Getting Error

[OlivioSarikas]

I'm getting this error: TypeError: LatentDiffusion.init() missing 3 required positional arguments: 'first_stage_config', 'cond_stage_config', and 'personalization_config'

I tried with a ckpt and safetensor model. neither seem to work for me

i'm on windows 11 with a 3080Ti

[comfyanonymous]

Can you give me the full traceback?

[OlivioSarikas]

got prompt
| 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
Traceback (most recent call last):
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 173, in execute
executed += recursive_execute(self.server, prompt, self.outputs, x, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 54, in recursive_execute
executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 54, in recursive_execute
executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 54, in recursive_execute
executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 62, in recursive_execute
outputs[unique_id] = getattr(obj, obj.FUNCTION)(**input_data_all)
File "C:\Users\Demo\Documents\ComfyUI\nodes.py", line 222, in load_checkpoint
out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=CheckpointLoader.embedding_directory)
File "C:\Users\Demo\Documents\ComfyUI\comfy\sd.py", line 788, in load_checkpoint_guess_config
model = instantiate_from_config(model_config)
File "C:\Users\Demo\AppData\Local\Programs\Python\Python310\lib\site-packages\ldm\util.py", line 93, in instantiate_from_config
return get_obj_from_str(config["target"])(**config.get("params", dict()), **kwargs)
TypeError: LatentDiffusion.init() missing 3 required positional arguments: 'first_stage_config', 'cond_stage_config', and 'personalization_config'

[comfyanonymous]

Can you try now with what I just pushed, I might have fixed it.

[OlivioSarikas]

Nope, same problem. That's why i asked you if we can do a call before i make my video about your UI

got prompt
Global Step: 840000
| 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
Traceback (most recent call last):
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 173, in execute
executed += recursive_execute(self.server, prompt, self.outputs, x, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 54, in recursive_execute
executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 54, in recursive_execute
executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 54, in recursive_execute
executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
File "C:\Users\Demo\Documents\ComfyUI\execution.py", line 62, in recursive_execute
outputs[unique_id] = getattr(obj, obj.FUNCTION)(**input_data_all)
File "C:\Users\Demo\Documents\ComfyUI\nodes.py", line 222, in load_checkpoint
out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=CheckpointLoader.embedding_directory)
File "C:\Users\Demo\Documents\ComfyUI\comfy\sd.py", line 788, in load_checkpoint_guess_config
model = instantiate_from_config(model_config)
File "C:\Users\Demo\AppData\Local\Programs\Python\Python310\lib\site-packages\ldm\util.py", line 93, in instantiate_from_config
return get_obj_from_str(config["target"])(**config.get("params", dict()), **kwargs)
TypeError: LatentDiffusion.init() missing 3 required positional arguments: 'first_stage_config', 'cond_stage_config', and 'personalization_config'

[comfyanonymous]

That's why i asked you if we can do a call before i make my video about your UI

I think this is the first time we interact and I don't remember someone asking me that.

The issue is that you have the ldm installed in your pip packages for some reason and it conflicts with the one in the ComfyUI directory.

Try: pip uninstall ldm

ComfyUI has some code to force it to look in the ComfyUI directory for my modified version of this library but apparently it doesn't work on your system.

[OlivioSarikas]

Hi, oh sorry, i thought you are "Said" form Facebook. My appologize for the mixup. Would you be able to assist me with this via Discord? Olivio#6552
Apparently my LDM isn't installed as i get this message:
WARNING: Skipping ldm as it is not installed.

[comfyanonymous]

ComfyUI now has a standalone portable build on the release page for windows/nvidia: https://github.com/comfyanonymous/ComfyUI/releases

Direct download link: https://github.com/comfyanonymous/ComfyUI/releases/download/latest/ComfyUI_windows_portable_nvidia_or_cpu.7z

That will probably fix your issue.