You get an instant subscribe for this because any time anyone tells us about a comfyui product or update, they NEED to include installs whether it is a short or long video however they clip it up. It won't work without this and this is CRUCIAL so THANK YOU!! HUGE thanks!!❣
Not yet! IPAdapter has a lot of uses. InstantID is mainly to add style and maintain the face similarity. Face detailer is to ensure that the face generated is anatomically correct.
Amazing video like always, very detailed thank you. man my poor 8GB vram can't handle the thing, i suggest using 4steps lightning models if you have less than 12GB, high quality 15second generations
Hi there!! I am following your guide on macOS. I am stuck at the Python Embedded. I don't seem to have that. Should I just skip it? Thanks in advance 😇
Hello, for MacOS, there are no python_embedded. Instead you will have to activate your python virtual environment. Since MacOS requires installing ComfyUI manually, this python virtual environment will differ from person to person. Typically you will do this in the terminal: source bin/activate. (you have to be in the correct folder; usually where you have ComfyUI installed.) This command may be different if you are using conda. Once you have the virtual environment activated, you can continue following the tutorial and replace python_embedded/python.exe with python3.
none of this is working for me. i tried to install through manager and got an error. then i went and installed the repo from github. still didnt work. then i opened my comfy window and typed in CMD and it opened a command window and I typed in LS and it told me D:\ComfyUI_windows_portable>ls 'ls' is not recognized as an internal or external command, operable program or batch file. but if i watch someone elses tutorial and type in python --version it says i have python 3.11.9
Hello, You are probably using windows cmd. try to open comfyui folder and type in "powershell" instead of "cmd". if this does not work, continue with cmd and replacing "ls" with "dir". I suggest getting powershell or windows terminal from the microsoft store. Alternatively, you can also install git bash. They are all free.
This is a great video and everything is very clear. However, I am still getting: 3.9 seconds (IMPORT FAILED): C:\Users\user\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_InstantID Do you have any insight or can you point me to a good place to troubleshoot this issue?
When you get the import failed message, look for any dependencies error such as no module found error. This will tell you what you are missing. Try looking the error message here: bit.ly/4c8YWBX
@@CodeCraftersCorner Thank you so much this got me going in the right direction! I wasn't loading onnyxruntime correctly so I carefully reviewed the necessary dependencies, installed CUDA and cuDNN correctly and added the various directories to Path and it is working now. Thanks again!
Hi thank for your tutorial. However when i run my workflow i have this error (in the Ksampler) despite using a SDXL checkpoint. Do you how to fix it please ? Expected query, key, and value to have the same dtype, but got query.dtype: float key.dtype: struct c10::Half and value.dtype: struct c10::Half instead. File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\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 "C:\ComfyUI\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 "C:ComfyUI\ComfyUI_windows_portable\ComfyUI odes.py", line 1371, in sample return common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI odes.py", line 1341, in common_ksampler samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
First, make sure that you have the correct controlnet and ipadapter.bin version. If that does not work, you can try add --force-fp16 at the end of the .bat file. You may get black images though. It may be a hardware incompatibility.
Most of this makes sense however, With all due respect: Where does one type the "python_embeded\python.exe pip Insightface ETC. STRING? Yes I see the string of which some is yellow and it is NOT exactly clear as to 1.) WHERE is this string supposed to be enterred 2.} is the yellow part supposed to be typed or what? Essentially What and exactly where PLEASE?
In the video, at 2:43 I showed how to open the terminal in Windows. If you are not using windows, you will have to look how to access the terminal on your system. These instructions should be typed in the terminal. Unfortunately, I do not have access to other operating systems.
Hello, on Windows, we use the .whl files because windows cannot build insightface. For Linux, you should be able to install it with pip. 1. Activate you virtual environment. 2. type in "python -m pip install insightface==0.7.3" and press enter. I cannot guarantee that this will work but that's how to install it in a Linux system.
Thank you for the amazing video. Could you please help me? I am using a GPU, but "-gpu" did not work for me. Now, in the workflow, I see I don't have the GPU option; how can I install it? I ran the code again in the confyui folder but did not change anything.
Hello @alinasiri9173! The command to install the onnxruntime is "python_embeded\python.exe -m pip install onnxruntime-gpu". If it does not work, try without the "-gpu". The overall workflow will still use GPU to run.
@@CodeCraftersCorner notice that there's no space between onnxruntime-gpu has no space before the dash - it's not an "option", it's a different filename.
Hello @moneygotmeblind! You may have to check the model checkpoint file. Click on "Refresh" below the "Queue Prompt" button and see which models you are missing.
hi, I have a problem with onnxruntime-gpu, I installed it with that command but I still encounter error trying to use cuda, I asked chatgpt and it said "it seems that the current latest stable release of ONNX Runtime (1.17.1) does not support ONNX IR version 10", now I can only use CPU in the "instantID face analysis" node which is slow since I'm using more than one image as source. Do you encounter this problem?
Im getitign this error ERROR: insightface-0.7.3-cp311-cp311-win_amd64.whl is not a supported wheel on this platform. Anybody come across this and how to solve this ?
@@CodeCraftersCorner I figured it out. I'm on window 11. My comfy install has ver 3.11 Python but my Python installed iin my OS win 11 is ver 3.10.6. So I ran the 310 version and now it works. Not sure why my standalone comfy is running ver 311 in its environment. But that was the issue
I got this problem after i press "Queue Prompt" How can i fix the problem? Error occurred when executing KSampler: mat1 and mat2 shapes cannot be multiplied (16x2048 and 768x320) File "D:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "D:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\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:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\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:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI odes.py", line 1344, in sample return common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "D:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI odes.py", line 1314, in common_ksampler samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "D:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack\modules\impact\sample_error_enhancer.py", line 22, in informative_sample raise e File "D:\ComfyUI ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack\modules\impact\sample_error_enhancer.py", line 9, in informative_sample return original_sample(*args, **kwargs) # This code helps interpret error messages that occur within exceptions but does not have any impact on other operations. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Hello @user-wg5ex8ho4z! You have to download one from civitai.com or huggingface website. Use the filter to select only SDXL checkpoints. Once you have a checkpoint, move the file to Comfy > models > checkpoints folder. Restart ComfyUI, select the SDXL model from the Load Checkpoint node list and try again.
Big thanks! Between official guides, troubleshooting threads, youtube guides, this is the only accurate & painless install tutorial.
Glad it helped!
You get an instant subscribe for this because any time anyone tells us about a comfyui product or update, they NEED to include installs whether it is a short or long video however they clip it up. It won't work without this and this is CRUCIAL so THANK YOU!! HUGE thanks!!❣
Thank you so much, @HAJJ101!
Thank you! This was what I needed. I couldn't figure out how to get the "Apply InstantID"-node to show up in ComfyUI.
Finally, I see it!!!
You da man!
Omg! Thank you very much for this great tutorial and your work.
Thank you so much, can you please make a similar video for PuLID installation 🙏
Hello, request completed: czcams.com/video/o_JOs56Q3zk/video.html
So Good! Thanks for the awesome walk through 🙏🙏
Glad you enjoyed it!
Thank you for doing this. I'm interested in being an end user versus computer troubleshooter.
Please do Pulid installation, i’am unable to install it
Very detailed video, really easy to follow! 🎉 thank you 🙏🏽
Glad it was helpful, @bozonight!
amazing!! thank you sooooo much
Glad you like it!
I love you ! Thank you!
Thank you! No problem 😊
thank for the video mate, have u done a videos between InstantId, ipadapter and face detailer diferences ?
Not yet! IPAdapter has a lot of uses. InstantID is mainly to add style and maintain the face similarity. Face detailer is to ensure that the face generated is anatomically correct.
ok i think the WHL part of your instructions mightve fixed it for me thank you!
Great to hear!
Thank you for another great video!
Thanks for watching, @58gpr!
pretty good tutorial!
Amazing video like always, very detailed thank you. man my poor 8GB vram can't handle the thing, i suggest using 4steps lightning models if you have less than 12GB, high quality 15second generations
Thanks for the tip, @0A01amir!
thank you for this video
Thank you!
Hi there!! I am following your guide on macOS. I am stuck at the Python Embedded. I don't seem to have that. Should I just skip it? Thanks in advance 😇
Hello, for MacOS, there are no python_embedded. Instead you will have to activate your python virtual environment. Since MacOS requires installing ComfyUI manually, this python virtual environment will differ from person to person. Typically you will do this in the terminal: source bin/activate. (you have to be in the correct folder; usually where you have ComfyUI installed.) This command may be different if you are using conda. Once you have the virtual environment activated, you can continue following the tutorial and replace python_embedded/python.exe with python3.
It works. Thank you very much!
Had to reinstall Comfy and was getting a failed to import, following the steps to install Insight Face via CMD fixed it :) Thanks.
Glad it helped!
I'm getting the same issue even after following all the tutorial steps, how exactly did you solve it?
none of this is working for me. i tried to install through manager and got an error. then i went and installed the repo from github. still didnt work. then i opened my comfy window and typed in CMD and it opened a command window and I typed in LS and it told me D:\ComfyUI_windows_portable>ls 'ls' is not recognized as an internal or external command, operable program or batch file.
but if i watch someone elses tutorial and type in python --version it says i have python 3.11.9
Hello, You are probably using windows cmd. try to open comfyui folder and type in "powershell" instead of "cmd". if this does not work, continue with cmd and replacing "ls" with "dir". I suggest getting powershell or windows terminal from the microsoft store. Alternatively, you can also install git bash. They are all free.
Thank you
Thank you, @francaleu7777!
This is a great video and everything is very clear. However, I am still getting:
3.9 seconds (IMPORT FAILED): C:\Users\user\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_InstantID
Do you have any insight or can you point me to a good place to troubleshoot this issue?
When you get the import failed message, look for any dependencies error such as no module found error. This will tell you what you are missing. Try looking the error message here: bit.ly/4c8YWBX
@@CodeCraftersCorner Thank you so much this got me going in the right direction! I wasn't loading onnyxruntime correctly so I carefully reviewed the necessary dependencies, installed CUDA and cuDNN correctly and added the various directories to Path and it is working now. Thanks again!
Glad you got it to work!
Hi thank for your tutorial. However when i run my workflow i have this error (in the Ksampler) despite using a SDXL checkpoint. Do you how to fix it please ?
Expected query, key, and value to have the same dtype, but got query.dtype: float key.dtype: struct c10::Half and value.dtype: struct c10::Half instead.
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\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 "C:\ComfyUI\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 "C:ComfyUI\ComfyUI_windows_portable\ComfyUI
odes.py", line 1371, in sample
return common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI
odes.py", line 1341, in common_ksampler
samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
First, make sure that you have the correct controlnet and ipadapter.bin version. If that does not work, you can try add --force-fp16 at the end of the .bat file. You may get black images though. It may be a hardware incompatibility.
Most of this makes sense however, With all due respect: Where does one type the "python_embeded\python.exe pip Insightface ETC. STRING? Yes I see the string of which some is yellow and it is NOT exactly clear as to 1.) WHERE is this string supposed to be enterred 2.} is the yellow part supposed to be typed or what? Essentially What and exactly where PLEASE?
In the video, at 2:43 I showed how to open the terminal in Windows. If you are not using windows, you will have to look how to access the terminal on your system. These instructions should be typed in the terminal. Unfortunately, I do not have access to other operating systems.
i cant install succsessfully Ignoring invalid distribution ~orch error
Are there any dependencies missing error when you check the terminal?
@@CodeCraftersCorner no
I'm trying to install this on ubuntu so the .whl files can't be installed because they are for windows. Is there a workaround for this?
Hello, on Windows, we use the .whl files because windows cannot build insightface. For Linux, you should be able to install it with pip.
1. Activate you virtual environment.
2. type in "python -m pip install insightface==0.7.3" and press enter.
I cannot guarantee that this will work but that's how to install it in a Linux system.
Wonderful Video, Step by Step guide. Thank you🙏. Subbed
Thanks for the sub, @priyadharshanprabhu4504!
Thank you for the amazing video. Could you please help me? I am using a GPU, but "-gpu" did not work for me. Now, in the workflow, I see I don't have the GPU option; how can I install it? I ran the code again in the confyui folder but did not change anything.
Hello @alinasiri9173! The command to install the onnxruntime is "python_embeded\python.exe -m pip install onnxruntime-gpu".
If it does not work, try without the "-gpu". The overall workflow will still use GPU to run.
@@CodeCraftersCorner notice that there's no space between onnxruntime-gpu has no space before the dash - it's not an "option", it's a different filename.
@maurypb Yes! there are two binaries which can be installed using pip. The "-gpu" is not a command flag.
Prompt outputs failed validation
Efficient Loader:
- Required input is missing: ckpt_name
please help :(
Hello @moneygotmeblind! You may have to check the model checkpoint file. Click on "Refresh" below the "Queue Prompt" button and see which models you are missing.
hi, I have a problem with onnxruntime-gpu, I installed it with that command but I still encounter error trying to use cuda, I asked chatgpt and it said "it seems that the current latest stable release of ONNX Runtime (1.17.1) does not support ONNX IR version 10", now I can only use CPU in the "instantID face analysis" node which is slow since I'm using more than one image as source. Do you encounter this problem?
Hello, the developer said to use the CPU version until there is support for the new version. I, too, am using the CPU version for now.
@@CodeCraftersCorner ok thanks for this information, hope that they release new version soon
@silverstone6849! Yes, me too.
@@CodeCraftersCorner hi, how can I know when they have update on this? I don't wanna miss it. Thank you
@silverstone6849 probably the best way would be to check the GitHub page.
How does this work for Kaggle??
Sorry, I do not use Kaggle for ComfyUI.
大哥，请我跪下，解决我4天没解决的问题。
Thank you so much
Im getitign this error
ERROR: insightface-0.7.3-cp311-cp311-win_amd64.whl is not a supported wheel on this platform.
Anybody come across this and how to solve this ?
Are you on windows and using python 3.11? If you are on a different platform, this solution will not work.
@@CodeCraftersCorner I figured it out. I'm on window 11. My comfy install has ver 3.11 Python but my Python installed iin my OS win 11 is ver 3.10.6. So I ran the 310 version and now it works. Not sure why my standalone comfy is running ver 311 in its environment. But that was the issue
@HiProfileAI Glad you were able to solve it.
I got this problem after i press "Queue Prompt"
How can i fix the problem?
Error occurred when executing KSampler:
mat1 and mat2 shapes cannot be multiplied (16x2048 and 768x320)
File "D:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\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:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\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:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI
odes.py", line 1344, in sample
return common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI
odes.py", line 1314, in common_ksampler
samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack\modules\impact\sample_error_enhancer.py", line 22, in informative_sample
raise e
File "D:\ComfyUI
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu_2\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack\modules\impact\sample_error_enhancer.py", line 9, in informative_sample
return original_sample(*args, **kwargs) # This code helps interpret error messages that occur within exceptions but does not have any impact on other operations.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Can you try with an SDXL checkpoint? Currently, InstantId only supports SDXL checkpoint.
@@CodeCraftersCorner how to do it sir?
Hello @user-wg5ex8ho4z! You have to download one from civitai.com or huggingface website. Use the filter to select only SDXL checkpoints. Once you have a checkpoint, move the file to Comfy > models > checkpoints folder. Restart ComfyUI, select the SDXL model from the Load Checkpoint node list and try again.
@@CodeCraftersCorner now i got this problem sir. please help
Error occurred when executing KSampler:
'NoneType' object has no attribute 'shape'
Can you share your workflow for me to test? Click on the Save options in ComfyUI menu and save it as a .json file.