Best & Easiest Way to Run Stable Diffusion for Free (WebUI)

Featured Image - Best & Easiest Way to Run Stable Diffusion for Free (WebUI)
[Generating 9 Batches of Images] Prompt: “hyperrealistic portrait of a cyberpunk man, medium hair, confident, cybernetics, immersed within a network, by Guy Denning, Derek Gores, Russ Mills, glitch art, hyper focus, fined detail, polished, complex, hacking effects, holographic, digital tech effects, color blocking!, green, realistic, acrylic on canvas, concept art, abstract, 8k. trending on cgsociety, trending on artstation”

Stable Diffusion is among the best AI art generators at the time of writing. It generates fantastic art, it has relatively low hardware requirements, and it’s fast. One of the best things about it is that it’s also available as free and open-source.

Thanks to Stable Diffusion being open-source, it can be integrated it into your own projects. Whether you want to use it as a standalone AI art generator, or as part of a larger AI art project, it’s easy to set up and use.

This has lead to users finding all sorts of interesting ways to use Stable Diffusion. This is excellent news, as it means that the software is constantly evolving and improving. As more people use it, and more ideas are shared, the software just gets better and better.

One of these projects is Stable Diffusion WebUI by AUTOMATIC1111, which allows us to use Stable Diffusion, on our computer or via Google Colab1Google Colab is a cloud-based Jupyter Notebook. Jupyter Notebooks are, in simple terms, interactive coding environments. Think of them as documents that allow you to write and execute code all in one place.

Google Colab is a service that provides free Jupyter Notebooks that are run on Google’s servers. This means that you can use Google Colab to write and execute code without having to download or install anything on your own computer.

Google Colab is a great tool for data scientists and machine learning engineers as it allows you to prototype and experiment with your code in a fast and convenient way. Additionally, Colab is perfect for sharing your work with others, as you can simply share the link to your notebook and anyone can view and execute the code.
, using an intuitive web interface with many interesting and time saving options and features, allowing us to continuously experiment with creating AI generated images for free.

In this tutorial we’ll cover what is, in my opinion, the best way to run Stable Diffusion for free. We’ll set up and run Fast Stable Diffusion WebUI by AUTOMATIC1111 on Google Colab, so we can generate AI art no matter our computer’s hardware.

Basically you can create hundreds of images per day for free, and they’ll all be stored in our Google Drive. This is invaluable, especially considering that a prompt won’t always result in the perfect image right away, and you may have to keep on experimenting.

Fast Stable Diffusion WebUI by AUTOMATIC1111 is a modified version of Stable Diffusion WebUI by AUTOMATIC1111, which is currently the most popular implementation of Stable Diffusion that comes with the most features out of the variants that I’ve tried.

The reason we’re using the Fast variant is because it’s a lot more user friendly, very easy to set up in Google Colab, and possibly faster. You can find its Github repository here. The author also optimizes other variants of Stable Diffusion implementations, such as the hlky version.

The reasons I consider this method to be the best are:

  1. It’s free.
  2. You’re running it in Google Colab. This means you don’t need to worry about your hardware. You can even run it from your phone. For a beginner friendly intro on Google Colab check out our related tutorial.
    • You can also run it locally, on your computer. Install instructions are available for Linux, Windows and Mac. I have an NVIDIA 1660 SUPER and a 512x512px image takes ~2 minutes, vs 15-20s on Google Colab with an NVIDIA Tesla T4 or P100.
  3. The setup is easy, beginner friendly, and relatively fast. Even though we’re using Google Colab to run it, it’s not complicated. You’ll just run a few initial steps, wait ~10 minutes, and then use Stable Diffusion from a very nice web interface.
  4. It has an easy to use and intuitive web interface, that you can access from your browser, and it has lots of features.
    • It displays an accurate image generation progress bar.
  5. It offers a lot very cool extra features such as:
    • Upscaling and face correction. It comes with two popular algorithms, to upscale images (increase their resolution), and to fix distorted faces (because Stable Diffusion may slightly distort some faces). The upscaling is done using ESRGAN and the restoration is done using GFPGAN.
    • The ability to write multiple prompts separated by |, and the system will create an image for each of them.
    • Downloading a fresh Stable Diffusion model or loading your own custom one. This is very useful because you can simply load an existing Stable Diffusion model from your Google Drive. And since recently we’ve been able to fine-tune Stable Diffusion, we can use the Web UI to use our fine tuned models.
    • You can check out more info and features in its Github page.

Here’s a quick preview of how it looks like:

Quick Setup & Image Generation Video Demo

This is a demo of what we’ll be doing to set it up and start using Stable Diffusion WebUI by AUTOMATIC1111. It’s not sped up so you can get an idea of how long it takes. As you can see, it’s very simple and straightforward.


tl;dr AI News Section Preview
tl;dr AI News Section Preview

Sidenote: AI art tools are developing so fast it’s hard to keep up.

We set up a newsletter called tl;dr AI News.

In this newsletter we distill the information that’s most valuable to you into a quick read to save you time. We cover the latest news and tutorials in the AI art world on a daily basis, so that you can stay up-to-date with the latest developments.

Check tl;dr AI News

Setting Up Fast Stable Diffusion by AUTOMATIC1111 in Google Colab

Open Colab Notebook

First we’ll open the Google Colab notebook for Fast Stable Diffusion by AUTOMATIC1111. To do this click here https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast_stable_diffusion_AUTOMATIC1111.ipynb to open it.

You can also copy it to your Google Drive by clicking the Copy to Drive button. This will open your new copy in a new tab and you can switch to it.

Copy Google Colab Notebook to Your Google Drive
Copy Google Colab Notebook to Your Google Drive

Enable GPU

Next we’ll need to enable the use of a GPU. The GPU will be allocated to us by Google.

To do this go in the menu and click Runtime > Change runtime type.

Runtime > Change runtime type
Runtime > Change runtime type

A small popup will appear with the title Notebook settings. We want to make sure in the dropdown under Hardware accelerator we have GPU selected.

Hardware accelerator > GPU
Hardware accelerator > GPU

Click Save when you’re done.

Run The First Cell & Connect Google Drive

Next we’ll run the first cell which will connect Google Colab to our Google Drive, so will have access to save our generated images directly into our drive.

Run First Cell to Connect Google Drive
Run First Cell to Connect Google Drive

You’ll get a popup asking if you’ll permit this notebook to access Google Drive. To allow it we’ll click Connect to Google Drive.

Connect to Google Drive
Connect to Google Drive

After which you’ll get another popup to select the Google account to connect with, and then another popup showing you the permissions you’re giving the notebook to your Google Drive.

After you’re done you’ll get a small output saying Mounted at /content/drive.

Mounted Google Drive
Mounted Google Drive

Input Your Hugging Face Token or Path to Pretrained Model

Before running every cell in order, we’ll first fill in the cell Model Download/Load.

Model Download/Load Section
Model Download/Load Section

This cell offers us three options. To either download the Stable Diffusion model, or to load it from our Google Drive, if we already have it downloaded. As such, we have one of two fields to fill:

Option 1: token (Download Stable Diffusion)

Option 1: Download the Stable Diffusion Model
Option 1: Download the Stable Diffusion Model

If we don’t already have the Stable Diffusion model downloaded, we can input a token from Hugging Face, that will allow the notebook to download it for us. If this is the first time you’re using Stable Diffusion, or just aren’t sure if you have it downloaded in your Google Drive, then I recommend we go with this option now. The next time you run it you won’t need to download it again, since we’ll already have Stable Diffusion in our Google Drive.

If you already downloaded the Stable Diffusion model, but for some reason want to redownload it, you can check the Redownload_the_original_model box.

  • To get your token just sign up for Hugging Face at https://huggingface.co/join. It’s very straightforward.
  • After you confirm your account just click on your profile picture in the upper right corner of the screen and go to your Settings > Tokens. Or you can just go there directly by clicking https://huggingface.co/settings/tokens.
  • There you can create an access token. A token is like a password. Just click on New Token and then the following popup will appear. Name it anything you want (the name is just for reference purposes), and click on Generate a token.
    Create New Access Token
    Create New Access Token
  • Next we have to agree to some terms for Hugging Face to allow us to download the Stable Diffusion model. Visit this URL https://huggingface.co/CompVis/stable-diffusion-v-1-4-original and you’ll have to click a checkbox that you agree to some terms, and then click Access repository. The terms allow access to our email and username to the authors of Stable Diffusion.
    Access repository
    Access repository
  • Finally we can input our token in the notebook in the token field.
    Input Token/Download Model
    Input Token/Download Model

Option 2: Path_to_trained_model (Load Existing Stable Diffusion from Google Drive)

If you had already downloaded Stable Diffusion before, of have a fine-tuned version of it (in the form of .ckpt) then you can easily load it. Just input the path to it in the Path_to_trained_model field.

Load Model. In my case the path is /content/gdrive/MyDrive/AI/models/sd-v1-4.ckpt
Load Model. In my case the path is /content/gdrive/MyDrive/AI/models/sd-v1-4.ckpt

If you have a Stable Diffusion model shared via your Google Drive with your current account, then you can insert the link to it in the Link_to_trained_model.  For example, say someone (or you, from a different Google account) shared a customized Stable Diffusion model with your current Google account.

Now it will show up in your Google Drive, but it won’t be physically stored there. To get the link to it, right click on it and click on Get Link.

A small popup will appear, with a button in the bottom left called Copy Link. Click on it and you’ll have your link.

Share Google Drive Model
Share Google Drive Model

Having the link, you can input it into the Link_to_trained_model field:

Model Load Shared Google Drive Model
Model Load Shared Google Drive Model

If you have already run this notebook before then you can leave both fields empty, because the notebook will automatically look for Stable Diffusion in /content/gdrive/MyDrive/sd/stable-diffusion-webui/models/Stable-diffusion/model.ckpt.

Run Every Other Cell & Wait for It to Finish

After you input either your token or your Path_to_trained_model you can run all the cells in order and wait.

We’ve already connected Google Drive, so we’ll run every cell starting with cell 2 – Installing AUTOMATIC1111 repo, until the final one, and then we’ll wait. The process should take ~10 minutes.

Running Every Other Cell
Running Every Other Cell

Access the Stable Diffusion WebUI by AUTOMATIC1111

After everything has finished running, underneath the last cell we’ll see some links generated. If you left Use_Gradio_Server unchecked, the link will look like https://fancy-spies-punch-34-150-175-108.loca.lt and if you checked it, it will look like  https://somenumber.gradio.app.

That is where we can access our user interface. Just click it and it will open a new tab and we can start generating images.

After clicking it a new tab will open with the WebUI where we can start generating images.

Access the Web UI by Automatic1111
Access the Web UI by Automatic1111

Where Are Images Stored in Google Drive

Images are stored in MyDrive > sd > stable-diffusion > outputs.

Fast Stable Diffusion by AUTOMATIC1111 Outputs
Fast Stable Diffusion by AUTOMATIC1111 Outputs

In our case, we just generated two images using the txt2img feature, and they are stored in MyDrive > sd > stable-diffusion > outputs > txt2img-images:

Fast Stable Diffusion by AUTOMATIC1111 txt2img Outputs
Fast Stable Diffusion by AUTOMATIC1111 txt2img Outputs

Troubleshooting

Update Repo

You may have seen the Update_repo checkbox. You’ll check this when you want to update your installation. The notebook is constantly updating with new features, this is an easy way to update your files to be up to date.

Conclusion

In this tutorial we covered how to set up Fast Stable Diffusion WebUI by AUTOMATIC1111 to create AI generated art using an intuitive web interface. As mentioned before, the WebUI offers you multiple useful and interesting features, and we recommend you check the original Github repository for more information on each of them.

With Stable Diffusion WebUI by AUTOMATIC1111 you’ll be able to have fun easily generating hundreds of images, to be sure you generate the right one for your prompt. You’ll conveniently have them stored in your Google Drive, and each image will have its’ settings stored in a file so you can share your settings, or tweak your settings and regenerate the image at a later date.

With ESRGAN you can upscale your images up to 4 times, and with GFPGAN you can fix distorted portraits. It also comes with with the Image-to-Image feature, where you can upload/draw an initial image to help guide Stable Diffusion to the desired result – as well as inpainting and outpainting for even more control.

And you can also load your own models during the setup, which means you can use the WebUI with your custom fine-tuned Stable Diffusion models.

If you encountered any issues or have any questions, please feel free to leave a comment and we’ll get back to you as soon as possible.

Very Useful Resources

  • TheLastBen Github Repository – the repository from where we’ve used our Google Colab notebook. It’s an implementation of the AUTOMATIC1111 Stable Diffusion WebUI implementation of Stable Diffusion. In this repository there are implementations of other Stable Diffusion variants, as well. Such as Stable Diffusion WebUI by hlky, which is another very popular implementation, as well as a variant of DreamBooth (for fine-tuning Stable Diffusion), which we cover in a separate tutorial.
  • AUTOMATIC1111 Github Repository – this is the repository for Stable Diffusion WebUI by AUTOMATIC1111. It’s the most popular Stable Diffusion implementation at the time of writing.
    • Feature Documentation Wiki – This section is an excellent and comprehensive presentation of the WebUI’s features, along with explanations and examples.
  • Lexica.art, OpenArt.ai, Kre.ai – these are some excellent search engines that allow you to search through millions of Stable Diffusion images and their prompts. They are invaluable for inspiration. I highly recommend them.
  • 1
    Google Colab is a cloud-based Jupyter Notebook. Jupyter Notebooks are, in simple terms, interactive coding environments. Think of them as documents that allow you to write and execute code all in one place.

    Google Colab is a service that provides free Jupyter Notebooks that are run on Google’s servers. This means that you can use Google Colab to write and execute code without having to download or install anything on your own computer.

    Google Colab is a great tool for data scientists and machine learning engineers as it allows you to prototype and experiment with your code in a fast and convenient way. Additionally, Colab is perfect for sharing your work with others, as you can simply share the link to your notebook and anyone can view and execute the code.
1 Shares:
Subscribe
Notify of
guest
Receive notifications when your comment receives a reply. (Optional)
Your username will link to your website. (Optional)

68 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Michael E
Michael E
2 months ago

You can also run with no setup at canva, see texttoimage.app

chibi_panda
chibi_panda
2 months ago

is it possible to host a persistent version of the webui on the huggingface spaces as it says?

Said S.
Said S.
2 months ago

you have forgotten to download the model file and place it in that directory

I did this but it’s not working, I created a folder called AI and a folder called models inside that folder and put the file inside that called sd-v1-4.ckpt.

the directory name doesn’t match, such as if it’s called /AI/Models instead of /ai/models

I did AI/models just as this tutorial but I still have this error.

What did I do wrong?

Tomer
Tomer
2 months ago
Reply to  EdXD

I get the same kind of error:
FileNotFoundError: [Errno 2] No such file or directory: ‘/content/models/sd-v1-4.ckpt’
I attached the error and my drive folder.
https://ibb.co/ZfJbbmD

2022-09-19 15.44.10 colab.research.google.com 590b5c0b1e1b.png
Tomer
Tomer
2 months ago
Reply to  EdXD

It is checked, it even asks for permission to access google drive as your instructions state. Still, it gives this error for some reason.

2022-09-20 08.21.21 colab.research.google.com 4e7c0f1370f8.png
nawnie
nawnie
2 months ago
Reply to  Tomer

capitalize /Models 😀

Tomer
Tomer
2 months ago
Reply to  EdXD

It is checked (even asks for GDrive permission) and I still receive the error

2022-09-20 08.21.21 colab.research.google.com 4e7c0f1370f8.png
Andrew
Andrew
2 months ago

Thank you for this fantastic guide! As someone who struggled briefly with Method 2: Step 3, I just want to mention, in case it’s helpful, that there appears to be a subtle autocorrect in the given code that causes an error when it’s copied verbatim. In the second line where it currently says “– output”, I believe it should be “–output”. I’m sure that this is merely a word-processing glitch, but hopefully a note here helps save any future head-scratching

Bardic
Bardic
2 months ago

Thanks for the helpful tutorial. A couple of points/questions:

1) I think there’s a typo in the code box:

!curl -L https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt – output "/content/drive/MyDrive/AI/models/sd-v1-4.ckpt" -H "Authorization: Bearer YOUR_ACCESS_TOKEN"

When I try to run this, it suggests that ‘– output’ is required rather than with just one dash.

2) I’m a total noob at this stuff, so probably a noob question: do I have to run the whole script each time I want to start using it, including the setup stage? It seems like it from my testing so far, but that seems counterintuitive to this noob, so maybe I’m missing something.

3) I’m running into an issue where the finished images aren’t showing up in the WebUI if I choose a batch number of higher than one. The processing finishes, which I can see on the Colab page, and the generated images show up in my Google Drive, but the image (and output) panes of the WebUI just keep on showing their ‘loading/working’ animation indefinitely. If I reload the page and turn the batch back down to 1, it works again. Any idea what I’m doing wrong?

Thanks again!

Jadrian
Jadrian
2 months ago
Reply to  Bardic

I, too, am seeing a bug where images fail to appear, but this is usually when I try upscaling. The first upscale works fine, but all subsequent upscales fail to load the resulting image, and also keep the Colab session resources marked as endlessly “Busy”.

Bardic
Bardic
2 months ago

…and now I think I know why the code I mentioned in 1) is messed up: this site software seems to be automatically converting two dashes into one, which also happened in my comment. 😛

To spell it out, then: the script seems to want two dashes in front of ‘output’, and there’s only one in front of it in the code block.

HekFar
HekFar
2 months ago

Hi EdXD,
My issue is that I wanted to upload the full ema checkpoint, after I uploaded to my google drive, I change the code in step 3 to –ckpt ‘{models_path}/sd-v1-4-full-ema.ckpt’ \
the screenshot shows that it stops running after “eps-prediction mode” leaving a ^C
and never loads the url for UI. Is it possible to use the full-ema.ckpt? If so could you help me fix the code to reflect the proper path?

Screenshot 2022-09-18 174844.png
Mr-J-Photoshooting
2 months ago

Thanks for this tuto, it seems the part where you need to add code to mount and download models aren’t necessary anymore, there is a part in the note book where you only need to put your token.

I’m still getting familiar with it, less convenient than the set-up i used before (but at least i don’t kill my computer^^) . Main issue is, when images are finished to be generated, the ui don’t show images, and seems to still be running, even in i see in notebook that nothing is running anymore. So between each generation i need to reload the page.

Mr-J-Photoshooting
2 months ago

Ok so it seems i can’t use it any more, i didn’t change anything, but when i tried to launch it to start a new creation session, i got this, i don’t find any way to start the notebook
No CUDA GPUs are available

Look like, notebook solution isn’t so free, google ask us to pay to be able to run it as we want i think.

Billions
Billions
2 months ago

Hey, great, easy to follow instructions! I got it working, but you have a little error in the curl function block in Method 2, step 3 where you have ” – output”
when it should be “–output”

Aside from that, my results are generally slower and less accurate than what I’m getting from Huggingface’s SD demo and even the Dawn AI app for ios. maybe it’s a GPU limitation of using colab or some setting that I have yet to properly tweak. also haven’t been able to get img2img working.
Anyway, I hope my little note helps some other noobs like myself.
Thanks!

Lucas E
Lucas E
2 months ago

I’m getting an error in Google collab during the final step,. I don’t really know how to address this, I’m a pleb!

This is the error it spits:

ImportError: this version of pandas is incompatible with numpy < 1.20.3
your numpy version is 1.19.2.
Please upgrade numpy to >= 1.20.3 to use this pandas version

TheUntrainedEye
TheUntrainedEye
2 months ago
Reply to  EdXD

I’m just commenting to say that I’m also running into this problem. And when I run “%pip install numpy” I get toldRequirement already satisfied: numpy in /usr/local/lib/python3.8/site-packages (1.23.3)
, while when I run “import numpyprint(numpy.version.version)” I get told 1.21.6, which only confuses me even more. I think though that the install package *is* only installing numpy 1.19.2, judging by how in the “1.1 Download repo and install” step has this in its output:Package numpy-base conflicts for:numpy-basemkl_fft -> numpy[version=’>=1.16, numpy-base==1.19.2[build=’py38hfa32c7d_0|py38h4c65ebe_1′]numpy -> numpy-base==1.19.2[build=’py38hfa32c7d_0|py38h4c65ebe_1′]torchvision -> numpy[version=’>=1.11′] -> numpy-base==1.19.2[build=’py38hfa32c7d_0|py38h4c65ebe_1′]mkl_random -> numpy[version=’>=1.16, numpy-base==1.19.2[build=’py38hfa32c7d_0|py38h4c65ebe_1′]
andPackage libgfortran-ng conflicts for:libgfortran-ngnumpy -> numpy-base==1.19.2=py38h4c65ebe_1 -> libgfortran-ngnumpy-base -> libgfortran-ng
This suggests that the error isn’t imaginary, the published version of the notebook really is only installing Numpy 1.19.2.

dispari
dispari
2 months ago
Reply to  EdXD

Same here :-/

Wim
Wim
2 months ago
Reply to  EdXD

Thanks,just install it with the numpy fix, and it works

jonathan
jonathan
2 months ago
Reply to  Lucas E

Did you manage to solve the problem of: Import: this version offer pandas is incompatible with numpy <1.20.3 ??? I am going crazy !!!
Your numpy version IS 1.19.2.
PLEASE UPGRESS NUMPY TO> = 1.20.3 TO US THIS PANDAS VERSION

jonathan
jonathan
2 months ago
Reply to  EdXD

perfect, it worked thank you very much, you have saved my life

Jadrian
Jadrian
2 months ago

Joining in with some other comments with a bit of findings. It seems the upscale functionality may have a bug of some kind. It works to upscale one image after loading up the Colab, but then fails to upscale others. I can see that the Colab session shows as busy when this is happening, as though there is still an upscale process running amok on the GPU session. I haven’t yet figured out what is holding up that process or how to move beyond it, but wanted to chime in that I am experiencing it, too. What a fantastic write-up, though, thanks for helping me get this running!

dispari
dispari
2 months ago

Hello, there is a way to stop batch processing once it is started (for corrections in prompt etc)?

Blue_Lotus
Blue_Lotus
2 months ago

OMG this might as well be written in Russian. I can’t even get the file to upload to drive!
All I want to do is make pretty art and not need to be some sort of code reading genius

Vic
Vic
1 month ago

Hi.I got this error when itried to start stable diffusion
Traceback (most recent call last):
File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/webui.py”, line 8, in <module>
from modules.paths import script_path
File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/modules/paths.py”, line 16, in <module>

assert sd_path is not None, “Couldn’t find Stable Diffusion in any of: ” + str(possible_sd_paths)

Assertion error:could’t find stable Diffusion in any of.[contentgdrivemydrivesd stable diffusion

    assert sd_path is not None, "Couldn't find Stable Diffusion in any of: " + str(pos
ProblemWithPickles
ProblemWithPickles
1 month ago

I’m stuck just after “Start stable-diffusion” and nobody else appears to have my issue.
Full Log: https://pastebin.com/0tJ5HKF2

It looks like I’m getting three errors:

1- File “/usr/lib/python3.7/zipfile.py”, line 1325, in _RealGetContents
raise BadZipFile(“File is not a zip file”)
zipfile.BadZipFile: File is not a zip file

2- File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/modules/safe.py”, line 50, in find_class
raise pickle.UnpicklingError(f”global ‘{module}/{name}’ is forbidden”)
_pickle.UnpicklingError: global ‘torch._utils/_rebuild_tensor’ is forbidden

3- The file may be malicious, so the program is not going to read it.

I have no idea what any of that means.

dmalessio
dmalessio
1 month ago

Hi, my fast_stable_diffusion_AUTOMATIC1111.ipynb page looks different from the one you have in the tutorial and in fact the process fails

Screenshot 2022-10-10 at 01-31-25 Google Colaboratory.png
dmalessio
dmalessio
1 month ago
Reply to  EdXD

Thanks for the reply.
This is the error I get aftee launching “Start stable-diffusion”.

Screenshot 2022-10-10 at 01-38-36 Google Colaboratory.png
skye
skye
1 month ago
Reply to  EdXD

Hi. I’m facing the same problem and I’ve inserted my hugging face token

dmalessio
dmalessio
1 month ago
Reply to  EdXD

Yes, the token field was correctly compiled. Anyway, I deleted everything (from gdrive), made a new token, started the process from zero and then everything is running fine. @skye you can try the same process

Nathan
Nathan
1 month ago

Do I have to run each cell every time I close and re-open my browser?

dmalessio
dmalessio
1 month ago

Hi again! Sometime when I check “Update_repo” I get this error:

Updating acedbe6..be1596c
error: Your local changes to the following files would be overwritten by merge:
	webui.sh
Please commit your changes or stash them before you merge.
Aborting

There is something that I can do before reinstalling everithing? Thank you again

dmalessio
dmalessio
1 month ago
Reply to  EdXD

yep, it works when I do that. I wanted to see if there was a way to not have to reinstall everything. Thank you very much anyway!

John
John
1 month ago
Reply to  EdXD

Is there any way to get Google Colab to reconnect my Stable Diffusion connection without the need to run through all the steps of the tutorial? I can’t seem to get a reconnection after I am disconnected. I then have to connect my Google Drive and run all the steps again to get a new link to the WebUI.

Franklyn
Franklyn
1 month ago

Can anyone explain why I get a 502 Bad Page after a while? Is it timing out? And when i attempt to run it again from colab, there’s always a folder webui/ldm/ missing and a missing module taming.modules.vqvae?

John
John
1 month ago

This tutorial was amazing! If you made a video tutorial you would probably get a million views! I have one issue I can not figure out how to correct. How do I get the WebUI to work after I close my browser? I am starting the Stable Diffusion WebUI with both boxes unchecked and I can clicking the link it provides. I am taken to localtunnel and I click a box to launch the WebUI. Everything works fine. I save the URL. If I close Google Colab page or the URL with the WebUI for Stable the URL gives me an error or a 404 and I have to do all the steps again to make the WebUI work again. I tried saving a copy of the Google Colab Copy of fast_stable_diffusion_AUTOMATIC1111.ipynb on my Google Drive as well as selected SAVE from the file list in Google Colab but it doesn’t work if I close the browser and try and launch it again.

Marc
Marc
1 month ago

Getting 404 from .loca.lt the one day after the install

Marc
Marc
1 month ago
Reply to  EdXD

I had to mount again, Installing Requirements and start the SD again. Is it expected?

Marc
Marc
1 month ago
Reply to  EdXD

After the reinstall the local worked again. Didn’t tried the Gradio.

Marc
Marc
1 month ago
Reply to  EdXD

I had to mount again, Installing Requirements and start the SD again. Is it expected?

riccardo
riccardo
1 month ago

I’ve got this error since this morning:
Traceback (most recent call last):
File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/webui.py”, line 14, in <module>
import modules.extras
File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/modules/extras.py”, line 13, in <module>
from modules.ui import plaintext_to_html
File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/modules/ui.py”, line 45, in <module>
from modules.generation_parameters_copypaste import image_from_url_text

ImportError: cannot import name ‘image_from_url_text’ from ‘modules.generation_parameters_copypaste’ (/content/gdrive/MyDrive/sd/stable-diffusion-webui/modules/generation_parameters_copypaste.py)

Any clue?

Juan
Juan
1 month ago
Reply to  EdXD

I had the same problem, and this solved the issue, thanks.

DefaultDanver
DefaultDanver
26 days ago

I seem to be getting some sort of error on the last step
Traceback (most recent call last):
File “/content/gdrive/MyDrive/sd/stable-diffusion-webui/webui.py”, line 10, in <module>
from modules.paths import script_path
ModuleNotFoundError: No module named ‘modules’

Carson
Carson
19 days ago

It works sometimes but most of the time I get “You cannot currently connect to a GPU due to usage limits in Colab”. How do you fix this? What is a compute unit and how many images does one generate?

You May Also Like