RunPod Auto DreamBooth Extension Of Automatic1111 Web UI & Latest Libraries Installer Script

ok i found the error and fixed. now works. please start a fresh pod and redownload installers :) ty

Furkan Gözükara

ok you are right currently installation is broken due to python packages. i will see if i can fix

Furkan Gözükara

hello. i dont suggest this extension anymore. do you need this exactly? by the way i don't see any breaking error. it still should be working from your messages.

Furkan Gözükara

Hello, After run this command apt update fuser -k 3000/tcp yes | apt install libcudnn8=8.9.5.29-1+cuda11.8 libcudnn8-dev=8.9.5.29-1+cuda11.8 --allow-change-held-packages cd /workspace/stable-diffusion-webui python relauncher.py I get this error TypeError: expected string or bytes-like object *** Error loading script: main.py Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/scripts.py", line 274, in load_scripts script_module = script_loading.load_module(scriptfile.path) File "/workspace/stable-diffusion-webui/modules/script_loading.py", line 10, in load_module module_spec.loader.exec_module(module) File "", line 883, in exec_module File "", line 241, in _call_with_frames_removed File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/main.py", line 8, in from dreambooth.diff_to_sd import compile_checkpoint File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/dreambooth/diff_to_sd.py", line 15, in from diffusers import UNet2DConditionModel File "/workspace/venv/lib/python3.10/site-packages/diffusers/__init__.py", line 58, in from .pipelines import ( File "/workspace/venv/lib/python3.10/site-packages/diffusers/pipelines/__init__.py", line 46, in from .audioldm import AudioLDMPipeline File "/workspace/venv/lib/python3.10/site-packages/diffusers/pipelines/audioldm/__init__.py", line 10, in if not (is_transformers_available() and is_torch_available() and is_transformers_version(">=", "4.27.0")): File "/workspace/venv/lib/python3.10/site-packages/diffusers/utils/import_utils.py", line 589, in is_transformers_version return compare_versions(parse(_transformers_version), operation, version) File "/workspace/venv/lib/python3.10/site-packages/packaging/version.py", line 52, in parse return Version(version) File "/workspace/venv/lib/python3.10/site-packages/packaging/version.py", line 196, in __init__ match = self._regex.search(version) TypeError: expected string or bytes-like object --- Applying attention optimization: xformers... done. *** Error reading github repository info from /workspace/stable-diffusion-webui/extensions/deforum Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/extensions.py", line 53, in do_read_info_from_repo repo = Repo(self.path) File "/workspace/venv/lib/python3.10/site-packages/git/repo/base.py", line 282, in __init__ self.working_dir: Optional[PathLike] = self._working_tree_dir or self.common_dir File "/workspace/venv/lib/python3.10/site-packages/git/repo/base.py", line 363, in common_dir raise InvalidGitRepositoryError() git.exc.InvalidGitRepositoryError --- *** Error reading github repository info from /workspace/stable-diffusion-webui/extensions/sd-webui-controlnet Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/extensions.py", line 53, in do_read_info_from_repo repo = Repo(self.path) File "/workspace/venv/lib/python3.10/site-packages/git/repo/base.py", line 282, in __init__ self.working_dir: Optional[PathLike] = self._working_tree_dir or self.common_dir File "/workspace/venv/lib/python3.10/site-packages/git/repo/base.py", line 363, in common_dir raise InvalidGitRepositoryError() git.exc.InvalidGitRepositoryError --- *** Failed reading extension data from Git repository (sd_dreambooth_extension) Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/extensions.py", line 65, in do_read_info_from_repo if repo.active_branch: File "/workspace/venv/lib/python3.10/site-packages/git/repo/base.py", line 881, in active_branch return self.head.reference File "/workspace/venv/lib/python3.10/site-packages/git/refs/symbolic.py", line 309, in _get_reference raise TypeError("%s is a detached symbolic reference as it points to %r" % (self, sha)) TypeError: HEAD is a detached symbolic reference as it points to 'dc413a14379b165355502d9f65856c40a4bb5b6f' I'm following this video https://www.youtube.com/watch?v=-QaJmS_Onts Do you know how can I fix it?

Pedro Atencio

just delete download models py files and it will work without downloading

Furkan Gözükara

Can you make another version of script, which doesn't download models, just install..?

Nenad Kuzmanovic

great

Furkan Gözükara

Thanks this post saved me huge amounts of time :)

Hazem

It is true. When the Web UI starts at port 3000, we connect it from port 3001 from RunPod proxy I need to see your commands that you executed on your RunPod terminal to help you. you can join discord and write in our Stable Diffusion or Patreon channel

Furkan Gözükara

Hello and thank you for your videos! I used your instruction from this video (manual installation): https://www.youtube.com/watch?v=c_S2kFAefTQ&ab_channel=SECourses After I finish all the steps and try to connect to HTTP Service, it brings me to the page with 502 error. So I just can't get into SD interface. Have tried it multiple times by turning the pod on and off. Tried on two different pods with different GPUs as well. By the way, the HTTP Service button contains Port 3001, not Port 3001, not sure if it should be that way. Is there any way to fix it? I'm a beginner user of Runpod and SD, so maybe there's something simple I'm missing. Thanks!

Ася Александрова

Hello. As I replied from Youtube since the extension is not maintained anymore we are installing older certain version of Automatic1111. So you need to do training with that version and then you can use those trained models on the latest version of Autoımatic1111 web ui.

Furkan Gözükara

Hello, I have updated automac1111 using your script. I thought that after the update I would be able to work with the ip adapter. But the ip adapter preprocessors did not appear in controlnet. Tell me, please, is it possible to download and upload these preprocessors to a folder without updating automatic 1111? And in which folder should they be added to work with them? Or is it necessary to update automatic1111 to the required version in order to be able to work with the ip adapter?

Dmitry

I just tested working First we will run install1 . sh file with below command cd /workspace chmod +x install1.sh ./install1.sh After above install you will get some errors. Ignore them. Below step will fix all errors. Then we will run install2 . sh file with below command cd /workspace chmod +x install2.sh ./install2.sh

Furkan Gözükara

hello welcome. did you watch this short video? can you try again and let me know? i will also check and verify : https://youtu.be/a8WESfPwlYw?si=Q0takCCKKUw1H8OT

Furkan Gözükara

Hi. First of all thank you for this. I'm having trouble when restarting the pod I get this message at the end of the automatic instalated. I use this command: fuser -k 3000/tcp cd /workspace/stable-diffusion-webui python relauncher.py Then I get this https://www.canva.com/design/DAFz83Uxjb8/ovMPM9tDfGltJW02tXVNbw/view?utm_content=DAFz83Uxjb8&utm_campaign=designshare&utm_medium=link&utm_source=editor Anyway you know what am I doing wrong? or if there is something I didn't update?

Michel W Tandane

for download models use wget. i explained in this video thoroughly : https://youtu.be/QN1vdGhjcRc

Furkan Gözükara

Thank you for your reply! it works🥳 I missed "/" in front of the path🤣 Also is upload checkpoint to the models file in Jupyter notebook the only way or can we paste link into the script so it can download the checkpoint model automatically like training process in many colabs did? Because now I need to download the model from internet, then upload it to Jupyter notebook, Im thinking if there's other way which is more direct🤔

han hsun shih

hello it is easy. start watching from 6:54. you can put into workspace and give its path : https://youtu.be/zA4LksIVas8?si=8TWyQZ5pRRMMJhYr

Furkan Gözükara

Hello, I couldn't run Dreambooth on mac so I eventually changed to runpod. Everything has doing well till I tried to train the dreambooth model. I followed this tutorial , https://youtu.be/g0wXIcRhkJk?t=899, after the setup the SD model with runpod but I'm not sure where should I put my dataset. I tried to put the dataset file in workspace in JupyterNotebook and in my local Mac, but none of them work. In the terminal in JupyterNotebook it shows: Invalid Dataset Directory: workspace/FileName or Invalid Dataset Directory: /Users/UserName/Desktop/FileName. Thank you for your answer in advance!

han hsun shih

I just did a fresh install. turned off pod and started again. after that execute below command without any install. web ui started perfectly fine no error apt update fuser -k 3000/tcp yes | apt install libcudnn8=8.9.5.29-1+cuda11.8 libcudnn8-dev=8.9.5.29-1+cuda11.8 --allow-change-held-packages cd /workspace/stable-diffusion-webui python relauncher.py

Furkan Gözükara

Fetch Walker

Thank you, The video @ "https://youtu.be/a8WESfPwlYw?si=o--_PJEv3GdEu5rV " Worked for me.

Fetch Walker

can you follow this video? https://youtu.be/a8WESfPwlYw?si=o--_PJEv3GdEu5rV - also try a new pod. if you still can't make it please record your screen and send me how you are trying

Furkan Gözükara

Fetch Walker

Thank you! 🙏

Manpreet Singh

SDXL is far better. you can safely switch to SDXL

Furkan Gözükara

Would images genrrated by SDXL in the current state be better than SD1.5 with webui or do you recommend to wait until you improve this further?

Manpreet Singh

yes it is available here as json file : https://www.patreon.com/posts/very-best-for-of-89213064

Furkan Gözükara

I see. Would SDXL be available in script form when you're done searching for parameters?

Manpreet Singh

SDXL is far superior. but SD 1.5 web UI better at the moment. I am searching for parameters

Furkan Gözükara

How would you rate/compare the quality of kohya generated dreambooth images vs WebUI?

Manpreet Singh

sadly i don't have. but for that purpose you can use kohya. it is run through command line interface

Furkan Gözükara

Do you have a script that allows us to use this functionality completed scripted / headless without using the Automatic Web UI?

Manpreet Singh

It was occupying a resolution of 768x1024. Right now I'm watching the videos on SDXL Training With Kohya to see if it gives me better results. Thanks for all the information you share!

Abdi Abdi

it depends on resolution. what resolution have you tested? by the way 24 s/it is 10 times slower than 2.5 s/it

Furkan Gözükara

Hello, good night. I was following the instructions in the unlisted video and also the instructions in the video "The END of Photography" and when training the model in Dreamboth it takes approximately 7 hours and goes at 2.52 s/it. It is normal? Since I saw in the video that it reached up to 24 s/it

Abdi Abdi

ok thanks

Astore Az

start on a new pod and follow this video please : https://youtu.be/MOVEcvn5h-0 - i may make a quick new tutorial hopefully tonight

Furkan Gözükara

Birinci defe açıldı. Amma sonradan 2 ci defe daxil oluram bele bir hata baş verir. Attempting uninstall: gradio 2023-10-16T15:45:07.086071521Z Found existing installation: gradio 3.41.2 2023-10-16T15:45:07.086074091Z 2023-10-16T15:45:07.086076652Z stderr: WARNING: Error parsing requirements for gradio: [Errno 2] No such file or directory: '/workspace/venv/lib/python3.10/site-packages/gradio-3.41.2.dist-info/METADATA' 2023-10-16T15:45:07.086079362Z WARNING: No metadata found in /workspace/venv/lib/python3.10/site-packages 2023-10-16T15:45:07.086082042Z WARNING: No metadata found in /workspace/venv/lib/python3.10/site-packages 2023-10-16T15:45:07.086084582Z WARNING: No metadata found in /workspace/venv/lib/python3.10/site-packages 2023-10-16T15:45:07.086088102Z WARNING: No metadata found in /workspace/venv/lib/python3.10/site-packages 2023-10-16T15:45:07.086090672Z WARNING: No metadata found in /workspace/venv/lib/python3.10/site-packages 2023-10-16T15:45:07.086093172Z WARNING: No metadata found in /workspace/venv/lib/python3.10/site-packages 2023-10-16T15:45:07.086095682Z ERROR: Cannot uninstall gradio 3.41.2, RECORD file not found. You might be able to recover from this via: 'pip install --force-reinstall --no-deps gradio==3.41.2'. 2023-10-16T15:45:07.086098272Z 2023-10-16T15:45:09.099243326Z Relauncher: Launching... 2023-10-16T15:45:09.099263416Z Relauncher: Process is ending. Relaunching in 2s...

Astore Az

select : stable-diffusion:web-ui-10.2.1

Furkan Gözükara

sure it is not an issue. Rica ederim.

Furkan Gözükara

which template i must select in runpud. RunPod Stable Diffusion runpod/stable-diffusion:web-ui-10.2.1 or RunPod Fast Stable Diffusion runpod/stable-diffusion :fast-stable-diffusion-2.4.0

Astore Az

I am from Azerbaijan i can speak turkish :) I am trying again. if is will be happen i write here. Uzurlu sayin deyerli vaxtınızı alıram.

Astore Az

hello. did you run install 1 and 2 separately? did you select web ui template? i can make a quick tutorial hopefully today for show how to use on runpod

Furkan Gözükara

I use runpod. When i want to instaling i get this error.

Astore Az

################################################################ Launching launch.py... ################################################################ Using TCMalloc: libtcmalloc_minimal.so.4 Python 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0] Version: v1.4.0 Commit hash: 394ffa7b0a7fff3ec484bcd084e673a8b301ccc8 Traceback (most recent call last): File "/workspace/stable-diffusion-webui/launch.py", line 38, in main() File "/workspace/stable-diffusion-webui/launch.py", line 29, in main prepare_environment() File "/workspace/stable-diffusion-webui/modules/launch_utils.py", line 268, in prepare_environment raise RuntimeError( RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check Relauncher: Process is ending. Relaunching in 2s...

Astore Az

awesome ty so much

Furkan Gözükara

Thank you very much. I see that this post is very recent. I will try to follow the instructions to the letter, and I will inform you if I fail again in search of your wise advice. I truly appreciate all your courses, knowledge, and above all, the time you dedicate.

Rafael Villalba

hello. you can message me your settings from discord along with dataset so i can comment. by the way on runpod i suggest dreambooth of SDXL much better : https://www.patreon.com/posts/very-best-for-of-89213064

Furkan Gözükara

Hello good day, I have been using this post, with the intention of doing a training that allows me to generate my own photos, Currently, I have chosen to use runpod for this purpose since my PC (it's a modest RTX 3050 4G), I have done all the configuration, and I have used the scripts to do it automatically, and I have managed to make the checkpoints (everything looks successful up to this point), I also checked to have photos dataset as recommended, but for some reason when generating the new photos using your prompts the results I get are not close to the fidelity I see in your photos, Currently, from your tutorials, would this be the one you would recommend to get the best quality in my own portrait photos? Could you help me and thank you very much

Rafael Villalba

just happened to me too. i recorded a video and shown how to fix. https://twitter.com/GozukaraFurkan/status/1710080934752436551 youtube video is also being uploaded also ignore HEAD is a detached symbolic reference as it points to

Furkan Gözükara

so you believe 2.0 is best?

Furkan Gözükara

just happened to me too. i recorded a video and shown how to fix. https://twitter.com/GozukaraFurkan/status/1710080934752436551 youtube video is also being uploaded also ignore HEAD is a detached symbolic reference as it points to

Furkan Gözükara

and it loops the process of installation to infinity

Andrey Reshetnikov

OSError: Port 3000 is in use. If a gradio.Blocks is running on the port, you can close() it or gradio.close_all(). *** Failed reading extension data from Git repository (sd_dreambooth_extension) Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/extensions.py", line 65, in do_read_info_from_repo if repo.active_branch: File "/workspace/venv/lib/python3.10/site-packages/git/repo/base.py", line 881, in active_branch return self.head.reference File "/workspace/venv/lib/python3.10/site-packages/git/refs/symbolic.py", line 309, in _get_reference raise TypeError("%s is a detached symbolic reference as it points to %r" % (self, sha)) TypeError: HEAD is a detached symbolic reference as it points to 'dc413a14379b165355502d9f65856c40a4bb5b6f'

Andrey Reshetnikov

You may want to pick a different source model to train from instead of RealisticVisionv5.1. I have trained 50+ LORAs locally (testing different settings, datasets etc..) and only recently discovered many of the quality issues I was having were when I switched from RealisticVisionv20 to v5.1 - of course your results may be different but thought that I'd share. I have not tested any other RealisticVision versions

Goldwaters

Hello. I just did a new training and it is working great with Realistic Vision. still not supporting SDXL sadly. But they are working on it.

Furkan Gözükara

ah yes they updated 1.6 just today or something. let me test and fix errors. so sorry for late reply. will message you once done

Furkan Gözükara

I am getting issue where all trained dreambooth models are being created with broken Clip and Broken Unet. This is using best Dreambooth settings you posted and has been reproduced 3 separate times on 3 separate pods. I'm assuming this might be due to web-ui v1.6 but was also curious if anyone else is having this issue?

john denver

that is a good solution. but if you use python relauncher py to restart web ui it will become permanent once again

Furkan Gözükara

Use this attribute to stop auto starts: RUNPOD_STOP_AUTO = 1

Bobbie

hello updated the script. can you download again and try again? use fast stable diffusion template not web ui template.

Furkan Gözükara

hello. going to fix it hopefully now and update script thanks for letting me know

Furkan Gözükara

I get the following error when I run install2. RuntimeError: Detected that PyTorch and torchvision were compiled with different CUDA versions. PyTorch has CUDA Version=11.7 and torchvision has CUDA Version=11.8. Please reinstall the torchvision that matches your PyTorch install. Relauncher: Process is ending. Relaunching in 2s... Then I guessed that the problem could be in this line of the script yes | pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 and I replaced it with: CUDA_VERSION=11.8 yes | pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu${CUDA_VERSION} Then the error I get is: Traceback (most recent call last): File "/workspace/stable-diffusion-webui/launch.py", line 39, in main() File "/workspace/stable-diffusion-webui/launch.py", line 30, in main prepare_environment() File "/workspace/stable-diffusion-webui/modules/launch_utils.py", line 314, in prepare_environment raise RuntimeError( RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check Relauncher: Process is ending. Relaunching in 2s... How can I solve this?

Don Quijote de la Mancha

You can ignore this warning. It is still working.

Furkan Gözükara

fuser -k 3000/tcp always gives an error. Cannot stat file /proc/43/fd/0: Permission denied Cannot stat file /proc/43/fd/1: Permission denied Cannot stat file /proc/43/fd/2: Permission denied Cannot stat file /proc/43/fd/4: Permission denied Cannot stat file /proc/43/fd/5: Permission denied Cannot stat file /proc/43/fd/6: Permission denied Cannot stat file /proc/43/fd/7: Permission denied Cannot stat file /proc/43/fd/8: Permission denied Cannot stat file /proc/43/fd/9: Permission denied Cannot stat file /proc/43/fd/10: Permission denied Cannot stat file /proc/43/fd/11: Permission denied Cannot stat file /proc/43/fd/12: Permission denied Cannot stat file /proc/43/fd/13: Permission denied

Ertan Bıçakçı

When you close runpod all files will remain. If you delete pod only then they will get deleted. But still you can download everything with runpodctl. And thanks a lot for Patreon support.

Furkan Gözükara

Thank you again for these great tutorials! I used runpod for the first time today and am using your latest "End of Photography..video settings" and everything is going smoothly and currently performing my second training set. When you close the runpod session i assume all files are deleted so I will make sure to copy the training parameters I used in the json files as well as download all of the chckpts.

Ec Jep

i am glad you solved

Furkan Gözükara

awesome.

Furkan Gözükara

Okay. Now I restarted the pod after editing relauncher.py and I changed the rm -r venv line in the script to full path. Then I got those error at the end of setup1.sh: Launching Web UI with arguments: -f --xformers --port 3000 --listen --enable-insecure-extension-access 2023-07-09 13:22:55,721 - ControlNet - INFO - ControlNet v1.1.231 ControlNet preprocessor location: /workspace/stable-diffusion-webui/extensions/sd-webui-controlnet/annotator/downloads 2023-07-09 13:22:55,802 - ControlNet - INFO - ControlNet v1.1.231 Loading weights [e1441589a6] from /workspace/stable-diffusion-webui/models/Stable-diffusion/v1-5-pruned.ckpt Creating model from config: /workspace/stable-diffusion-webui/configs/v1-inference.yaml LatentDiffusion: Running in eps-prediction mode DiffusionWrapper has 859.52 M params. Textual inversion embeddings loaded(0): Model loaded in 5.6s (load weights from disk: 2.8s, create model: 0.4s, apply weights to model: 0.5s, apply half(): 0.3s, load VAE: 0.9s, move model to device: 0.2s, calculate empty prompt: 0.5s). Exception importing api Traceback (most recent call last): File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/api.py", line 27, in from dreambooth.dataclasses.db_config import from_file, DreamboothConfig File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/dreambooth/dataclasses/db_config.py", line 12, in from dreambooth.utils.image_utils import get_scheduler_names # noqa File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/dreambooth/utils/image_utils.py", line 11, in from diffusers.schedulers import KarrasDiffusionSchedulers ModuleNotFoundError: No module named 'diffusers' *** Error loading script: main.py Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/scripts.py", line 274, in load_scripts script_module = script_loading.load_module(scriptfile.path) File "/workspace/stable-diffusion-webui/modules/script_loading.py", line 10, in load_module module_spec.loader.exec_module(module) File "", line 883, in exec_module File "", line 241, in _call_with_frames_removed File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/main.py", line 7, in from dreambooth.dataclasses.db_config import from_file, save_config File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/dreambooth/dataclasses/db_config.py", line 12, in from dreambooth.utils.image_utils import get_scheduler_names # noqa File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/dreambooth/utils/image_utils.py", line 11, in from diffusers.schedulers import KarrasDiffusionSchedulers ModuleNotFoundError: No module named 'diffusers' --- Applying attention optimization: xformers... done. *Deforum ControlNet support: enabled* preload_extensions_git_metadata for 10 extensions took 0.43s Running on local URL: http://0.0.0.0:3000 To create a public link, set `share=True` in `launch()`. *** Error executing callback app_started_callback for /workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/api.py Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/script_callbacks.py", line 126, in app_started_callback c.callback(demo, app) File "/workspace/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/api.py", line 530, in dreambooth_api model_cfg: DreamboothConfig = Body(description="The config to save"), NameError: name 'DreamboothConfig' is not defined --- Startup time: 11.1s (import torch: 1.2s, import gradio: 0.9s, import ldm: 0.9s, other imports: 0.7s, load scripts: 6.5s, create ui: 0.6s, gradio launch: 0.1s). But I ignored them and started setup2.sh after that and after removing --skip-install and restarted webui the error were gone! :) Thank you very much!

Nerdlicht

Ah, I forgot to restart the pod before proceeding. I'll try that in some minutes. Btw, could it be that the setup1.sh script tries to remove the venv dir after it changed the dir to /workspace/stable-diffusion-webui, but the venv dir is in /workspace

Nerdlicht

ok here the steps. start a fresh pod. 1st change relauncher py. while (n<1): - after this dont forget to restart pod then upload scripts to workspace first execute install1.sh once it is done execute install2.sh i also added 1 more line to the github file i just did a fresh install and working amazing if you still cant make just upgrade the patreon level and i will connect your pc and show you how to install if you wish

Furkan Gözükara

Hello. I will test and update. Sorry for delay. Did you change relauncher py and made it n<1 and restart pod?

Furkan Gözükara

After "python relauncher.py" from manual steps I get this error at the end: Running setup.py install for lit ... done Successfully installed MarkupSafe-2.1.2 certifi-2022.12.7 charset-normalizer-2.1.1 cmake-3.25.0 filelock-3.9.0 idna-3.4 jinja2-3.1.2 lit-15.0.7 mpmath-1.2.1 networkx-3.0 numpy-1.24.1 pillow-9.3.0 requests-2.28.1 sympy-1.11.1 torch-2.0.1+cu118 torchaudio-2.0.2+cu118 torchvision-0.15.2+cu118 triton-2.0.0 typing-extensions-4.4.0 urllib3-1.26.13 Traceback (most recent call last): File "/workspace/stable-diffusion-webui/launch.py", line 38, in main() File "/workspace/stable-diffusion-webui/launch.py", line 29, in main prepare_environment() File "/workspace/stable-diffusion-webui/modules/launch_utils.py", line 268, in prepare_environment raise RuntimeError( RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check Relauncher: Process is ending. Relaunching in 2s... root@04e3a1a68393:/workspace/stable-diffusion-webui# I hope I can ignore it :)

Nerdlicht

Great that is better

Furkan Gözükara

Thank you I will message you in discord

John Vee

Hopefully I can try later and tell you solution. Can you show me screenshot of your settings please? You can also message me from discord.

Furkan Gözükara

ok I've tested the tiling method - unfortunately same error. I wonder how I can achieve 6000x6000px without error. Thank you for your time sir

John Vee

thank you sir I'll check your patreon posts and youtube video for this. Do you think Controlnet can handle this resolution?

John Vee

OK looks like your only option is using controlnet tiling feature. You know how to use it for upscaling? I have a whole new video of controlnet very recent. I show how to use tiling feature in that video. Look video chapters. Looks like that vram not sufficient for that resolution direct upscale

Furkan Gözükara

hello sir yes I removed --xformers (I didn't have --opt-sdp-attention) from webui-user.sh and even restarted pod - still getting error: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.43 GiB (GPU 0; 23.69 GiB total capacity; 13.94 GiB already allocated; 5.85 GiB free; 17.53 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF I'm using RTX3090 and trying to do a 2 x upscale of 3000x3000 using img2img / 4 x Ultrasharp. I have tried the other upscalers too with the same issue

John Vee

Awesome

Furkan Gözükara

Hello. Turn off xformers and opt sdp attention from starting command line arguments in web ui user.sh file. Remove them

Furkan Gözükara

Hello when I upscale img2img higher than 4000x4000px I keep getting error "OutOfMemoryError: CUDA out of memory. Tried to allocate 39.45 GiB (GPU 0; 47.54 GiB total capacity; 11.90 GiB already allocated; 34.15 GiB free; 13.05 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF" it doesn't matter even if I choose A6000 GPU. thanks

John Vee

I just wanted to write that it works now! I removed the skip install, it installed the requirements (dont know which and from where), but then the error was gone.

Nerdlicht

i found the problem. extension developer has to fix it. opened an issue thread : https://github.com/AlUlkesh/sd_grid_add_image_number/issues/10

Furkan Gözükara

can you please remove it and restart web ui and try again? and let me know? i wonder if that is the reason. by the way i will test the extension ty for full link

Furkan Gözükara

--skip-install is set but the extensions has no requuirements.txt: https://github.com/AlUlkesh/sd_grid_add_image_number

Nerdlicht

I will test the extension. can you verify that --skip-install is set or not in your webui user sh file. if it is set that it wont install requirements of the extension

Furkan Gözükara

I used this guide to install DB/SD and wanted to install the "sd_grid_add_image_number" extension. Additionally I had to install the "fonts" module, but nonetheless I get an error: Error loading script: sd_grid_add_image_number.py Traceback (most recent call last): File "/workspace/stable-diffusion-webui/modules/scripts.py", line 263, in load_scripts script_module = script_loading.load_module(scriptfile.path) File "/workspace/stable-diffusion-webui/modules/script_loading.py", line 10, in load_module module_spec.loader.exec_module(module) File "", line 883, in exec_module File "", line 241, in _call_with_frames_removed File "/workspace/stable-diffusion-webui/extensions/sd_grid_add_image_number/scripts/sd_grid_add_image_number.py", line 5, in from fonts.ttf import Roboto ImportError: cannot import name 'Roboto' from 'fonts.ttf' (/workspace/venv/lib/python3.10/site-packages/fonts/ttf/__init__.py) I never had this problem in local installations :/

Nerdlicht

we don't have such script on dreambooth or kohya web ui. it can only be made by finding the training script of these repos and modifying them. pretty complex to do. your above solution much better.

Furkan Gözükara

i think this is your best option. look for estimated training time and set according to that. this is a good way i think

Furkan Gözükara

I got the answer from runpod as follows "Hi, if you submit your training request from the command line, you can do something like this to stop your Pod once it's over: ./submit_training_request.sh && runpodctl stop pod $RUNPOD_POD_ID Where submit_training_request.sh is a bash script which begins the training process."

Thịnh Nguyễn

can you guide me on Automatically Stop My Pod after the training is over. I found this command but it is on predefined time (sleep 2h; runpodctl stop pod $RUNPOD_POD_ID) &

Thịnh Nguyễn