Closed FennelFetish closed 2 months ago
CrossAttentionPatch.init() got an unexpected keyword argument 'cond_alt' This error is due to a modification made to the CrossAttentionPatch of the plugin, specifically in:https://github.com/cubiq/ComfyUI_IPAdapter_plus/blob/0d0a7b3693baf8903fe2028ff218b557d619a93d/CrossAttentionPatch.py#L9
@FennelFetish This has been addressed in pull request https://github.com/siliconflow/onediff/pull/814 and is currently being progressed for merging. You can try it out using https://github.com/siliconflow/onediff/pull/814
## Install
### 1. onediff
git clone https://github.com/siliconflow/onediff.git
cd onediff
git checkout -b dev_ipa+isntantid
pip install -e .
### 2. onediff_comfy_nodes
cp -r onediff_comfy_nodes path/to/ComfyUI/custom_nodes/
Describe the bug
Exception when executing workflow in ComfyUI with OneDiff and IPAdapter. Happens with all the nodes: IPAdapter FaceID, IPAdapter Advanced, IPAdapter/IPAdapter Unified Loader... IPAdapter does work without OneDiff.
Your environment
ComfyUI_IPAdapter_plus commit
b257aecbedeb0c1367c6f67671afdeadd6e81fef
(2024-04-13)Kubuntu 22.04 LTS AMD Ryzen 9 5950X 16-Core Processor NVIDIA GeForce RTX 4090
OneDiff git commit id
d7e448773f7d666fc3ad865c46d448532614e8df
(2024-04-12)OneFlow version info
How To Reproduce
Import ComfyUI workflow from this image here: https://github.com/siliconflow/onediff/pull/732
The complete error message
Don't know if this line is part of the error:
Failed to update obj_entity in place. Exception: Not support <function CondFunc.__new__.<locals>.<lambda> at 0x7f07a456f010> with parameters Module: <module 'onediff_comfy_nodes.modules.sd_hijack_utils' from '/mnt/firlefanz/StableDiffusion/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/sd_hijack_utils.py'>
Additional context
patch_kwargs
dictionary on line 361 inComfyUI/custom_nodes/ComfyUI_IPAdapter_plus/IPAdapterPlus.py
or477217b364de427827f181cf11404bfc34181c41
(the one used in https://github.com/siliconflow/onediff/pull/732 ) I get the following error instead:Thank you for your work on OneDiff