I got a Machine Learning Deformer from https://github.com/dgovil/MLDeform. I got it to work in Maya 2022 only with example project. If I use the same model and add my own deformer (PSD). I can get the data to write but training I get errors.
# MLDeform._training.train : Training for joint1 #
# MLDeform._training.train : Normalizing values #
Train on 0 samples, validate on 1 samples
Epoch 1/300
# Error: Empty training data.
# Traceback (most recent call last):
# File “”, line 9, in
# File “C:/Users/EdC/Documents/maya/2022/scripts\MLDeform_training\train.py”, line 179, in train
# batch_size=batch_size)
# File “C:\Program Files\Autodesk\Maya2022\Python37\lib\site-packages\tensorflow\python\keras\engine\training.py”, line 780, in fit
# steps_name=‘steps_per_epoch’)
# File “C:\Program Files\Autodesk\Maya2022\Python37\lib\site-packages\tensorflow\python\keras\engine\training_arrays.py”, line 380, in model_iteration
# aggregator.finalize()
# File “C:\Program Files\Autodesk\Maya2022\Python37\lib\site-packages\tensorflow\python\keras\engine\training_utils.py”, line 111, in finalize
# raise ValueError(‘Empty training data.’)
# ValueError: Empty training data. #
Can you guys help?
Maybe you can look into https://pytorch.org ?
MLDeform has its own deformer. I sound to me that if you replace the deformer with something else you need to train it again? But maybe you can’t just replace the deformer as this is done by the ML. I guess it’s not creating a skin map to be use by any other deformer. The deformation is just based on the mesh data and bone data.
The data was empty for the first joint. I have to to train the whole skeleton to get the ML deformer to get it to work.