How to add additional models and loras?

[martinenkoEduard]

How to add additional models and loras?

[WTFox1]

Download the lora/ the SAFETENSORS File [Like Realistic Vision V5.1 https://civitai.com/models/4201/realistic-vision-v51,.]

Drop that into your ...\Fooocus\models\checkpoints [or Lora].

It should then become an option as a Refiner or Model. I believe it needs to be SDLX or SD 1.5

[NicolasGoeddel]

Download the lora/ the SAFETENSORS File [Like Realistic Vision V5.1 https://civitai.com/models/4201/realistic-vision-v51,.]

Drop that into your ...\Fooocus\models\checkpoints [or Lora].

It should then become an option as a Refiner or Model. I believe it needs to be SDLX or SD 1.5

Can you please give me a better explanation on how to do that? I only get this error:

Traceback (most recent call last):
  File "/usr/lib/python3.10/threading.py", line 1016, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.10/threading.py", line 953, in run
    self._target(*self._args, **self._kwargs)
  File "/home/nicolas/git/AI/Fooocus/modules/async_worker.py", line 809, in worker
    pipeline.prepare_text_encoder(async_call=True)
  File "/home/nicolas/git/AI/Fooocus/fooocus_env/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/nicolas/git/AI/Fooocus/fooocus_env/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/nicolas/git/AI/Fooocus/modules/default_pipeline.py", line 210, in prepare_text_encoder
    assert_model_integrity()
  File "/home/nicolas/git/AI/Fooocus/fooocus_env/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/nicolas/git/AI/Fooocus/fooocus_env/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/nicolas/git/AI/Fooocus/modules/default_pipeline.py", line 53, in assert_model_integrity
    raise NotImplementedError(error_message)
NotImplementedError: You have selected base model other than SDXL. This is not supported yet.

[WTFox1]

You are only able to use SDXL as the Base Model at this point. You can use SDXL or SD1.5 as a Refiner.

NotImplementedError: You have selected base model other than SDXL. This is not supported yet.
this just tells you that you are attempting to use something other than an SDXL.

When you go to CivitAI, for example, you can see the type on the right of the Lora/Checkpoint. It's either SD or SDXL or SVD XT etc. You can use most SDXL as a Base model, but that's it for now.

The "Async Style" Lora you are trying to load is SD 1.5. Try putting that on as a Refiner with something like BluePencil (as BaseModel), and play around with the weight. That might work.

[NicolasGoeddel]

I was not able to find a hint about SDXL or not. I am very new to this stuff. I just followed your link and downloaded a model with the .safetensors extension, to be honest. 😅
I figured I can use the model as a refiner. I tried the default and the realistic preset as a base and it seems to work. However I do not know if that was the right thing to do.

However I will try to fiddle around a bit more. Thank you.

[samboa86]

I too came with the same question, read these replies, but I cannot select a "refined" model. it is greyed out cant select. I have the SDXL as base model but cant select the new 1.5 model as refined in order to use some LORAs that I guess can only be used with 1.5

[lllyasviel]

if refiner is greyed out then user is using Extreme Speed mode.
Extreme Speed mode cannot load complicated models to ensure the speed.