[Bug]: black screen on output

[lgwjames]

Checklist

• The issue has not been resolved by following the troubleshooting guide
• The issue exists on a clean installation of Fooocus
• The issue exists in the current version of Fooocus
• The issue has not been reported before recently
• The issue has been reported before but has not been fixed yet

What happened?

when trying to cartoon or cute photos with image prompt getting errors

Steps to reproduce the problem

man meets woman, Cutest cartoon illustration, empty background, cute, colorful, beautiful, art station, deviant art, pinterest in prompt

add photo prompt

What should have happened?

a good photo and not a blank black one

What browsers do you use to access Fooocus?

Mozilla Firefox

Where are you running Fooocus?

Locally

What operating system are you using?

mac os 14.5

Console logs

[Fooocus] Downloading control models ...
[Fooocus] Loading control models ...
extra clip vision: ['vision_model.embeddings.position_ids']
[Parameters] Sampler = dpmpp_2m_sde_gpu - karras
[Parameters] Steps = 60 - 30
[Fooocus] Initializing ...
[Fooocus] Loading models ...
Refiner unloaded.
[Fooocus] Processing prompts ...
[Wildcards] processing: __neg__
[Wildcards] Image compression
[Wildcards] processing: __neg__
[Wildcards] Out of Focus
[Wildcards] processing: __neg__
[Wildcards] Banding
[Wildcards] processing: __neg__
[Wildcards] tasteless
[Fooocus] Preparing Fooocus text #1 ...
[Prompt Expansion] man meets woman, Cutest cartoon illustration, empty background, cute, colorful, beautiful, art station, deviant art, pinterest, amazing, gorgeous, intricate, vibrant, highly detailed, very sharp focus, clear, artistic, fabulous, elegant, epic, stunning, full bright, brilliant, optimistic colors, glowing, shining atmosphere, radiant light, vivid, magic
[Fooocus] Preparing Fooocus text #2 ...
[Prompt Expansion] man meets woman, Cutest cartoon illustration, empty background, cute, colorful, beautiful, art station, deviant art, pinterest, highly detailed, coherent, sharp focus, rich bright colors, optimistic, futuristic, color, intricate, saturated, vivid, dramatic, romantic, determined, generous, cinematic, loving, pretty, inspiring, epic, stunning, creative, passionate
[Fooocus] Preparing Fooocus text #3 ...
[Prompt Expansion] man meets woman, Cutest cartoon illustration, empty background, cute, colorful, beautiful, art station, deviant art, pinterest, highly detailed, very sharp, focus, inspired, rich bright colors, optimistic, glowing, elegant, intricate, innocent, light, magical, pretty, illuminated, vivid, color, coherent, strong, complex, iconic, amazing composition
[Fooocus] Preparing Fooocus text #4 ...
[Prompt Expansion] man meets woman, Cutest cartoon illustration, empty background, cute, colorful, beautiful, art station, deviant art, pinterest, highly detailed, sharp focus, intricate, elegant, innocent, generous, lovely, magical, incredible, epic light, cinematic, stunning, coherent, amazing composition, creative, positive, joyful, unique, attractive, very inspirational, artistic
[Fooocus] Encoding positive #1 ...
[Fooocus] Encoding positive #2 ...
[Fooocus] Encoding positive #3 ...
[Fooocus] Encoding positive #4 ...
[Fooocus] Encoding negative #1 ...
[Fooocus] Encoding negative #2 ...
[Fooocus] Encoding negative #3 ...
[Fooocus] Encoding negative #4 ...
[Fooocus] Image processing ...
Requested to load CLIPVisionModelWithProjection
Loading 1 new model
Requested to load Resampler
Loading 1 new model
Requested to load To_KV
Loading 1 new model
[Fooocus Model Management] Moving model(s) has taken 0.14 seconds
[Parameters] Denoising Strength = 1.0
[Parameters] Initial Latent shape: Image Space (896, 1152)
Preparation time: 30.85 seconds
[Fooocus] Preparing task 1/4 ...
[Sampler] refiner_swap_method = joint
[Sampler] sigma_min = 0.0291671771556139, sigma_max = 14.614643096923828
Requested to load SDXL
Loading 1 new model
unload clone 5
[Fooocus Model Management] Moving model(s) has taken 4.02 seconds
  0%|                                                                                                                                                                 | 0/60 [00:00<?, ?it/s]/Users/Quad-Core/Fooocus/modules/core.py:257: RuntimeWarning: invalid value encountered in cast
  x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)
 23%|███████████████████████████████████▍                                                                                                                    | 14/60 [03:28<11:22, 14.85s/it]

Additional information

also another error from today

/Users/Quad-Core/Fooocus/modules/core.py:331: RuntimeWarning: invalid value encountered in cast
return [np.clip(255. * y.cpu().numpy(), 0, 255).astype(np.uint8) for y in x]

this is in my start up

[System ARGV] ['entry_with_update.py', '--theme', 'dark', '--all-in-fp16', '--attention-pytorch', '--disable-offload-from-vram', '--always-high-vram', '--gpu-device-id', '0', '--async-cuda-allocation', '--unet-in-fp16', '--vae-in-fp16', '--clip-in-fp16']
Python 3.11.5 (main, Sep 11 2023, 08:19:27) [Clang 14.0.6 ]
Fooocus version: 2.4.1
Set device to: 0
[Cleanup] Attempting to delete content of temp dir /var/folders/6l/xrkxgs657g7_vq50479x7lc40000gp/T/fooocus
[Cleanup] Cleanup successful
Total VRAM 65536 MB, total RAM 65536 MB
Forcing FP16.
Set vram state to: SHARED
Device: mps
VAE dtype: torch.float16
Using pytorch cross attention
Refiner unloaded.
IMPORTANT: You are using gradio version 3.41.2, however version 4.29.0 is available, please upgrade.

[mashb1t]

SDXL is prone to generating black images after VAE if there are issues in its internal calculations when using FP16 on some devices. As you're using the MPS backend there might be some issues with this combination. Please check if this can be solved by using --vae-in-fp32 as i can't reproduce the issues. You can also download and use a separate VAE with FP16 support, which might work more consistently.

EDIT: as of #2171 upgrading the OS did help.

See also:

• [Issue]: SDXL 1.0 Producing Black Images vladmandic/automatic#1858 (comment)
• [Bug]: Black screen on many outputs anywhere from 8 to 50+ steps #2747
• [Bug]: v2.30 Preset lightning, the resulting images are all black. #2580
• entirely black image generated #1184
• Black images OR very poor performance on M1 / Silicon MacBook Pro – "invalid value encountered in cast" #2171
• VAE problem? #1840
• Entirely black image generated on a Intel Mac book pro #1176

[lgwjames]

have updated as requested. python entry_with_update.py --theme dark --all-in-fp16 --attention-pytorch --disable-offload-from-vram --always-high-vram --gpu-device-id 0 --async-cuda-allocation --unet-in-fp16 --vae-in-fp32 --clip-in-fp16

just need to re download some files

[lgwjames]

so my time has gone from about 11/12 s/it to 14/15 so a small increase.

first image test

and now adding the image prompt

[lgwjames]

blank screen
no preview

[Fooocus Model Management] Moving model(s) has taken 2.90 seconds
0%| | 0/60 [00:00<?, ?it/s]/Users/Quad-Core/Fooocus/modules/core.py:257: RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)
17%|██████████████████████████ | 10/60 [02:27<12:15, 14.71s/it]

[lgwjames]

also on term cmd window different bars

ended as nothing

100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 60/60 [14:30<00:00, 14.50s/it]
/Users/Quad-Core/Fooocus/modules/core.py:331: RuntimeWarning: invalid value encountered in cast
return [np.clip(255. * y.cpu().numpy(), 0, 255).astype(np.uint8) for y in x]
[Fooocus] Saving image 1/1 to system ...
Image generated with private log at: /Volumes/1TB Caddy bay 5/2024-06-02/log.html
Generating and saving time: 877.44 seconds

[lgwjames]

remove base model to a new one
and straight away error

/Users/Quad-Core/Fooocus/modules/core.py:257: RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)

seems to be when i add a photo prompt so far

[lgwjames]

downgraded to 2.4.0

same error /Users/Quad-Core/Fooocus/modules/core.py:257: RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)

[lgwjames]

removing extra code and will report back

[lgwjames]

ok removed all code and still getting black screens.

/Users/Quad-Core/Fooocus/modules/core.py:257: RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)

tried all 7 preset and as soon as i put a image in prompt error

[mashb1t]

Please check out the MacOS support issue here and ask for help: #129

[mashb1t]

Please link to this issue in #129 to clarify with tother MacOS users.

[lgwjames]

today tested and was ok as soon with image prompt as i add my sister and bf, errors
Fooocus Model Management] Moving model(s) has taken 4.45 seconds
0%| | 0/60 [00:00<?, ?it/s]/Users/Quad-Core/Fooocus/modules/core.py:257: RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)

[lgwjames]

still errors, new install using 2.3.1
simple add 2 photos, left one me, right one generated one for face swop

prompt

Create a cool picture of a business man, age 40, in front of a London skyline with futuristic lighting, relaxed, serious, determined, color, dramatic light, detailed, sharp focus, intricate, very inspirational, innocent, attractive, sublime, pretty, fine detail, epic composition, elegant, perfect formal atmosphere, extremely stylish, creative, appealing, cheerful, beautiful, stunning

Loading 1 new model
[Fooocus Model Management] Moving model(s) has taken 4.57 seconds
0%| | 0/60 [00:00<?, ?it/s]huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...
To disable this warning, you can either:
- Avoid using tokenizers before the fork if possible
- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)
/Users/Quad-Core/Fooocus/modules/anisotropic.py:132: UserWarning: The operator 'aten::std_mean.correction' is not currently supported on the MPS backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at /Users/runner/work/_temp/anaconda/conda-bld/pytorch_1711403210267/work/aten/src/ATen/mps/MPSFallback.mm:13.)
s, m = torch.std_mean(g, dim=(1, 2, 3), keepdim=True)
/Users/Quad-Core/Fooocus/modules/core.py:255: RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.cpu().numpy().clip(0, 255).astype(np.uint8)
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 60/60 [17:01<00:00, 17.02s/it]
Requested to load AutoencoderKL
Loading 1 new model
[Fooocus Model Management] Moving model(s) has taken 0.18 seconds
/Users/Quad-Core/Fooocus/modules/core.py:329: RuntimeWarning: invalid value encountered in cast
return [np.clip(255. * y.cpu().numpy(), 0, 255).astype(np.uint8) for y in x]
Image generated with private log at: /Volumes/1TB Caddy bay 5/2024-07-27/log.html
Generating and saving time: 1031.29 seconds
Total time: 1104.49 seconds

[mashb1t]

Did you check out #129 / #1184?

[lgwjames]

I have really never had any problem with face swap, it only seems to be in the last three or four months that I've had problems with this software. i have a 12gb rtx 3060 on my desk and just waiting on my epgu case to test external should add, tested also on my bootcamp windows and same when i started here , i never had problems at all, everything just worked. Sent from Outlook for Android<https://aka.ms/AAb9ysg>

…

________________________________ From: Manuel Schmid ***@***.***> Sent: Saturday, July 27, 2024 12:29:27 PM To: lllyasviel/Fooocus ***@***.***> Cc: James ***@***.***>; Author ***@***.***> Subject: Re: [lllyasviel/Fooocus] [Bug]: black screen on output (Issue #3070) Did you check out #129<#129> / #1184<#1184>? — Reply to this email directly, view it on GitHub<#3070 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AIPK7J7ZVPDPKGIPJZ72OL3ZOOABPAVCNFSM6AAAAABI4PMD5SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENJUGEZDEMBRHA>. You are receiving this because you authored the thread.Message ID: ***@***.***>