ComfyUI InsightFace Windows Fast Installation (2024) | NO MORE ERRORS FOR IPADAPTERS / ROOP 

Hi, did you manage to solve it? I have not seen that error before but I think it might be a using wrong models in the wrong node kinda error.

I see, maybe can try this: 1. Under windows search, search 'Terminal'. 2. Right click on it, select 'Run as Administrator'. 3. type in 'cd D: ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable' 4. type in the same command from the video. Let me know if it works!

Close ComfyUI terminal

it works! Ty@@Ernest2002

@@Ernest2002 Wow, this random comment just saved me a headache. Thank you.

@@Ernest2002 wow thank you guy ! you save my night !!!!!!

Same. Is there a colab notebook I can just clone?

Yup that is expected, you are probably using a faceid model in an IPAdapter node, but faceid models only work in IPAdapter FaceID nodes. You can check out my other IPAdapter video where I talked about which models to use for which node: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-oYjEFHb--RA.html

Hi, just to be sure the installation is successful, you could run this command if you are using windows (replace findstr with grep for linux): .\python_embeded\python.exe -m pip list | findstr insightface If there is an output like insightface 0.7.3, it means insightface has installed correctly and the error lies in either wrong model placement or using wrong python executable.

Hi, sorry I do not use stability matrix, but after some searching, I think this might help you out: github.com/LykosAI/StabilityMatrix/issues/344

chad

I'm getting "ERROR: Invalid requirement: '[path-to-wheel]'" Is this related to your issue and fix?

Facing the same issue on Kaggle

Hi sorry I do not use a mac, you may want to refer to github.com/deepinsight/insightface/issues/2493 and see if the solutions work :)

Hello, I think have seen others facing the same issue in the comments. Their case was that after checking, the python ver they are running from for ComfyUI is not 3.11. I'm not sure on your case if its the same!

This means you used the wrong model for the wrong node, models with 'faceid' goes to apply faceid ipadapter, those without faceid in the model name goes to apply ipadapter.

Will share more soon in my next vid!

Hi brother, yes the dev removed the node from the repository.

@@DataLeveling thanks bro for letting me know

Hi, yes I heard from others that if they installed reactor first before doing the steps in this video, it has an error. I suggest uninstalling the reactor that you have from the manager, install insightface as per this video, then follow my Reactor installation guide: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-PQLttrd-Epc.html This should make sure everything runs correctly.

Hi, you could click on the windows address at the top and type cmd and it will open the terminal :)

I see, okay try this, put the wheel file in the ComfyUI_windows_portable folder, then same thing, right click open in terminal and run this: .\python_embeded\python.exe -m pip install insightface-0.7.3-cp311-cp311-win_amd64.whl onnxruntime

thanks so much - it worked !! You're a legend@@DataLeveling

Hi, are you using ComfyUI? The base installation should have it installed already. If you are using ComfyUI, you can check where your python is running from, from the run_cpu / run_nvidia_gpu file.

@@DataLeveling Thanks, but how cai i know where python is running from? Yes i'm a noob

@@user-lv9vs9il7z no worries, if lets say you are not using the portable version, it is more complicated as it requires you to create your own python virtual environment then launch python main.py. If you haven't been using a virual environment, then you can find your python by typing in cmd: 'where python', the one you are using is probably something like this: \AppData\Local\Programs\Python\Python311\python.exe It is recommended to use virtual environment, so that it will not affect your base python when you continuously install new libraries. A simpler alternative is to use the portable version, they created for us the python_embedded folder, and if you right click the 'run_nvidia_gpu.bat' and open it with notepad, you can see it runs the python from the python_embedded folder: '.\python_embeded\python.exe -s ComfyUI\main.py' You can check out my video on installing portable ComfyUI and ComfyUI Manager: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-_1fSVcNIOBw.html

@@user-lv9vs9il7z no worries, if lets say you are not using the portable version, you can find your python by typing in cmd: 'where python'. It is recommended to use a python virtual environment, so that it will not affect your base python when you continuously install new libraries. To create a virtual env: docs.python.org/3/library/venv.html A simpler alternative is to use the portable version, they created for us the python_embedded folder, and if you right click the 'run_nvidia_gpu.bat' and open it with notepad, you can see it runs the python from the python_embedded folder: '.\python_embeded\python.exe -s ComfyUI\main.py' You can check out my video on installing portable ComfyUI and ComfyUI Manager: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-_1fSVcNIOBw.html

@@DataLeveling Excuse me for responding very late, thanks a lot

Hi, to do that, change the python path in your comfyui's run_nvidia_gpu.bat file to the a1111 python path.

​@@DataLeveling thank you. Btw i got it working with 3.11 thx to you. 🎉

i have the same error and i have not find any solution :(

@@JipandCie I solved it! Open the file downloaded from the official website, right-click the mouse, and turn on all permissions in the properties!