Vision node
Body Landmark
Extracts a single body joint for a chosen person from Vision Tracking. Pick the person and the joint; outputs its position and confidence (plus depth in 3D mode).
Inputs
| Port | Type | Default |
|---|---|---|
| BodyConnect Vision Tracking's Body output here — the pose data every joint position is read from. Nothing connected reports not found. | json | — |
| PersonWhich tracked person to read, by index. Negative values count as 0; an index nobody occupies simply reports not found. | int | 0 |
| JointWhich body joint to follow, chosen from the tracker's joint list. A name that is not on the list falls back to the first joint. | string | |
Outputs
| Port | Type |
|---|---|
| X | float |
| Y | float |
| Confidence | float |
| Found | bool |
What It Does
Vision Tracking outputs all the landmarks bundled together. Body Landmark is a simple "picker" that pulls out one joint of one person — say, person 0's right wrist — and gives you its X and Y position as plain numbers you can wire anywhere.
It's the bridge between "tracking data" and "a value that drives a shape, a color, a particle emitter, or a parameter."
How It Works
Connect Vision Tracking's Body output to this node. Two dropdowns then let you choose the Person (0, 1, 2…) and the Joint (nose, neck, left/right shoulder, elbow, wrist, hip, knee, ankle, eyes, ears). The menus are filled from the live data, so you only see joints that actually exist. Outputs are the joint's X, Y and a Confidence (0–1) telling you how sure the tracker is. In 3D mode a depth value is also available.
Parameters
| Parameter | What it does |
|---|---|
| Person | Which tracked person to read from. |
| Joint | Which body point to extract. |
Examples
- Follow the head — pick "nose", wire X/Y into a Circle's position → the circle sits on the face.
- Reach trigger — pick "right wrist", feed Y into a Compare node ("Y > 0.5") → fire an event when the hand is raised.
- Particle hand — wire the wrist X/Y into a particle emitter's position so sparks fly from your hand.
Tips
- Use the Confidence output to ignore unreliable points (e.g. only act when Confidence > 0.3).
- Smooth the X/Y with a Smooth node for extra-fluid motion on top of Vision Tracking's own smoothing.
Open Body Landmark in the interactive catalog →
Body Landmark is one of 381 nodes in Grafit Pro.
Free download. Every feature for 7 days, no card. Then €99.99 once — or keep playing your projects as a Player.