search

AUTOMATIC1111 / stable-diffusion-webui

Stable Diffusion web UI
GNU Affero General Public License v3.0
136.87k stars 26.05k forks source link

[Bug]: Output image does not display on latest commit #8904

Open psykokwak-com opened 1 year ago

psykokwak-com commented 1 year ago

Is there an existing issue for this?

What happened?

I try to do a simple txt2img generation. The firt one works well. The next one does not display the final output image if I change the output resolution. The previous image still remains.

Steps to reproduce the problem

  1. Launch UI
  2. Write anything you want in the prompt
  3. Click on generate
  4. Wait until the end of image generation
  5. change output resolution to something else (1024x1024)
  6. Click on generate

What should have happened?

The intermediate generated images are good but it switch to the previous generated final image at the end.

Commit where the problem happens

009bc9f534a4f6d19ece5b5dafe3421421085fb1

What platforms do you use to access the UI ?

Windows

What browsers do you use to access the UI ?

Google Chrome

Command Line Arguments

--xformers --api --gradio-img2img-tool color-sketch

List of extensions

sd-webui-3d-open-pose-editor sd-webui-aspect-ratio-helper sd-webui-controlnet

Console logs

venv "C:\Users\Jyce\Desktop\stable-diffusion-webui\venv\Scripts\Python.exe"
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug  1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Commit hash: 009bc9f534a4f6d19ece5b5dafe3421421085fb1
Installing requirements for Web UI

Launching Web UI with arguments: --xformers --api --gradio-img2img-tool color-sketch
Loading weights [f93e6a50ac] from C:\Users\Jyce\Desktop\stable-diffusion-webui\models\Stable-diffusion\uberRealisticPornMerge_urpmv13.safetensors
Creating model from config: C:\Users\Jyce\Desktop\stable-diffusion-webui\configs\v1-inference.yaml
LatentDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.52 M params.
Applying xformers cross attention optimization.
Textual inversion embeddings loaded(0):
Model loaded in 2.6s (load weights from disk: 0.1s, create model: 0.3s, apply weights to model: 0.6s, apply half(): 0.4s, move model to device: 0.4s, load textual inversion embeddings: 0.7s).
Running on local URL:  http://127.0.0.1:7860

To create a public link, set `share=True` in `launch()`.
Startup time: 7.1s (import gradio: 1.0s, import ldm: 0.6s, other imports: 1.1s, list extensions: 0.4s, load scripts: 0.8s, load SD checkpoint: 2.8s, create ui: 0.2s).
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  8.51it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 11.30it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 10.02it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 10.55it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 10.21it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 11.00it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:09<00:00,  2.16it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:09<00:00,  2.20it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 11.45it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 11.45it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  9.85it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 10.93it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00,  9.94it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 10.93it/s]
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:09<00:00,  2.21it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:09<00:00,  2.21it/s]
Total progress: 100%|██████████████████████████████████████████████████████████████████| 20/20 [00:09<00:00,  2.19it/s]

Additional information

No response

Wevann commented 1 year ago

I have same the problem T__T

wmooring commented 1 year ago

I'm having a similar issue, the thumbnails for the generated images are blank even though they get written to disk, it's also showing blank thumbnails for the latest ones in the Image Browser extension tab.

Golevka2001 commented 1 year ago

The same problem, but it has been solved after reinstalling.

I'm still trying to figure out where this problem originated. But it is certain that the project itself is not problematic.

Currently, it can be determined that it is not related to the models or launch argument, but may be due to extensions or setting conflicts. I am still testing.

Golevka2001 commented 1 year ago

Got it.

After reinstalling, I met this issue again.

I deleted a few files and folders in ./venv/Lib/site-packages. (What should be deleted depends on when you encounter this problem.)

Now the output images appear again.

What extensions did I install

wmooring commented 1 year ago

I'm still having the issue even after a reinstall and disabling all the non-default extensions.

Is it related to general recent issues like these?

https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/8987

https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/9012

(Also, I tested in both Firefox and Chrome and have the same issue)

Golevka2001 commented 1 year ago

I'm still having the issue even after a reinstall and disabling all the non-default extensions.

Is it related to general recent issues like these?

8987

9012

(Also, I tested in both Firefox and Chrome and have the same issue)

No... I am also using the latest version, but have not encountered the problem in the two issues you mentioned.

The problem of this issue has nothing to do with browsers(I also tested in different ones). I think it's a conflict caused by some extensions, although I still can't explain it clearly🥲

wmooring commented 1 year ago

This is the full extent of my extensions list right now:

Capture

Unless you're saying some of the built-in ones need to be disabled too?

(Also, if I do "Inspect Element" on the area where the generated image thumbnails are supposed to be, I can see the links to the local image files, and opening those in another browser tab works fine, so the images exist, they just aren't being displayed properly as thumbnails for some reason)

Golevka2001 commented 1 year ago

I don't know...

It just return to normal after some packages(installed by some extensions) and extensions are removed.

Of course not the built-in extensions.

wmooring commented 1 year ago

I can do another fresh install I guess, last time I reinstalled all my extensions before disabling them and then removing their folders

psykokwak-com commented 1 year ago

Retried today and still have the issue.

wmooring commented 1 year ago

I was able to fix it on a fresh install with nothing else loaded, but I found out that once the Image Browser extension is loaded, I have a new issue with nothing on the Settings page being able to load on desktop Firefox or either of my mobile browsers: https://github.com/AlUlkesh/stable-diffusion-webui-images-browser/issues/100#issuecomment-1485897571

wmooring commented 1 year ago

The original issue with blank previews after generating images returned, and after some testing it seems like the issue is at least partly affected by the "Images filename pattern" setting.

If I use my desired filename pattern of:

[steps]_[sampler]_[seed]_c[cfg]_AD1.0_[prompt_spaces]

...I'll get blank panels in the results. If I remove the period (so just "AD10" for that part) and reload the browser tab, then the images will be shown normally after generating. At first I thought it was just the presence of any period and some part of the UI was being dumb and mistaking anything after that for the file extension, but it turns out I can put periods elsewhere in the pattern, or with different text around it, and it works most of the time.

It seems to be specifically an underscore followed by the the letters "AD" (or "ad" lowercase) then by a single number and a period that hits the bug. Baffled, I tried other samplers and models, and it seems to be consistent and have nothing to do with the Analog Diffusion model specifically. I imagine there are other strings that would hit the same bug, I just didn't happen to stumble across any.

_AD1.

/\ As the full filename pattern is the shortest string I've gotten the bug to trigger on.

Can anyone else test it, to see if they get the same results? This is happening across both Firefox and Chrome. Just set that as your filename pattern, generate some images and see if you get blank previews. If you do, set a different filename pattern, hit Apply Settings and then reload your browser tab (the Reload UI button doesn't seem to affect it), and see if you can see new generations.

Gepage commented 1 year ago

Solved. The problem is the "translate of prompts". Disable extension. Delete files with data of installed extension (uninstal extension doesn't work). Delete this files and start with "webui --update-check"

stable-diffusion-webui/venv/bin/ chardetect get_objgraph pathos_connect pox translate unidecode docx2txt nltk portpicker ppserver undill

/stable-diffusion-webui/venv/lib/python3.10/site-packages/ cffi hstspreload pathos-0.3.0.dist-info cffi-1.15.1.dist-info hstspreload-2023.1.1.dist-info pox chardet httpcore pox-0.3.2.dist-info chardet-3.0.4.dist-info httpcore-0.9.1.dist-info pp cryptography httpx ppft cryptography-40.0.1.dist-info httpx-0.13.3.dist-info ppft-1.7.6.6.dist-info dill hyperframe pycparser dill-0.3.6.dist-info hyperframe-5.2.0.dist-info pycparser-2.21.dist-info docx2txt joblib PyExecJS-1.5.1.dist-info docx2txt-0.8.dist-info joblib-1.2.0.dist-info pygoogletranslation execjs lru_dict-1.1.8.dist-info pygoogletranslation-2.0.6.dist-info googletrans lxml PyPDF2 googletrans-4.0.0rc1.dist-info lxml-4.9.2.dist-info pypdf2-3.0.1.dist-info h11 _multiprocess requests h11-0.9.0.dist-info multiprocess requests-2.25.1.dist-info h2 multiprocess-0.70.14.dist-info translators h2-3.2.0.dist-info nltk translators-5.6.3.dist-info hpack nltk-3.8.1.dist-info unidecode hpack-3.0.0.dist-info pathos Unidecode-1.3.6.dist-info

That are the files of MY problem...

codyrchildress commented 1 year ago

I am getting this issue after a fresh install, without ever installing "translate of prompts".

Gepage commented 1 year ago

I am getting this issue after a fresh install, without ever installing "translate of prompts".

Look in for files modified since the installation of SD. On linux they may be persistent in the ~/.local directory (name in my language, python settings). I don't touch any code to solve. I don't know in windows.

psykokwak-com commented 1 year ago

What about this issue ? I still have it on a fresh update.

psykokwak-com commented 1 year ago

Ok. It seems that deleting the "venv" directory fixes the issue.

feixiaomi commented 1 year ago

thx!

karry1155 commented 12 months ago

I'm having a similar issue, the thumbnails for the generated images are blank even though they get written to disk, it's also showing blank thumbnails for the latest ones in the Image Browser extension tab.

exactly, I can click the file buttion to open the finder, the photos are in the folder, but not display on the WebUI

karry1155 commented 11 months ago

After conpletely reintall the hole system, I finally fix the bug!

000ev commented 10 months ago

this is pretty frustrating. no images still.

Crosius1 commented 10 months ago

Same problem. Tried reinstalling everything and changing optimization settings. Tried other browsers and changing antivirus settings and disabling extensions. Regardless of everything, when some image resolution threshold or sampling steps are exceeded, a "connection errored out" error appears. My build has nothing to do with public internet connection. Now max is 3 low resolution images with standard sampling steps.

Before this problem, I used to be able to create 9 with sampling steps of 30 and resolutions well above the recommended ones.

Now the web interface has become close to useless for me and I don't understand what's wrong at all.

The images are there in the preview and even saved, but they are not in the web interface in any form.

Upd. Ok, while experimenting I disabled Kaspersky antivirus and THAT fixed the issue. I don't know why it didn't work when I added the generation page to the exceptions. Looks like more experimentation is required.

Gilded-Edge commented 10 months ago

Same problem. Tried reinstalling everything and changing optimization settings. Tried other browsers and changing antivirus settings and disabling extensions. Regardless of everything, when some image resolution threshold or sampling steps are exceeded, a "connection errored out" error appears. My build has nothing to do with public internet connection. Now max is 3 low resolution images with standard sampling steps.

Before this problem, I used to be able to create 9 with sampling steps of 30 and resolutions well above the recommended ones.

Now the web interface has become close to useless for me and I don't understand what's wrong at all.

The images are there in the preview and even saved, but they are not in the web interface in any form.

Upd. Ok, while experimenting I disabled Kaspersky antivirus and THAT fixed the issue. I don't know why it didn't work when I added the generation page to the exceptions. Looks like more experimentation is required.

Thank you so much! this was driving me mad!

michaelrinderle commented 8 months ago

@Crosius1 thanks, it was kaspersky for me, too. I just added an exclusion to auto1111 folder.