API Reference — dna/layers/MachineLearnedBehaviorExt

MachineLearnedBehaviorOperationType

Tag the type of operation a machine-learned behavior node performs within the DNA rig evaluation graph.

Why this exists

ML behavior evaluation in a DNA rig is decomposed into discrete typed steps — gather inputs, run inference, scatter outputs, combine results — rather than a single monolithic operation. This enum makes the step type explicit so the evaluator can dispatch each node to the correct kernel without runtime type inspection. Using a typed enum over a raw integer or string prevents mis-dispatch and makes the operation pipeline readable at a glance.

Fields

Name Type Description
Unspecified MachineLearnedBehaviorOperationType Default sentinel; operation type has not been assigned. Treat a node with this value as misconfigured.
Gather MachineLearnedBehaviorOperationType Reads input values from a source buffer and feeds them into the ML node.
Scatter MachineLearnedBehaviorOperationType Writes output values produced by the ML node back to a target buffer.
MLP MachineLearnedBehaviorOperationType Executes a multi-layer perceptron inference pass on the gathered inputs.
WeightedSum MachineLearnedBehaviorOperationType Combines multiple input contributions using per-weight summation.

Relationships

  • MachineLearnedBehaviorParameterKey — key enum used to configure parameters on the same ML behavior node

MachineLearnedBehaviorParameterKey

Identify which joint-space configuration parameter to read or write on a machine-learned behavior node.

Why this exists

ML behavior nodes drive joint transforms, and those transforms depend on coordinate conventions that vary per rig and per engine — rotation representation, axis orientation, sign handedness, and Euler order. Rather than passing configuration as untyped key strings or raw indices, this enum makes every configurable aspect a named constant. Callers pass a MachineLearnedBehaviorParameterKey value to a parameter accessor, and the type system prevents passing an invalid key at compile time.

Fields

Name Type Description
JointTranslationType MachineLearnedBehaviorParameterKey Specifies the translation convention the joint uses.
JointRotationType MachineLearnedBehaviorParameterKey Specifies the rotation representation (e.g. quaternion, Euler angles).
JointScaleType MachineLearnedBehaviorParameterKey Specifies the scale convention the joint uses.
JointCoordinateSystemAxisX MachineLearnedBehaviorParameterKey Maps the joint's local X axis to a world-space axis.
JointCoordinateSystemAxisY MachineLearnedBehaviorParameterKey Maps the joint's local Y axis to a world-space axis.
JointCoordinateSystemAxisZ MachineLearnedBehaviorParameterKey Maps the joint's local Z axis to a world-space axis.
JointRotationSignAxisX MachineLearnedBehaviorParameterKey Sign correction applied to rotation about the joint's X axis.
JointRotationSignAxisY MachineLearnedBehaviorParameterKey Sign correction applied to rotation about the joint's Y axis.
JointRotationSignAxisZ MachineLearnedBehaviorParameterKey Sign correction applied to rotation about the joint's Z axis.
JointRotationSequence MachineLearnedBehaviorParameterKey Euler rotation application order (e.g. XYZ, ZYX) for this joint.

Relationships

  • MachineLearnedBehaviorOperationType — operation-type enum for the same ML behavior node