Pythoner: problem with numpy 2.5.2 on mac M1 with/-out Rosetta

zimt

dear forum members,

i'm trying to run the pythoner examples from https://community.troikatronix... (DX quick start) + "pythoner_examples_v0.4-release" from the add-ons page, and followed also the video on pythoner on youtube python mediapipe and every other tutorial that i could find.

my pythoner works fine with the json and beautifulsoup4 libraries but as soon as numpy wants to be loaded i get the errors attached below.

it seems that i have a problem with the arm / x86 libs for numpy that came with the "install-packages-into-virtual-environment.scpt" in the "Python_Support_Scripts" folder. i tried to install the VENV to the "virtual_env" folder of the project and also to the standard plugin folder + activating the path but with the same result.

i'm on a MBP M1 with osx 15.6.1 and isadora 4.5.1 - in Activity Monitor Isadora shows up as "Intel", so it's using Rosetta in x86-mode.

--> update: i checked in Finder / Info that for the Universal Binary of Isadora 4.5.1 after a fresh installation the "Run with Rosetta" was checked. is there a reason to run it in Rosetta ? when running in ARM-mode numpy works BUT...

A) in the "python_mediapipe_pose_detection_v10 v4.5.1.izz" i still get a red [!] sign on the pythoner and no tracking
(Monitor Windows says: >> File "Module_Oo0FftFuw4.py", line 118, in python_init >> AttributeError: module 'mediapipe' has no attribute 'solutions')

and

B) OpenNI Tracker does not work... is there a way to make numpy and openNI work in ARM mode? READING openNIpage i guess the answer is NO. --> is the Processing workaround (following your Tutorial) the best way then?

is there a possibility to install Numpy as x86 version to the virtual_env folder?
i'd prefer to have everything inside one software without Processing/Syphon...

thanks!

============ A ====================

ImportError:

IMPORTANT: PLEASE READ THIS FOR ADVICE ON HOW TO SOLVE THIS ISSUE!

Importing the numpy C-extensions failed. This error can happen for
many reasons, often due to issues with your setup or how NumPy was
installed.

We have compiled some common reasons and troubleshooting tips at:

https://numpy.org/devdocs/user...

Please note and check the following:

* The Python version is: Python 3.12 from "/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/bin/python"
* The NumPy version is: "2.5.2"

and make sure that they are the versions you expect.

Please carefully study the information and documentation linked above.
This is unlikely to be a NumPy issue but will be caused by a bad install
or environment on your machine.

Original error was: dlopen(/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/numpy/_core/_multiarray_umath.cpython-312-darwin.so, 0x0002): tried: '/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/numpy/_core/_multiarray_umath.cpython-312-darwin.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64')), '/System/Volumes/Preboot/Cryptexes/OS/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/numpy/_core/_multiarray_umath.cpython-312-darwin.so' (no such file), '/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/numpy/_core/_multiarray_umath.cpython-312-darwin.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64'))

====== B ===================

ImportError: dlopen(/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/PIL/_imaging.cpython-312-darwin.so, 0x0002): tried: '/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/PIL/_imaging.cpython-312-darwin.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64')), '/System/Volumes/Preboot/Cryptexes/OS/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/PIL/_imaging.cpython-312-darwin.so' (no such file), '/Users/z/Documents/Isadora/Example Patches and Tutorials/pythoner_examples_v0.4-release/virtual_env/lib/python3.12/site-packages/PIL/_imaging.cpython-312-darwin.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64')) (Module_o__8XJymcD.py, line 1)

=============================================================================

bonemap

@zimt

Hi,

I don't have a specific solution for your issue, unfortunately. I have projects using mediapipe that I have come back to and found that there have been a number of changes to both Mediapipe and Pythoner compatibility that required rewinding and using an older, supported Mediapipe and Python version (such as Python 3.10) to avoid dependency conflicts.

My Mediapipe projects have been usable again after doing that. Linked here is a Mediapipe face-tracking patch in Isadora 4.5.1 but it was made before a json/pyhthoner skeleton decoder actor was available, so the face marker decoding is scripted manually using a Text Parser actor.

Mediapipe Face by Bonemap

Best Wishes

Russell

Woland

@zimt said:

every other tutorial that i could find.

Did you find and read through this article about Pythoner and NumPy yet? https://support.troikatronix.com/support/solutions/articles/13000118388

DusX

@zimt said:

is there a reason to run it in Rosetta ?

The only real reason to run Isadora in Rosetta mode is to use the 'OpenNI Tracker' actor which is EOL.
The core libraries used for the OpenNI tracker, haven't been and will not be updated for use with Apple Arm processors, so the OpenNI Tracker can't be updated either unfortunately.

@zimt said:

is there a way to make numpy and openNI work in ARM mode?

Maybe, this configuration isn't tested. You may be able to install a python virtual environment via terminal forcing the process to run under x86_64.
That will give you an 'intel' version of Python, and if the version is exact (it is 3.12.10 currently) Isadora (running under Rosetta) may be able to use it, and when you install the required modules via PIP, you could use something like:

arch -x86_64 /path/to/python3 -m pip install --only-binary=:all: numpy==2.5.2

to force the intel version of numpy to be installed.

...As I said, this is untested, and definitely a more advanced install/config. I would avoid it to be honest.