Load the torinmb MediaPipe engine once to enable any combination of face, hand, body, and segmentation pipelines, avoiding conflicts from separate setup tools.
Set up selfie segmentation in TouchDesigner to isolate a person from the webcam as an alpha mask or transparent person cutout for direct use in keying or matte tools.
Build a stick-figure skeleton in TouchDesigner from 33 pose landmarks with glowing bones. Sources: synthetic, MediaPipe, OSC, or an existing pose CHOP. Adjust line color, width, camera distance.
Create a depth-driven GPU POP scatter field in TouchDesigner, using depth/mask TOPs to displace or emit particles, with automatic MediaPipe setup when needed.
Set up real-time body tracking from a webcam in TouchDesigner: loads the MediaPipe engine, creates a 33-landmark pose CHOP, and builds a live skeleton. Just pick your webcam and enable Pose.
Set up real-time hand tracking from a webcam in TouchDesigner, generating a CHOP with 21 landmarks per hand for gesture detection and pose skeleton binding.
Integrates Square sandbox catalog and orders management with voice commands (via ElevenLabs) and hand gesture controls (via MediaPipe) through an interactive seller dashboard in Goose Desktop.
Builds a body-reactive visual from pose tracking: glowing marks on 33 landmarks with point, glow, or trail styles. Uses a synthetic pose for instant preview; supports MediaPipe, OSC, or a pose CHOP.
Generate reactive scalar channels from MediaPipe body-pose landmarks—such as hand height, arm openness, and joint angles—for direct binding to TouchDesigner parameters.
Set up full-body pose tracking to drive body-reactive visuals. Use the synthetic animated pose for instant preview without a camera, then switch to MediaPipe, OSC, or an existing CHOP for live input.
Sets up webcam-based face-landmark tracking in TouchDesigner using MediaPipe, generating a 468/478-point face-landmark CHOP with confidence scores for use in data binding or visualization.
Maps hand gestures to TDAbleton TDA_Mapper controls: outputs left/right pinch and wrist-roll maps, builds a skeleton overlay, and links the mapper to an existing setup for real-time performance.
Create an interactive TouchDesigner installation where an open palm emits soap bubbles, with body and hand landmarks acting as colliders to bat them, plus a visible body contour overlay.
Run any model from the Replicate catalog by its owner/name identifier. Supports all categories including embeddings, segmentation, depth estimation, and more.
Converts MediaPipe pose tracking into reactive scalar channels (hand height, arm openness, hand velocity) on a Null CHOP, with optional parameter binding.
Create a live stick-figure skeleton from pose data in TouchDesigner. Supports synthetic, MediaPipe, OSC, or existing CHOP sources; outputs a compositable render with adjustable color, width, camera.
Create live visuals driven by full-body pose tracking: glowing marks follow 33 landmarks in points, glow, or trail styles, using synthetic, MediaPipe, OSC, or existing pose data.
Sets up real-time body tracking from a webcam in TouchDesigner: loads the MediaPipe engine, runs the timeline, and outputs a 33-landmark pose CHOP with an optional visual skeleton.
Sets up real-time face-landmark tracking from a webcam: loads MediaPipe, starts timeline, and creates an adapter CHOP outputting 468/478 landmarks (tx/ty/tz/confidence) for visualization and binding.
Configures MediaPipe selfie segmentation in TouchDesigner, generating an alpha mask (invertible/featherable) and pre-keyed person-on-transparent output for keyers or compositors.