Navigation
API > API/Plugins > API/Plugins/MetaHumanPipelineCore
Model output variants — each backend (ONNX, CoreML) may use different tensor names and shapes for the same logical outputs. Resolved at load time in CheckModels.
An FModelVariant maps logical output roles (e.g. "Scores", "Landmarks") to tensor descriptors. The role keys are constant across backends; the tensor names and shapes within each FModelOutputDesc may differ between ONNX, CoreML, etc.
| Name | FModelOutputDesc |
| Type | struct |
| Header File | /Engine/Plugins/MetaHuman/MetaHumanCoreTechLib/Source/MetaHumanPipelineCore/Public/Nodes/HyprsenseRealtimeNode.h |
| Include Path | #include "Nodes/HyprsenseRealtimeNode.h" |
Syntax
struct FModelOutputDesc
Variables
Public
| Name | Type | Remarks | Include Path | Unreal Specifiers |
|---|---|---|---|---|
| ActiveControlFlags | TConstArrayView< const bool > | Solver variants only. | Nodes/HyprsenseRealtimeNode.h | |
| Name | FString | Nodes/HyprsenseRealtimeNode.h | ||
| Shape | UE::NNE::FTensorShape | Nodes/HyprsenseRealtimeNode.h |