fine in txt2img, but with a little bug, If you change the output size, you have to move a bit script's multiply slider forward and back - without that the preview size - that one with gray square - never changing.
Thanks!
I don't understand how it works, I select the enabled checkbox, I place the gray square in the place that interests me in the image, I press the generate button and it creates a totally different image instead of improving the area of the image that I wanted . What am I doing wrong? Can you explain to me what are the exact steps of use? Thank you so much.
For more than a month now people have been asking about this issue and we have sadly no info. I tried this on A1111, Vlad, after multiple clean installs too, and no, the grey box is gone.
Should we move on? Without that functionality this extensions is sadly pointless
Love this ext. Is it possible to add:
cheers!
I'm not sure if my understanding is right
This extension is using default Latent model to upscale specific area
But can you add an options that user can select different model (or upscale method)?
Like Hires. fix can let user select model from "stable-diffusion-webui\models\ESRGAN"
Could you add the position of each side of the square or rectangle that appears inside the canvas? That position could also be saved as generation info as it also changes how the image is composed.
Error running process: F:\WBC\sdwb\extensions\sd-webui-llul\scripts\llul.py
Traceback (most recent call last):
File "F:\WBC\sdwb\modules\scripts.py", line 386, in process
script.process(p, *script_args)
File "F:\WBC\sdwb\extensions\sd-webui-llul\scripts\llul.py", line 112, in process
raise ValueError(f'Image size must be multiple of 64 for LLuL, but {p.width}x{p.height}.')
ValueError: Image size must be multiple of 64 for LLuL, but 608x448.
So, it's not possible to use LLuL with an image size that's not a multiple of 64?
upd1: Please, if you know how to fix this, I would appreciate it. If it can't be fixed, how do I apply llul to images whose resolutions are not a multiple of 64? Inpainting?
sd-webui-llul issues value to controlnet extension, while it should not, breaking controlnet in img2img mode.
Problem may be in this part of code that deals with gradio buttons:
https://github.com/hnmr293/sd-webui-llul/blob/master/scripts/llul.py#L30-L50
I kept getting error about ffmpeg module not found after installing the extension. How to fix this?
I did try pip3 install ffmpeg-python, but since I am using webUI Automatic1111, I can't seem to install it there in their virtual environment.
Traceback (most recent call last):
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/ffmpy.py", line 93, in run
self.process = subprocess.Popen(
File "/opt/homebrew/Cellar/[email protected]/3.10.10_1/Frameworks/Python.framework/Versions/3.10/lib/python3.10/subprocess.py", line 971, in __init__
self._execute_child(args, executable, preexec_fn, close_fds,
File "/opt/homebrew/Cellar/[email protected]/3.10.10_1/Frameworks/Python.framework/Versions/3.10/lib/python3.10/subprocess.py", line 1847, in _execute_child
raise child_exception_type(errno_num, err_msg, err_filename)
FileNotFoundError: [Errno 2] No such file or directory: 'ffmpeg'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/routes.py", line 337, in run_predict
output = await app.get_blocks().process_api(
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/blocks.py", line 1013, in process_api
inputs = self.preprocess_data(fn_index, inputs, state)
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/blocks.py", line 911, in preprocess_data
processed_input.append(block.preprocess(inputs[i]))
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/components.py", line 1830, in preprocess
ff.run()
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/ffmpy.py", line 98, in run
raise FFExecutableNotFoundError(
ffmpy.FFExecutableNotFoundError: Executable 'ffmpeg' not found
Traceback (most recent call last):
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/ffmpy.py", line 93, in run
self.process = subprocess.Popen(
File "/opt/homebrew/Cellar/[email protected]/3.10.10_1/Frameworks/Python.framework/Versions/3.10/lib/python3.10/subprocess.py", line 971, in __init__
self._execute_child(args, executable, preexec_fn, close_fds,
File "/opt/homebrew/Cellar/[email protected]/3.10.10_1/Frameworks/Python.framework/Versions/3.10/lib/python3.10/subprocess.py", line 1847, in _execute_child
raise child_exception_type(errno_num, err_msg, err_filename)
FileNotFoundError: [Errno 2] No such file or directory: 'ffmpeg'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/routes.py", line 337, in run_predict
output = await app.get_blocks().process_api(
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/blocks.py", line 1013, in process_api
inputs = self.preprocess_data(fn_index, inputs, state)
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/blocks.py", line 911, in preprocess_data
processed_input.append(block.preprocess(inputs[i]))
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/gradio/components.py", line 1830, in preprocess
ff.run()
File "/Users/jimmygunawan/stable-diffusion-webui/venv/lib/python3.10/site-packages/ffmpy.py", line 98, in run
raise FFExecutableNotFoundError(
ffmpy.FFExecutableNotFoundError: Executable 'ffmpeg' not found
^Z
zsh: suspended ./webui.sh --no-half
My image is in ratio 3:2 or 2:3, I can't seem to get the LLUL to work properly, as it is complaining:
100%|█████████████████████████████████████████████████████████████████████████████| 16/16 [00:26<00:00, 1.67s/it]
activating extra network lora with arguments [<modules.extra_networks.ExtraNetworkParams object at 0x3be1e31f0>]: RuntimeError
Traceback (most recent call last):
File "/Users/jimmygunawan/stable-diffusion-webui/modules/extra_networks.py", line 75, in activate
extra_network.activate(p, extra_network_args)
File "/Users/jimmygunawan/stable-diffusion-webui/extensions-builtin/Lora/extra_networks_lora.py", line 23, in activate
lora.load_loras(names, multipliers)
File "/Users/jimmygunawan/stable-diffusion-webui/extensions-builtin/Lora/lora.py", line 151, in load_loras
lora = load_lora(name, lora_on_disk.filename)
File "/Users/jimmygunawan/stable-diffusion-webui/extensions-builtin/Lora/lora.py", line 113, in load_lora
module.weight.copy_(weight)
RuntimeError: output with shape [96, 320] doesn't match the broadcast shape [96, 320, 96, 320]
Thx for making a nice ext! However, moving the gray box does not seem to work. Is this a bug?
Env
python: 3.9.16
torch: 1.13.1+cu117
xformers: 0.0.14.dev
gradio: 3.16.2
commit: 0cc0ee1b
Error
Error running process: /storage/stable-diffusion/stable-diffusion-webui/extensions/sd-webui-llul/scripts/llul.py
Traceback (most recent call last):
File "/storage/stable-diffusion/stable-diffusion-webui/modules/scripts.py", line 386, in process
script.process(p, *script_args)
File "/storage/stable-diffusion/stable-diffusion-webui/extensions/sd-webui-llul/scripts/llul.py", line 154, in process
x=int(x) / p.width,
ValueError: invalid literal for int() with base 10: '123.55087280273438'
woops
edge browser No error reported by cmd
python: 3.10.7 • torch: 2.0.0+cu118 • xformers: 0.0.17 • gradio: 3.23.0 • commit: [22bcc7b]
Hello, first off I'm on the newest Auromatic1111 version and downloaded this extension yesterday. I am on Windows 10.
The issue:
As described in the title, the selection boss you can drag around to select the area for enriching the detail disappears if you close the menu (by clicking on the LLuL header) and then reopen it. It cannot be recovered until the UI is reloaded. It works in both txt2img and img2img.
Now I have to guess if the gray box is where I want it
No gray box after update A1111. But is it possible to somehow disable the gray box by making the field 100%? So that could load masks and work in the mask area? (maybe make Multiplication (2^N) less then 1 or just zero).
Now the mask works, but it is cut off by the area in the center of the image.
llul wasn't showing up and I don't understand any of these words to troubleshoot for myself.
Error loading script: llul.py Traceback (most recent call last): File "E:\stable-diffusion-webui\modules\scripts.py", line 248, in load_scripts script_module = script_loading.load_module(scriptfile.path) File "E:\stable-diffusion-webui\modules\script_loading.py", line 11, in load_module module_spec.loader.exec_module(module) File "<frozen importlib._bootstrap_external>", line 883, in exec_module File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed File "E:\stable-diffusion-webui\extensions\sd-webui-llul\scripts\llul.py", line 194, in <module> init_xyz(Script) File "E:\stable-diffusion-webui\extensions\sd-webui-llul\scripts\llul_xyz.py", line 61, in init_xyz define('Enabled', 0, to_bool, choices=['false', 'true']) File "E:\stable-diffusion-webui\extensions\sd-webui-llul\scripts\llul_xyz.py", line 59, in define data.module.axis_options.append(AxisOption(f'LLuL {param}', type, fn, choices=lambda: choices)) TypeError: AxisOption.__new__() got an unexpected keyword argument 'choices'
This function looks cool, it makes it possible to let img2img use latent .
Why the mask is the size of the grey box, why not the mask of the whole image?,
it's not convinient for mask editing, if you change that mask pic to the resolution of the full image,
it's very easy to make the mask picture, in Photoshop, just take a screen shot of your grey box position, then every child can find that mask area. finally, we just need to cut it to right size.
Hey thanks for your work, I am still getting
RuntimeError: "lerp_kernel_scalar" not implemented for 'Half'
I am running on M1 mac.
A long time ago when I used this, when there was no loading of mask pictures yet, everything was fine, you could choose the location of the interaction (for example, the very top of the picture or the very bottom), but now it is always in the middle, even if you load a mask picture where the white one is only up, it will be somewhere in the middle, and never at the top of the image. I changed the resolution of the mask, made it the same size as the picture, made it square, big and small and nothing helps... it's just always in the middle
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.