Unable to generate results even after matching sample settings

[Nesuhst]

Pulled from sample on extension page:

My attempt with matching settings, including seed, steps, size, and checkpoint on txt2img:

The ouput picture never seems to relate to the input image in any of the options.

I'm using forge on the latest version. On windows 10. I'd followed the instructions to create a "Unet" folder in the forge > webui > models > folder where I'd put the ic-light safetensors, as there wasn't one yet.

running on a 980ti 6gb

for some reason, at the start of the generation, the input ic-light image is shown but then fades completely to an unrelated image (in txt2img at least, but I think it's similar in img2img)

[Dreamz-Dziner]

For me also nothing seems to be working even after doing exactly the same things mentioned in the repo page. The resulting image does not match with the foreground uploaded. Not sure what went wrong.

[huchenlei]

Can you attach full console log? Thanks

[Dreamz-Dziner]

Can you attach full console log? Thanks

There is no error message as such in the console which is the confusing part.

[Dreamz-Dziner]

Here is what I get in the UI section

[CasualDev242]

Multiple reasons this can be caused, as it happened to me and I finally fixed it. I put both the models instructed to in my models/unet directory, and it had no effect. Then I noticed this in the log: Cannot access gated repo for url https://huggingface.co/briaai/RMBG-1.4/resolve/main/model.safetensors. Access to model briaai/RMBG-1.4 is restricted. You must be authenticated to access it.

You have to create a hugging face account, and download that file manually. Put it in the same unet folder. Now it should work. Also worth noting that it didn't work well on my SDXL checkpoint, but does work on SD1,5 checkpoints.

I still can't remove the background though for some reason.

Also when following the instructions of cfg scale 2 and 1.0 denoise, it works but it does subtly change the foreground character such as mild clothes changes and mild facial feature changes.

[Ignacio50]

Same here, this is what I got on the console log ` *** Error running process_before_every_sampling: E:\stable-diffusion-webui\extensions\sd-forge-ic-light\scripts\forge_ic_light.py Traceback (most recent call last): File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\utils_errors.py", line 304, in hf_raise_for_status response.raise_for_status() File "E:\stable-diffusion-webui\venv\lib\site-packages\requests\models.py", line 1021, in raise_for_status raise HTTPError(http_error_msg, response=self) requests.exceptions.HTTPError: 401 Client Error: Unauthorized for url: https://huggingface.co/briaai/RMBG-1.4/resolve/main/model.safetensors

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "E:\stable-diffusion-webui\modules\scripts.py", line 835, in process_before_every_sampling
    script.process_before_every_sampling(p, *script_args, **kwargs)
  File "E:\stable-diffusion-webui\extensions\sd-forge-ic-light\scripts\forge_ic_light.py", line 403, in process_before_every_sampling
    rmbg = BriaRMBG.from_pretrained("briaai/RMBG-1.4").to(device=device)
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\utils\_validators.py", line 118, in _inner_fn
    return fn(*args, **kwargs)
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\hub_mixin.py", line 277, in from_pretrained
    instance = cls._from_pretrained(
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\hub_mixin.py", line 492, in _from_pretrained
    model_file = hf_hub_download(
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\utils\_validators.py", line 118, in _inner_fn
    return fn(*args, **kwargs)
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\file_download.py", line 1403, in hf_hub_download
    raise head_call_error
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\file_download.py", line 1261, in hf_hub_download
    metadata = get_hf_file_metadata(
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\utils\_validators.py", line 118, in _inner_fn
    return fn(*args, **kwargs)
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\file_download.py", line 1667, in get_hf_file_metadata
    r = _request_wrapper(
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\file_download.py", line 385, in _request_wrapper
    response = _request_wrapper(
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\file_download.py", line 409, in _request_wrapper
    hf_raise_for_status(response)
  File "E:\stable-diffusion-webui\venv\lib\site-packages\huggingface_hub\utils\_errors.py", line 321, in hf_raise_for_status
    raise GatedRepoError(message, response) from e
huggingface_hub.utils._errors.GatedRepoError: 401 Client Error. 

Cannot access gated repo for url https://huggingface.co/briaai/RMBG-1.4/resolve/main/model.safetensors.
Access to model briaai/RMBG-1.4 is restricted. You must be authenticated to access it.

---

` Now I did like @CasualDev242 mentions, create a hugging face account, download the model.safetensors, put it in the same unet folder, and I still can't get it to work.

[quangchi]

Has anyone been able to fix this problem yet? I see a few people on YouTube have installed it (https://www.youtube.com/watch?v=WSBkgXnQBkU) . But I do not

[hiend0001]

Hi,quangchi What GPU are you using?

[Haoming02]

Hello, you may try out my fork in the meantime, which uses the same rembg package as stable-diffusion-webui-rembg that automatically downloads the models, among some other UI improvements.

[huchenlei]

Hello, you may try out my fork in the meantime, which uses the same rembg package as stable-diffusion-webui-rembg that automatically downloads the models, among some other UI improvements.

Thanks! Would you like your change merged here as well? I can add you as collaborator of this repo.

[huchenlei]

Amazing, you even implemented the detail restore that I planned to do a while ago. 👍

[Haoming02]

Well, I did make a few fundmental changes (eg. the model loading parts). But if you're fine with them, then merging is fine ig