Depth cameras are one of the most powerful tools you can bring into a TouchDesigner project. Whether you want to track people moving through a space, generate real-time point clouds, or create interactive experiences that respond to the physical world, a depth camera opens doors that a standard webcam simply can’t. In this blog, we’ll break down what a depth camera is, how it differs from related technologies, and walk through the most common options you’ll encounter as a TD practitioner.
What Is a Depth Sensor?
A depth sensor is a camera that captures not just the 2D appearance of a scene, but also how far away every point in that scene is from the camera. Instead of returning a standard color image, it produces a depth map — a grayscale or floating-point image where each pixel’s value represents a distance in physical space. In TouchDesigner, this depth data can be used to isolate foreground objects, generate point clouds, track bodies in 3D, or feed into generative systems that react to movement and presence.
Depth Camera vs. Pose Tracking vs. LiDAR
These three terms get used interchangeably sometimes, but they refer to different things — and knowing the distinction will save you a lot of confusion when shopping.
Depth cameras
A depth camera captures a per-pixel distance map of a scene. They give you the shape and position of everything in frame, but they don’t inherently know what a person is. They’re your raw spatial data.
Pose tracking
Pose tracking uses machine learning to detect a person’s pose. It can be layered on depth camera data, but it can also be information from any image. MediaPipe uses machine learning. The hurdle of relying on pose tracing or machine learning in an installation is that it’s not reliable in low-light situations if the camera can’t identify what it’s seeing.
LiDAR
(Light Detection and Ranging) works similarly to a depth camera but uses laser pulses rather than structured light or stereo vision to measure distances. LiDAR typically offers a much longer range and performs well outdoors, making it a common choice for large-venue installations or architectural scanning.
Get Our 7 Core TouchDesigner Templates, FREE
We’re making our 7 core project file templates available – for free.
These templates shed light into the most useful and sometimes obtuse features of TouchDesigner.
They’re designed to be immediately applicable for the complete TouchDesigner beginner, while also providing inspiration for the advanced user.
Depth Sensors to Know for TouchDesigner
Kinect 2
The Kinect 2 was, for a long time, the go-to depth sensor for TouchDesigner. Before the current wave of options, it was widely considered the best accessible choice: affordable, reliable, and with tight TD integration via the Kinect TOP and Kinect CHOP. The body tracking was particularly impressive for its price point, and it became a staple of interactive installations throughout the 2010s. Today it’s still a capable sensor, but it’s harder to find since it’s been discontinued, and it only works on Windows. If you have one sitting in a drawer, it’s still worth using — just know you may be chasing down secondhand units if it ever needs replacing.
Kinect Azure
The Kinect Azure was Microsoft’s follow-up: a significant step up in depth quality, using time-of-flight technology for more accurate and longer-range sensing. In TouchDesigner, you’ll use the Kinect Azure TOP and Kinect Azure CHOP, the latter of which provides skeleton and body-tracking data. Like its predecessor, it’s Windows-only, and Microsoft has since discontinued this line as well. That said, the Kinect Azure still performs extremely well, and units are still available. It’s also notable because Orbbec now produces cameras that are fully compatible with the Kinect Azure SDK, which means you can use the Kinect Azure nodes in TD with Orbbec hardware.
The Interactive & Immersive HQ did a video comparing the two Kinects a while ago:
Orbbec
After Microsoft discontinued the Kinect line, Orbbec stepped up as the most accessible and actively developed alternative. Their Femto Bolt and Femto Mega cameras use the same underlying Microsoft ToF technology as the Kinect Azure and are fully compatible with the Kinect Azure TOP and Kinect Azure CHOP in TouchDesigner — including body tracking. Orbbec also has its own native operator, the Orbbec TOP, which supports a wider range of their cameras (including the Astra and Gemini lines) and works on both Windows and macOS. One thing to be aware of: you shouldn’t use the Orbbec TOP and the Kinect Azure TOP in the same project, as it can cause instabilities. Pick one approach and stick with it for a given patch.
The Interactive & Immersive HQ also did a video comparing the Orbbec with the Kinect:
ZED (Stereolabs)
The ZED cameras from Stereolabs take a different approach — rather than structured light or time-of-flight, they use stereo vision: two lenses, like eyes, that calculate depth by comparing slightly different views of the same scene. TouchDesigner supports several ZED models (ZED, ZED 2, ZED 2i, ZED Mini) via the ZED TOP, ZED CHOP, and ZED SOP. The ZED SOP is particularly interesting for TD work — it allows you to do real-time spatial mapping, generating actual 3D geometry from your environment. The ZED cameras are also capable of very long-range depth sensing (up to 20 meters), which makes them well-suited for large venues or outdoor installations. They require a capable NVIDIA GPU to run and are Windows-only for the full feature set, but the depth quality and range they offer is hard to beat.
The Interactive & Immersive HQ did a deep dive on the ZED camera here:
Intel RealSense
RealSense cameras have a dedicated TOP in TouchDesigner and are a solid, affordable option for short-to-medium range depth work. Intel has discontinued active development of the consumer line, so availability is becoming more limited, but they’re still widely used and easy to find secondhand.
Wrap Up
Depth cameras have become an essential part of the interactive and immersive toolkit, and TouchDesigner’s native support for most major sensors makes getting started more accessible than ever. The right sensor depends on your environment, your range requirements, and whether you need skeleton tracking or just raw depth data. Start with what fits your budget and project scope (even an older Kinect can produce compelling work), and let the sensor inform what kinds of interactions become possible in your patch.
