[Bug]: txt2img or img2img generate a green pi

[inspire-boy]

What happened?

A bug happened!

Version

0.0.1 (Default)

What browsers are you seeing the problem on?

Chrome

Where are you running the webui?

Windows

Custom settings

Relauncher: Launching...
Found GFPGAN
Found RealESRGAN
Error loading LDSR:
Traceback (most recent call last):
  File "scripts/webui.py", line 384, in try_loading_LDSR
    LDSR = load_LDSR(checking=True) # TODO: Should try to load both models before giving up
  File "scripts/webui.py", line 299, in load_LDSR
    raise Exception("LDSR model not found at path "+model_path)
Exception: LDSR model not found at path ./src/latent-diffusion\experiments/pretrained_models\model.ckpt

Loading model from models/ldm/stable-diffusion-v1/model.ckpt
Global Step: 470000
LatentDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.52 M params.
making attention of type 'vanilla' with 512 in_channels
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
making attention of type 'vanilla' with 512 in_channels
C:\Users\pc\.conda\envs\ldm\lib\site-packages\gradio\deprecation.py:43: UserWarning: You have unused kwarg parameters in File, please remove them: {'source': 'upload'}
  warnings.warn(
Running on local URL:  http://localhost:7860/

To create a public link, set `share=True` in `launch()`.
Iteration: 1/1[MemMon] Recording max memory usage...


100%|████████████████████████████████████████████████████████████████████████████████| 50/50 [02:30<00:00,  3.01s/it]
[MemMon] Stopped recording.

[MemMon] Recording max memory usage...
Iteration: 1/1

Warning: Subprompt weights add up to zero. Discarding and using even weights instead.
100%|████████████████████████████████████████████████████████████████████████████████| 38/38 [01:51<00:00,  2.94s/it]
[MemMon] Stopped recording.

Relevant log output

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[taylor85345]

A similar issue was posted on the CompVis stable-diffusion repository. The solution seems to be setting --precision=full.

This can be accomplished by launching webui with python scripts/webui.py --no-half --precision=full, or by modifying the additional_arguments section of your relauncher.py:

additional_arguments = "--no-half --precision=full "

I was having the same issue, and this worked for me.

EDIT: Note that this does increase VRAM usage, so you may need to set optimized mode on a system with limited VRAM.

[inspire-boy]

Thank you very much, I change my way because I am 6G vram.

[Zenahr]

@taylor85345 is there a way to set no-half and full-precision when working with it programmatically?

[taylor85345]

@taylor85345 is there a way to set no-half and full-precision when working with it programmatically?

@Zenahr Can you tell me how you are using it? I'm fairly new to all this too, and not really a programmer, but I might have an idea depending on what you mean.

[Zenahr]

@taylor85345 I switched to the official fork as that one allows me to set the flags I need to.