| On this page |
There are several Machine learning recipes available through the tab menu. These are similar to shelf tools that put down networks of nodes for learning purposes and plug and play processes. The ML Train Deformer recipe creates a setup with two entities for which ML Deformer models can be trained.
Structure ¶
Capybara
This subnetwork prepares the Capybara test geometry for training as an ML Deformer.
Tube
This subnetwork prepares a flexible tube for training as an ML Deformer.
ML_Train_Deformer
This subnetwork performs the data generation and training for the entities contained in the input geometry.
ML_Deform
This subnetwork allows a trained ML Deformer to be used.
Important nodes ¶
ML Pose SerializeHelps to convert each pose into an array of numbers that can be absorbed into the data set.
Principal Component AnalysisHelps to encode each residual skin deformation (relative to linear blend skinning) as a small array of numbers that can be absorbed into the data set.
Armature Capture Prepares an entity for quasistatic tissue simulation.
Armature DeformPerforms a quasistatic tissue simulation on an entity.
ML Train RegressionTrains a model for an ML Deformer using a data set consisting of encoded poses and deformations.
Learning from this example ¶
| To... | Do this |
|---|---|
|
Train ML Deformers for the entities connected to the ML_Train_Deformer network |
Click Cook Output Node and wait for it to finish. |
|
Perform inference on a specific entity |
Change the Entity Name parameter to that of the entity. |
|
Hook up a new creature to train |
Look at the |