I've been using Blender to create a dataset for a machine vision problem. My current method is randomising the bone rotations of a hand for each frame (using Fcurves with the noise modifier), and hoping to use the Fcurve data and the rendered hands as input and output data, so that I could train a neural network to take in an image of a hand and estimate the rotation of each joint.
The image generation worked fine, but unfortunately I can't find a way to get the value of the Fcurves at each frame. I would be fine with either saving them all as matrices (with a value for every Fcurve at each frame) or save the formulas for them, but I'm not sure how to do either. I've been trying to have a look at developer.blender.org and I've found how noise in the Fcurves are implemented (https://developer.blender.org/rBS4b025d68657c197b302516b96f69405aa957accf) but I don't understand how to read that code, I only work in python and I think it's some sort of pseudocode anyway so I really don't know how to access the functions its referencing. I've been trying to work on this for a while but have gotten nowhere. I would be really grateful for any insight. Thank you.