Automatic1111

#6
by Edmo - opened

So you can't use this model in Automatic1111?

no. was going to try implementing this myself but i have yet to be approved. heard it works on comfyUI

no. was going to try implementing this myself but i have yet to be approved. heard it works on comfyUI

Thanks for the heads up, it's working on comfy

no. was going to try implementing this myself but i have yet to be approved. heard it works on comfyUI

Yep works without any probs

no. was going to try implementing this myself but i have yet to be approved. heard it works on comfyUI

Yep works without any probs

how? Im getting with comfy

making attention of type 'vanilla-xformers' with 512 in_channels
building MemoryEfficientAttnBlock with 512 in_channels...
Traceback (most recent call last):
  File "/stable-diffusion/execution.py", line 184, in execute
    executed += recursive_execute(self.server, prompt, self.outputs, x, extra_data)
  File "/stable-diffusion/execution.py", line 60, in recursive_execute
    executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
  File "/stable-diffusion/execution.py", line 60, in recursive_execute
    executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
  File "/stable-diffusion/execution.py", line 60, in recursive_execute
    executed += recursive_execute(server, prompt, outputs, input_unique_id, extra_data)
  File "/stable-diffusion/execution.py", line 69, in recursive_execute
    outputs[unique_id] = getattr(obj, obj.FUNCTION)(**input_data_all)
  File "/stable-diffusion/nodes.py", line 220, in load_checkpoint
    out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths("embeddings"))
  File "/stable-diffusion/comfy/sd.py", line 921, in load_checkpoint_guess_config
    if len(sd['model.diffusion_model.input_blocks.1.1.proj_in.weight'].shape) == 2:
KeyError: 'model.diffusion_model.input_blocks.1.1.proj_in.weight'
This comment has been hidden

no. was going to try implementing this myself but i have yet to be approved. heard it works on comfyUI

Is there any way to make it run on a11111? I imported config, but it still doesn't work properly

no. was going to try implementing this myself but i have yet to be approved. heard it works on comfyUI

Is there any way to make it run on a11111? I imported config, but it still doesn't work properly

U might check out the kardinsky extension for auto1111 and program a similar ext for sdxl but I recommend to use comfy

Comfy seems more complex.

This is how you can use it with SD.Next / Vlad's fork of A1111: https://github.com/vladmandic/automatic/wiki/Diffusers#sd-xl-notes

This is how you can use it with SD.Next / Vlad's fork of A1111: https://github.com/vladmandic/automatic/wiki/Diffusers#sd-xl-notes

In my opinion it's better to run it on comfy because u can't still use high Res fix on Vlads automatic. In comfy I have a separate node for this and it's a massive quality improvement.

You can be sure: the developers of auto1111 are implementing sdxl as we speak.

In my opinion it's better to run it on comfy because u can't still use high Res fix on Vlads automatic. In comfy I have a separate node for this and it's a massive quality improvement.

How do ? which node? what's the flow please

As a heads-up for anyone following this thread, support was merged into A111 today with version 1.5.0 of the UI

Sign up or log in to comment