API Reference — dna/layers/RBFBehaviorWriter

---

class DNAAPI RBFBehaviorWriter : public virtual BehaviorWriter

Write RBF solver and pose data into a rig's DNA — names, scale, control mappings, solver configuration, and LOD associations.

When to use this

Use this interface when constructing or mutating the RBF behavior layer of a rig: populating poses, binding pose controls, configuring solvers, and mapping solvers to levels of detail. Do not inherit directly from this class — inherit from Writer instead, as noted in the class @warning.

Method groups

Group	Methods
Poses	clearRBFPoses, setRBFPoseName, setRBFPoseScale
Pose controls	clearRBFPoseControlNames, setRBFPoseControlName, setRBFPoseInputControlIndices, setRBFPoseOutputControlIndices, setRBFPoseOutputControlWeights
Solvers	clearRBFSolvers, clearRBFSolverIndices, setRBFSolverIndices, setRBFSolverName, setRBFSolverRawControlIndices, setRBFSolverPoseIndices, setRBFSolverRawControlValues
Solver configuration	setRBFSolverType, setRBFSolverRadius, setRBFSolverAutomaticRadius, setRBFSolverWeightThreshold, setRBFSolverDistanceMethod, setRBFSolverNormalizeMethod, setRBFSolverFunctionType, setRBFSolverTwistAxis
LOD mapping	clearLODRBFSolverMappings, setLODRBFSolverMapping

Example

// Populate one pose and one solver, then map to LOD 0
RBFBehaviorWriter* writer = ...; // obtained via a Writer subclass

// Pose
writer->setRBFPoseName(0, "jaw_open_pose");
writer->setRBFPoseScale(0, 1.0f);

// Pose controls — inputs that multiply the pose output, and outputs the pose drives
const uint16_t inputControls[] = {11, 12, 13, 14};
writer->setRBFPoseInputControlIndices(0, inputControls, 4);
const uint16_t outputControls[] = {7};
writer->setRBFPoseOutputControlIndices(0, outputControls, 1);
const float outputWeights[] = {0.85f};
writer->setRBFPoseOutputControlWeights(0, outputWeights, 1);

// Solver
writer->setRBFSolverName(0, "jaw_rbf_solver");
writer->setRBFSolverType(0, RBFSolverType::RBF);
writer->setRBFSolverDistanceMethod(0, RBFDistanceMethod::Quaternion);
writer->setRBFSolverNormalizeMethod(0, RBFNormalizeMethod::Softmax);
writer->setRBFSolverRadius(0, 0.5f);
writer->setRBFSolverWeightThreshold(0, 0.01f);
const uint16_t rawControls[] = {11, 12, 13, 14};
writer->setRBFSolverRawControlIndices(0, rawControls, 4);
const uint16_t poseIndices[] = {0};
writer->setRBFSolverPoseIndices(0, poseIndices, 1);

// LOD mapping — associate solver index list 0 with LOD 0
const uint16_t solverIdx[] = {0};
writer->setRBFSolverIndices(0, solverIdx, 1);
writer->setLODRBFSolverMapping(0, 0);

Watch out for

• Storage arrays resize implicitly when writing to an index beyond the current size, but setRBFPoseInputControlIndices, setRBFPoseOutputControlIndices, setRBFPoseOutputControlWeights, setRBFSolverRawControlIndices, setRBFSolverPoseIndices, and setRBFSolverRawControlValues require poseIndex / solverIndex to already exist (must be less than the current pose/solver count). Create the pose or solver entry first before assigning its data arrays.
• setRBFSolverRawControlIndices expects the raw control count to match the distance method's arity: Quaternion, SwingAngle, and TwistAngle consume 4 raw controls per driving joint; Euclidean consumes 1 per scalar input.
• Names passed to setRBFPoseName, setRBFPoseControlName, and setRBFSolverName are copied on write — the caller may free the pointer immediately after the call returns.