Media Pipe input error. Image input crashes on all nodes except default image load.

[firespark81]

Same issue as I was having with your base media pipe nodes. It will not accept input from any node that is not the default image load node.

Here is the node I have plugged into it but I have tried many. Same error every time.

If you need more info just let me know what you need. I'll do my best to provide it.

ComfyUI Error Report

Error Details

• Node ID: 247
• Node Type: MediaPipeHandLandmarkerNode
• Exception Type: TypeError
• Exception Message: init(): incompatible constructor arguments. The following argument types are supported:
  1. mediapipe.python._framework_bindings.image.Image(image_format: mediapipe::ImageFormat_Format, data: numpy.ndarray[numpy.uint8])
  2. mediapipe.python._framework_bindings.image.Image(image_format: mediapipe::ImageFormat_Format, data: numpy.ndarray[numpy.uint16])
  3. mediapipe.python._framework_bindings.image.Image(image_format: mediapipe::ImageFormat_Format, data: numpy.ndarray[numpy.float32])

Invoked with: kwargs: image_format=<ImageFormat.SRGB: 1>, data=array([[[ 21, 23, 18],
[ 22, 25, 21],
[ 22, 25, 19],
...,
[ 84, 100, 51],
[ 66, 80, 31],
[ 60, 75, 32]],

   [[ 23,  25,  20],
    [ 22,  24,  21],
    [ 23,  25,  22],
    ...,
    [ 70,  91,  45],
    [ 67,  89,  38],
    [ 48,  71,  28]],

   [[ 21,  23,  19],
    [ 23,  25,  21],
    [ 23,  25,  22],
    ...,
    [ 59,  81,  41],
    [ 67,  94,  48],
    [ 44,  59,  28]],

   ...,

   [[ 59,  53,  37],
    [ 52,  49,  33],
    [ 49,  45,  29],
    ...,
    [ 48,  52,  38],
    [ 45,  51,  38],
    [ 45,  50,  40]],

   [[ 54,  51,  34],
    [ 48,  45,  30],
    [ 55,  51,  34],
    ...,
    [ 50,  56,  44],
    [ 47,  54,  44],
    [ 54,  61,  48]],

   [[ 58,  58,  40],
    [ 52,  51,  37],
    [ 55,  52,  37],
    ...,
    [ 64,  70,  54],
    [ 64,  70,  58],
    [ 63,  72,  54]]], dtype=uint8)

Stack Trace

  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 327, in execute
    output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
                                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 202, in get_output_data
    return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 174, in _map_node_over_list
    process_inputs(input_dict, i)

  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 163, in process_inputs
    results.append(getattr(obj, func)(**inputs))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_realtimenodes\node_wrappers\mediapipe_vision\hand\landmark\hand_landmark.py", line 62, in detect
    hand_landmarks_batch = detector.detect(
                           ^^^^^^^^^^^^^^^^

  File "D:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_realtimenodes\src\mediapipe_vision\hand_landmark\detector.py", line 93, in detect
    mp_image = mp.Image(image_format=mp.ImageFormat.SRGB, data=np_image)

[firespark81]

detector.txt

hand_landmark.txt

I fixed it but I used gemini 2.5 so the code is probably trash, but I wanted to share it with you anyway so you could see how it went about it. I don't know enough to understand what it did to explain it to you so I attached the 2 files I modified to get it working. Had to change the extension to be able to attach them.
Would assume you probably need to give all the other nodes from media pipe the same treatment but I don't know, have not experimented with them yet. Just wanted the hands one. If you need anything else from me like the chat log with gemini or whatever feel free to let me know.

[ryanontheinside]

I assume youre working with batches of images as opposed to real-time (like real-time with comfystream)?

I did try to make this work with batches as well, but it was an afterthought. Thank you for letting me know! Def let me know if you have more issues

[firespark81]

Nope not batches, just single image. It doesn't matter where you patch it in from I can take it right out of a vae decode and it gives the error. But you are correct I am working with static images and not real time. I do select the option for image and not video.

[ryanontheinside]

I am not able to reproduce this error

[RodrigoSKohl]

I am not able to reproduce this error

this error is happening in any Media Pipe pipeline on Windows, only when an image is selected. MediaPipe 0.10.x only accepts protobuf < 5

[Pythonpa]

I am not able to reproduce this error

this error is happening in any Media Pipe pipeline on Windows, only when an image is selected. MediaPipe 0.10.x only accepts protobuf < 5

So, you mean we should downgrade the version of protobuf to <5?

[RodrigoSKohl]

I am not able to reproduce this error

this error is happening in any Media Pipe pipeline on Windows, only when an image is selected. MediaPipe 0.10.x only accepts protobuf < 5

So, you mean we should downgrade the version of protobuf to <5?

Yes

[Pythonpa]

I am not able to reproduce this error

this error is happening in any Media Pipe pipeline on Windows, only when an image is selected. MediaPipe 0.10.x only accepts protobuf < 5

So, you mean we should downgrade the version of protobuf to <5?

Yes

Yeah,It works for me,