J’s tudio    File   Blog   CV   Contact
       
Kinect w/ Touch Designer

Jan 27th, 2026



1) Things to prepare:


1) a windows computer
2) kinect v2 and related cables

2) Install Kinect SDK 2.0


here is the link.

Writing this just in case. If you have the issue “A driver cannot load this device”, click learn more and follow the steps.

Once you did the installations, you can open the kinect app, and you will get this 3D Monitor View.


3) Install Touch Designer (?) if you don’t have it


You got this. :)

4) Test the camera in Touch Designer 


Use Kinect TOP

5) Play with Touch Designer!




find td file here

6) How does the kinect work as a 3D camera?


At a high level, a 3D camera detects distance by figuring out how far light has traveled from the camera to an object and back. The key difference between a normal camera and a 3D camera is that a normal camera only records color and brightness, while a 3D camera actively or passively reconstructs geometry.

Step 1: Kinect sends out invisible light

Step 2: The light bounces back

Step 3: Kinect measures the time

Step 4: Kinect creates a depth image





If you want to make life harder:



1) Stereo vision (depth from two eyes)


This works almost exactly like human vision.

Two cameras are placed a fixed distance apart. Each camera sees the same scene from a slightly different angle. The system looks for the same feature in both images and measures how much it shifts horizontally (this shift is called disparity).

Objects that are closer have a larger shift; objects farther away have a smaller shift. With known camera spacing and lens parameters, the system triangulates distance.

This approach is passive (no light projection needed), but it struggles with blank surfaces, low light, or scenes with little texture. It’s common in robotics, AR headsets, and computer vision research.

2) Time-of-Flight (ToF) This is the most physically direct method.

The camera emits infrared light and measures how long it takes for that light to bounce back from each point in the scene. Since the speed of light is constant, distance is calculated as:

distance = (speed of light Ă— time) Ă· 2

Modern ToF sensors don’t measure single pulses directly (the timing would be too small). Instead, they use modulated light waves and measure phase shifts or very precise timing differences.

This approach produces clean, real-time depth maps and works well for full-body tracking, gesture detection, and interactive installations. It’s the foundation of many contemporary depth sensors.

3) Structured light (project-and-observe)


Here the camera actively projects a known infrared pattern (dots or stripes) onto the scene.

The pattern deforms when it hits objects at different depths. An infrared camera observes how the pattern warps compared to a reference pattern captured on a flat surface. From this distortion, the system computes distance.

This method is very good indoors and works even on textureless surfaces. Early consumer depth cameras (including early generation Kinect-style sensors) used this technique.

Its main limitation is sensitivity to strong sunlight and a shorter effective range.