ComfyUI
ComfyUI copied to clipboard
comfyanonymous
Error when using cascade
Hi....... When I using cascade, error shows below. I'm sure my comfyui is the latest and my models put in the right place. It confuse me for a long time. Thank you everyone!
Error occurred when executing UNETLoader:
'conv_in.weight'
File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\nodes.py", line 814, in load_unet model = comfy.sd.load_unet(unet_path) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\comfy\sd.py", line 607, in load_unet model = load_unet_state_dict(sd) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\comfy\sd.py", line 580, in load_unet_state_dict model_config = model_detection.model_config_from_diffusers_unet(sd) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\comfy\model_detection.py", line 380, in model_config_from_diffusers_unet unet_config = unet_config_from_diffusers_unet(state_dict) File "D:\AI\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\comfy\model_detection.py", line 262, in unet_config_from_diffusers_unet match["model_channels"] = state_dict["conv_in.weight"].shape[0]
same problem
Any update on this? So far seems all my cascade workflows are having this issue as well. My bad, was using the wrong clip model for text encoding lol. My bad. Resolved by using correct clip model.
@GeekyGhost can you share your workflow please ? I'm having the same issue.
@GeekyGhost can you share your workflow please ? I'm having the same issue.
Sure, https://civitai.com/models/432889/geeky-ghost-all-in-one-evolved
@GeekyGhost can you share your workflow please ? I'm having the same issue.
Basically, needed to use full models instead of of the unet models, then used the original cascade clip model, and stage A for the VAE, but, you can use C for clip abd Stage B for VAE if you don't have them, it will throw an error, but still generate the image.
Hi....I'm not an export. What's the full model means? Thanks a lot!
oops, so the original models were just like unet models and needed a special loader, the full models are merges that made them useable in regular checkpoint loaders. The merged models work, the unet models were giving me issues, but there's always a work around lol