DX – Mediapipe – QuickStart

MediaPipe Pose Tracking for Isadora – Quickstart Example
This example project demonstrates how to integrate MediaPipe Pose tracking into Isadora using the Pythoner actor.
It provides a simple, self-contained setup for real-time full-body tracking with nothing more than your webcam — no depth camera or external hardware required.
Using Python’s MediaPipe library, the patch captures 2D pose landmarks from the live video feed and sends them into Isadora as structured data you can use for creative control, visual interaction, or motion-driven effects.
What’s Included
-
Pythoner-based Pose Tracker
Captures all 33 MediaPipe Pose landmarks (head, limbs, hands, and torso) in real time. -
OpenNI-Compatible Conversion Tools
Converts MediaPipe’s full landmark set into the standard 15-joint OpenNI skeleton used by Isadora’s built-in Skeleton actors. -
Skeleton JSON Wrapping
Outputs properly formatted data for use with the Skeleton From JSON actor, allowing seamless integration with Isadora’s native skeleton system. -
Automatic Environment Setup
A simple tool automatically creates a Python virtual environment and installs the required dependencies (MediaPipe, NumPy, OpenCV).
Getting Started
-
Unzip the example package — keep all included files together in a single folder.
-
Open the Isadora project file (
.izz). -
When prompted, allow the Pythoner actor to run.
-
Enter Scene 1 and click the trigger to run the setup process.
-
This will automatically build and configure your Python environment.
-
-
When setup is complete, close and reopen Isadora to finalize.
-
Your system is now ready for real-time body tracking via your webcam.
Why Use It
This example provides a lightweight, cross-platform solution for body tracking inside Isadora — ideal for:
-
Interactive installations
-
Live VJ and performance setups
-
Motion-driven visual or lighting systems
-
Research and prototyping for gesture interfaces
With the included conversion and skeleton tools, you can go from camera feed to full Isadora skeleton control in minutes.
🔗 Watch the Quickstart on YouTube
Once you have completed the tutorial, you should look at this additional approach with Multi Body and Hand Tracking: media-pipe-pose-estimation.zip
Log in to Reply
You must be logged in to post a comment.