Skip to main content

Class: DetectedBodyPose

Defined in: src/world/humans/DetectedBodyPose.ts:81

Represents a single human body pose detected in physical space. Inherits from THREE.Object3D to fit naturally into the Three.js scene graph, positioning itself at the estimated hips/center of the tracked human.

Extends​

Constructors​

Constructor​

new DetectedBodyPose(poseId, landmarks, detection2DBoundingBox): DetectedBodyPose

Defined in: src/world/humans/DetectedBodyPose.ts:89

Creates an instance of DetectedBodyPose.

Parameters​

poseId​

number

A unique tracking identifier for this body pose.

landmarks​

PoseLandmark[]

The list of raw and 3D-projected anatomical landmarks.

detection2DBoundingBox​

Box2

The 2D bounding box of the person in normalized screen space.

Returns​

DetectedBodyPose

Overrides​

THREE.Object3D.constructor

Properties​

detection2DBoundingBox​

detection2DBoundingBox: Box2

Defined in: src/world/humans/DetectedBodyPose.ts:92

The 2D bounding box of the person in normalized screen space.

landmarks​

landmarks: PoseLandmark[]

Defined in: src/world/humans/DetectedBodyPose.ts:91

The list of raw and 3D-projected anatomical landmarks.

poseId​

poseId: number

Defined in: src/world/humans/DetectedBodyPose.ts:90

A unique tracking identifier for this body pose.

xb?​

optional xb: XBObjectOptions

Defined in: src/interaction/InteractionTypes.ts:32

Inherited from​

THREE.Object3D.xb

Methods​

getJointPosition()​

getJointPosition(name, options): Vector3 | null

Defined in: src/world/humans/DetectedBodyPose.ts:115

Returns the 3D world space position of a specific joint/landmark in meters. Exposes both standard MediaPipe landmark mappings and composite VRM/humanoid landmarks.

The pose model always returns all landmarks, including ones it could not actually see, so a body that is only half in frame still reports legs. Pass minVisibility to drop those guesses instead of drawing them.

Parameters​

name​

string

The name of the joint (standard or composite).

options​

Set minVisibility to reject landmarks the model is not confident about. Defaults to 0, which keeps every landmark.

minVisibility?​

number = 0

Returns​

Vector3 | null

A clone of the 3D world space position vector, or null if the joint is undetected, unprojected, or below minVisibility.