lllyasviel / lllyasviel/stable-diffusion-webui-forge
[Bug]: Highres-fix button crashes
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 13k
- Forks
- 1.7k
- PR merge metrics
- No merged PRs in 30d
Description
Checklist
- The issue exists after disabling all extensions
- The issue exists on a clean installation of webui
- The issue is caused by an extension, but I believe it is caused by a bug in the webui
- The issue exists in the current version of the webui
- The issue has not been reported before recently
- The issue has been reported before but has not been fixed yet
What happened?
Forge has this Highres fix button ✨ but for some reason it now started to crash when I try to use it on a generated image. It worked before, maybe some update went wrong? Highres-fix (checkmark tab) works however. The button was just handy so I don't have to run highres-fix everytime I try to generate an image with txt2img.
traceback (most recent call last): | 0/15 [00:00<?, ?it/s]
File "W:\forge.git\app\modules_forge\main_thread.py", line 37, in loop
task.work()
File "W:\forge.git\app\modules_forge\main_thread.py", line 26, in work
self.result = self.func(*self.args, **self.kwargs)
File "W:\forge.git\app\modules\txt2img.py", line 85, in txt2img_upscale_function
processed = processing.process_images(p)
File "W:\forge.git\app\modules\processing.py", line 752, in process_images
res = process_images_inner(p)
File "W:\forge.git\app\modules\processing.py", line 922, in process_images_inner
samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts)
File "W:\forge.git\app\modules\processing.py", line 1291, in sample
return self.sample_hr_pass(samples, decoded_samples, seeds, subseeds, subseed_strength, prompts)
File "W:\forge.git\app\modules\processing.py", line 1321, in sample_hr_pass
samples = torch.nn.functional.interpolate(samples, size=(target_height // opt_f, target_width // opt_f), mode=self.latent_scale_mode["mode"], antialias=self.latent_scale_mode["antialias"])
File "W:\automatic1111.git\app\venv\lib\site-packages\torch\nn\functional.py", line 3959, in interpolate
return torch._C._nn.upsample_bilinear2d(input, output_size, align_corners, scale_factors)
RuntimeError: "upsample_bilinear2d_out_frame" not implemented for 'BFloat16'
"upsample_bilinear2d_out_frame" not implemented for 'BFloat16'
This seems to be a pytorch issue however it started to appear in Forge after updating it to the latest version.
https://github.com/pytorch/pytorch/issues/88536
Steps to reproduce the problem
- Generate image in txt2img
- Hit the star button under the image
What should have happened?
Not crash
What browsers do you use to access the UI ?
Mozilla Firefox
Sysinfo
...
Console logs
...
Additional information
No response
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start in app/modules/txt2img.py at txt2img_upscale_function, then follow the traceback into app/modules/processing.py and the sample_hr_pass path. Reproduce the crash by generating an image in txt2img and pressing the star button; done means the high-res fix action completes without the reported BFloat16 interpolation error.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Stale
- Clarity
- Clearly specified
- Newbie friendliness
- 45/100