[Nodes] IPAdapterUnifiedLoader, IPAdapter not found on fresh install ComfyUI,ComfyUI_IPAdapter_plus

[svenhimmelvarg]

Platform: Linux Python: v.3.10.2

I have a new installation of ComfyUI and ComfyUI_IPAdapter_plus, both at the latest as of 30/04/2024. No other modules are installed and I get the error where the following Nodes are not found (see screenshot of default workflow at the bottom):

• IPAdapterUnifiedLoader
• IPAdapter

The startup screen show this (extra model loading output is editing down for conciseness):

(ai)linuxmachin@pop-os:/mnt/sdc1/apps/ComfyUI-Minimal$ python main.py  --listen 0.0.0.0 --multi-user  --disable-xformers 
Total VRAM 24257 MB, total RAM 48091 MB
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 3090 : cudaMallocAsync
VAE dtype: torch.bfloat16
Using pytorch cross attention
Adding extra search path checkpoints /mnt/sdc1/apps/ComfyUI/models/models/checkpoints/
   ...
Adding extra search path vae /mnt/sdc1/apps/autoimport/vae

Import times for custom nodes:
   0.0 seconds: /mnt/sdc1/apps/ComfyUI-Minimal/custom_nodes/websocket_image_save.py
   0.0 seconds: /mnt/sdc1/apps/ComfyUI-Minimal/custom_nodes/ComfyUI_IPAdapter_plus

Starting server

To see the GUI go to: http://0.0.0.0:8188

Note that the example custom node and the IP Adapter plus are the only ones installed.

All the requirements are met. In fact the requirements are complete enough to run about 30 other custom nodes. This environment is being used to run this minimal set up above.

Is there any other debug information I can provide to help diagnose this?

[cubiq]

can I see the directory structure of the IPAdapter extension you installed?

[svenhimmelvarg]

Sure, here you go. The directory tree taken from $COMFY_HOME/custom_nodes:

linuxmachin@pop-os:/mnt/sdc1/apps/ComfyUI-Minimal/custom_nodes$ tree -L 3
.
├── ComfyUI_IPAdapter_plus
│   ├── examples
│   │   ├── batch_images.jpg
│   │   ├── canny_controlnet.jpg
│   │   ├── FaceID_basic.json
│   │   ├── FaceID_enhanced.json
│   │   ├── FaceID_Plus.json
│   │   ├── FaceID_Plus_v2.json
│   │   ├── face_id_wf.jpg
│   │   ├── face_swap.jpg
│   │   ├── image_weighting.jpg
│   │   ├── inpainting.jpg
│   │   ├── IPAdapter_2_masks.json
│   │   ├── IPAdapter_batch_images.json
│   │   ├── IPAdapter_Canny.json
│   │   ├── IPAdapter_face.json
│   │   ├── IPAdapter_inpaint.json
│   │   ├── IPAdapter_mask.json
│   │   ├── IPAdapter_prepped.json
│   │   ├── IPAdapter_sdxl_vit-h.json
│   │   ├── IPAdapter_weighted.json
│   │   ├── IPAdapter_weight_types.json
│   │   ├── masking.jpg
│   │   ├── noise_example.jpg
│   │   ├── prep_images.jpg
│   │   ├── timestepping.jpg
│   │   └── weight_types.jpg
│   ├── __init__.py
│   ├── IPAdapterPlus.py
│   ├── IPAdapter_workflow.json
│   ├── ipadapter_workflow.png
│   ├── LICENSE
│   ├── models
│   │   └── legacy_models_location.txt
│   ├── __pycache__
│   │   ├── __init__.cpython-310.pyc
│   │   ├── IPAdapterPlus.cpython-310.pyc
│   │   └── resampler.cpython-310.pyc
│   ├── README.md
│   └── resampler.py
├── example
│   ├── example_node.py.example
│   ├── __init__.py
│   ├── __pycache__

[emourdavid]

I have similar problem even comfy ipadapter plus node state loaded successfully:

Windows 11 Python 3.10.11 Torch 12.1.1cuda121 Latest ComfyUI + IPAdapter Plus

Updates: When I replace old nodes to new nodes as your suggestion in main page, I have new problem. I changed from light, standard, strength... and it shows like following

[cubiq]

@svenhimmelvarg your installation looks broken. delete and reinstall the extension

[PresTrembleyIIIEsq]

I'm getting the same issue as OP, and like they did, I completely re-installed ComfyUI and then ComfyUI_IPAdapter_plus. When that still didn't work, I also used the "Try update" button to try updating IPAdapter_plus from within the ComfyUI Manager, with the same result.

I'm also on Linux (Manjaro), but I'm using Python 3.10.11.

[aleph65]

Same exact issue here.

[VIPspawn]

I'm having a same problem with latest update. When I load the old update (In that case, this issue doesn't occur.) I only get an error saying "H265" output.

[cubiq]

I don't know what the manager is doing. just download this repository and put it into the custom_nodes directory. then download and rename all the models

[VIPspawn]

I don't know what the manager is doing. just download this repository and put it into the custom_nodes directory. then download and rename all the nodes

Not Fixed

[cubiq]

that is an info box, not an error

[TheMindExpansionNetwork]

same issue

[2PeteShakur]

I'm having a same problem with latest update. When I load the old update (In that case, this issue doesn't occur.) I only get an error saying "H265" output.

make sure to have a folder named "ipadapter" inside the "model" folder. Put your ipadapter model files inside it, resfresh/reload and it should be fixed.

[PresTrembleyIIIEsq]

Not sure if I did anything differently this time, but it worked again for me this morning after doing the following:

1. Reinstall ComfyUI
2. Reinstall ComfyUI_IPAdapter_plus using git clone in the ComfyUI/custom_nodes folder
3. Re-download all of the models and make sure they have the correct names and locations

[aleph65]

@everyone, I watched Olivio's video and was able to fix it. I think I had to press Update All in the manager (top level menu), and I downloaded all IP Adapter files manually into the required directories. I'm running this on Runpod.

https://youtu.be/c8HWDQ67Dvg?si=9bimuy0HLf5AUi5W

Cheers

[VIPspawn]

I'm having a same problem with latest update. When I load the old update (In that case, this issue doesn't occur.) I only get an error saying "H265" output.

make sure to have a folder named "ipadapter" inside the "model" folder. Put your ipadapter model files inside it, resfresh/reload and it should be fixed.

The solution you provided is correct; however, when I replaced the node with a new one, my issue was resolved.

[ye7iaserag]

Here is a solution that works for now to support the workflow hoarders:

• Go to custom_nodes folder.
• Make a copy of ComfyUI_IPAdapter_plus and name it something like ComfyUI_IPAdapter_plus_legacy.
• Checkout commit 6a411dcb2c6c3b91a3aac97adfb080a77ade7d38
• Restart ComfyUI
• ???
• PROFIT???

This is a work around, but I hope the guys here add support to accommodate legacy workflows.

[garbe-github-support]

me too，+1

[samehra]

What worked for me is adding the line below to the ComfyUI/folder_paths.py file when the "folder_names_and_paths" dict is created.

_folder_names_and_paths["ipadapter"] = ([os.path.join(models_dir, "ipadapter")], supported_ptextensions)

The issue I think is ipadpter path is not getting picked correctly from the extra_model_paths.yaml file

[Sunsolele]

I'm having a same problem with latest update. When I load the old update (In that case, this issue doesn't occur.) I only get an error saying "H265" output.

make sure to have a folder named "ipadapter" inside the "model" folder. Put your ipadapter model files inside it, resfresh/reload and it should be fixed.

You saved my life! Thanks, buddy!

[justwjx]

same issue!!!!!!!!!!

[Dabhand404]

I'm not a pro in this but this works for me, Step1- Uninstall ipadapterplus with comfyUI manager Step 2- Go to -https://github.com/cubiq/ComfyUI_IPAdapter_plus step 3- Download - /ComfyUI/models/clip_vision (note you might need to rename this) Step 4- Download- /ComfyUI/models/ipadapter, (create it if not present) Step 5- FaceID models require insightface, download in /ComfyUI/models/ipadapter Step 6- Most FaceID models require a LoRA. download in /ComfyUI/models/loras

once all the downloads are done. Step 7- Install ipadapterplus with comfyUI manager

hope this helps

[kernel2048]

This did the trick for me without a full reinstall:

1. Go to ComfyUI\custom_nodes\ComfyUI_IPAdapter_plus
2. Copy or move all the bin files to ComfyUI\models\ipadapter
3. Queue your workflow again.

Hope this helps too.

[raochinmay6]

see this !!! Exception during processing!!! ClipVision model not found. Traceback (most recent call last): File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "E:\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 "E:\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 "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_IPAdapter_plus\IPAdapterPlus.py", line 529, in load_models raise Exception("ClipVision model not found.") Exception: ClipVision model not found.

[shimib577]

this worked for me:

go to '.../ComfyUI/custom_nodes/ComfyUI_IPAdapter_plus', then open terminal and put:

git checkout 6a411dcb2c6c3b91a3aac97adfb080a77ade7d38

this switch code to last working state in previous version.

[KhanBen]

Try renaming ip-adapter to IP adapter, it worked for me

[jonnax-dev]

(1)adding path to the ComfyUI/folder_paths.py as following:

What worked for me is adding the line below to the ComfyUI/folder_paths.py file when the "folder_names_and_paths" dict is created.

_folder_names_and_paths["ipadapter"] = ([os.path.join(models_dir, "ipadapter")], supported_ptextensions)

The issue I think is ipadpter path is not getting picked correctly from the extra_model_paths.yaml file

(2)then the files in check custom_nodes\ComfyUI_IPAdapter_plus: if the "utils.py；CrossAttentionPatch.py“ does not exist, that means ComfyUI_IPAdapter_plus have not been updated to latest verion. 2.1 reinstall ComfyUI_IPAdapter_plus will be ok.