Is it possible to train some of the existing open source neural networks to modify stl files?

Question

3D

Situation: a friend got a photopolymer 3D printer (clone b9 creator) and prints a lot of models.
Each model needs to be placed on supports, which is time-consuming and terribly boring in itself.
Existing programs to automate this business do not cope with the task, tk. photopolymer printing is a rather specific thing.
He already has about 200 models placed on supports (ASCII STL) and is it realistic to train any of the existing free networks on them for automatic placement of supports?

Answer 1

Neural networks are by no means a panacea. If there is a clear algorithm, then it is better to use it.
If you find it difficult to formalize the algorithm, then most likely you will not be able to train the neural network either, because there the task of formalizing the input data will be more difficult.