Microsoft Kinect
| Microsoft Kinect | |
|---|---|
| Basic Info | |
| VR/AR | Augmented Reality, Virtual Reality (research and accessory use) |
| Type | Depth-sensing camera and motion input device |
| Subtype | RGB-D camera with markerless skeletal tracking |
| Platform | Xbox 360, Xbox One, Windows, Linux |
| Creator | Microsoft |
| Developer | Microsoft, PrimeSense (original sensor) |
| Manufacturer | Microsoft |
| Announcement Date | June 1, 2009 (as "Project Natal") |
| Release Date | November 4, 2010 (Kinect for Xbox 360) |
| Price | $149.99 (Kinect for Xbox 360, US launch) |
| Website | https://learn.microsoft.com/en-us/azure/kinect-dk/ |
| Versions | Kinect for Xbox 360 (2010), Kinect for Windows (2012), Kinect for Xbox One / Kinect v2 (2013), Kinect for Windows v2 (2014), Azure Kinect DK (2019) |
| System | |
| Operating System | Xbox 360 system software, Xbox One system software, Windows, Linux |
| Storage | |
| Display | |
| Display | N/A |
| Resolution | Depth 640x480 @ 30 Hz (v1); 512x424 depth, 1920x1080 RGB (v2); 1 MP ToF depth, 12 MP RGB (Azure Kinect) |
| Refresh Rate | 30 Hz (depth) |
| Image | |
| Field of View | N/A |
| Optics | |
| Optics | Infrared projector and camera, RGB camera |
| Passthrough | N/A |
| Tracking | |
| Tracking | Markerless skeletal tracking (20 joints, up to 2 active players in v1; 25 joints, up to 6 skeletons in v2) |
| Eye Tracking | N/A |
| Face Tracking | Yes (Kinect v2: facial expression and head orientation) |
| Hand Tracking | Yes (gesture and open/closed hand state) |
| Body Tracking | Yes (markerless full-body skeletal tracking) |
| Rotational Tracking | N/A (fixed sensor) |
| Positional Tracking | Yes (depth-based, of subjects in front of the sensor) |
| Audio | |
| Audio | N/A |
| Microphone | Multi-microphone array (4 capsules in v1; 7-microphone array in Azure Kinect DK) |
| Camera | RGB camera plus infrared depth camera |
| Connectivity | |
| Connectivity | USB (proprietary connector on Xbox models; USB-C on Azure Kinect DK) |
| Device | |
| Sensors | RGB camera, infrared depth camera, microphone array, accelerometer (v1) / IMU (v2 and Azure Kinect DK) |
| Input | Voice, gesture, and full-body motion |
Property "VR/AR" (as page type) with input value "Virtual Reality]] (research and accessory use)" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process. Property "Type" (as page type) with input value "Depth sensing|Depth-sensing camera]] and motion input device" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process. Property "Developer" (as page type) with input value "PrimeSense]] (original sensor)" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.
The Microsoft Kinect is a family of depth-sensing cameras and markerless motion-input devices developed by Microsoft. The first model, Kinect for Xbox 360, was previewed at the E3 trade show in June 2009 under the code name "Project Natal", renamed Kinect at E3 in June 2010, and released on November 4, 2010 as a controller-free accessory for the Xbox 360 game console.[1][2][3] Each Kinect combines a color (RGB) camera, an infrared depth sensor, and a microphone array, and uses computer-vision software to recognize a user's body, gestures, and voice without a handheld controller.[4][1]
Although Kinect was created for gaming, its low cost and built-in markerless skeletal tracking made it a widely used sensor in Virtual Reality and Augmented Reality research, where it has been applied to full-body avatars, motion capture, and rehabilitation systems.[5][6] The third generation, the Azure Kinect DK (announced 2019), used the same time-of-flight depth sensor as Microsoft's HoloLens 2 augmented reality headset.[7][8]
Microsoft discontinued consumer Kinect hardware for the Xbox One in October 2017 after selling about 35 million units across all versions, and ended sales of the Azure Kinect DK, the last device in the line, in October 2023.[3][8]
History and development
The depth sensor in the original Kinect came from the Israeli 3D-sensing company PrimeSense, which developed a depth-sensing method that projected a near-infrared pattern onto a scene and read the distorted pattern back with a standard image sensor to compute distance.[9][10] PrimeSense licensed the hardware design and chip used in Microsoft's Kinect for the Xbox 360.[9][11]
Microsoft previewed the device as "Project Natal" at the E3 trade show in June 2009 and announced the retail name "Kinect" at E3 in June 2010, demonstrating a camera that recognized joints of the human body and mapped them to an on-screen avatar or in-game character without a held controller.[1][2] Kinect for Xbox 360 went on sale in November 2010 and became the fastest-selling consumer electronics device on record, a milestone Microsoft cited again when it later reported total Kinect sales.[3]
In November 2013, Apple acquired PrimeSense for a reported figure of around $360 million, after which the original sensor supplier was absorbed into Apple.[9][11]
Versions
| Model | Released | Depth method | Notable specifications |
|---|---|---|---|
| Kinect for Xbox 360 | November 2010 | PrimeSense structured light (light coding) | 640x480 depth and RGB at 30 Hz; multi-capsule microphone array; motorized tilt; markerless body and joint tracking[1][10] |
| Kinect for Windows | February 2012 | Structured light | PC version with a near-range mode and an official Windows SDK for developers[4] |
| Kinect for Xbox One (Kinect v2) | November 2013 | Time-of-flight | 1080p RGB camera; higher-resolution time-of-flight depth; tracks more joints per person than v1, plus facial expression and hand state[10][4] |
| Kinect for Windows v2 | 2014 | Time-of-flight | Windows version of the Xbox One sensor[4] |
| Azure Kinect DK | March 2020 (announced February 24, 2019) | Time-of-flight (shared with HoloLens 2) | 1 MP ToF depth camera, 12 MP RGB camera, 7-microphone array, IMU; $399; aimed at developers and enterprise rather than gaming[7][8] |
The first two generations used PrimeSense structured-light sensing, in which a fixed infrared pattern is projected and its deformation is measured to build a depth image.[9][10] Starting with Kinect for Xbox One in 2013, Microsoft switched to a time-of-flight sensor, which measures depth from the phase shift of reflected infrared light and supported higher-resolution skeletal tracking.[10]
How it works
A Kinect sensor produces three synchronized data streams: a color image, a depth image (a per-pixel distance map of the scene), and audio from its microphone array. The depth image is generated by an infrared subsystem, structured light in the first two generations and time-of-flight from Kinect v2 onward.[4][10] Microsoft's runtime software then fits a skeleton to the depth data, estimating the 3D position of a person's joints in real time without any worn markers or controllers, a technique known as markerless motion capture or skeletal tracking.[4][5]
For the original Kinect, the system recognized dozens of joints on the human body and matched them to an on-screen avatar or in-game character in real time, tracking more than one person at once.[1] Kinect for Xbox One increased the resolution of this tracking and added facial-expression and hand-state detection.[10][4] The microphone array supports voice commands and speech recognition, and the original sensor included a motorized tilt to adjust its vertical aim.[1]
VR and AR relevance
Kinect became a common research and prototyping tool for virtual reality and augmented reality because it offered depth sensing and markerless full-body motion capture at consumer prices.[5][6] Its skeletal data has been used to drive avatars, to add full-body tracking beyond the head and hands provided by a headset and controllers, and to capture user motion for Locomotion and interaction studies; one approach fuses several Kinect units to improve full-body tracking for VR-aided assembly simulation.[6] Researchers have also published accuracy and precision comparisons of the Kinect v1, Kinect v2, and Azure Kinect for skeleton tracking, which has informed their use in clinical and laboratory settings.[5]
A frequent application is VR rehabilitation. Because a depth camera captures motion without markers or worn sensors, systems built on Kinect let patients perform exercises at home or in clinics inside game-like virtual environments; studies have evaluated Kinect skeletal tracking in VR systems for upper-limb hemiparesis, among other conditions.[12][5]
The connection to AR is most direct in the Azure Kinect DK, announced alongside the HoloLens 2 headset on February 24, 2019. Microsoft built the Azure Kinect around the same 1-megapixel time-of-flight depth sensor that gives HoloLens 2 its environmental and human understanding, so the developer kit effectively exposed the HoloLens 2 depth technology as a standalone PC peripheral.[7][8] The Azure Kinect shipped with a Body Tracking SDK and was aimed at developers building computer-vision, robotics, and spatial-computing applications rather than at gamers.[8][4]
Discontinuation
Microsoft discontinued each generation of Kinect in turn. In 2017, Microsoft confirmed it had stopped manufacturing the Kinect for Xbox One, with retailers selling remaining stock; the company said about 35 million Kinect units had been sold to that point.[3]
The Azure Kinect DK was the final device in the family. In August 2023 Microsoft announced that the Azure Kinect DK would be discontinued, with the hardware available to buy only until October 2023.[8] Microsoft said it would license the underlying time-of-flight depth-sensing technology to partners building similar products, and depth-camera maker Orbbec became a manufacturer continuing the technology, while the Azure Kinect SDK remained available for download.[8][13]
See also
- Depth sensing
- Time-of-flight camera
- HoloLens
- PrimeSense
- Motion capture
- Markerless tracking
- Leap Motion
References
- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 Template:Cite news
- ↑ 2.0 2.1 Template:Cite news
- ↑ 3.0 3.1 3.2 3.3 Template:Cite news
- ↑ 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7 "Azure Kinect DK documentation". 2023. https://learn.microsoft.com/en-us/azure/kinect-dk/.
- ↑ 5.0 5.1 5.2 5.3 5.4 (2021). "Skeleton Tracking Accuracy and Precision Evaluation of Kinect V1, Kinect V2, and the Azure Kinect".{Template:Journal. https://www.mdpi.com/2076-3417/11/12/5756. Retrieved 2026-06-15.
- ↑ 6.0 6.1 6.2 (2022). "Multi-Kinects fusion for full-body tracking in virtual reality-aided assembly simulation".{Template:Journal. https://journals.sagepub.com/doi/full/10.1177/15501329221097591. Retrieved 2026-06-15.
- ↑ 7.0 7.1 7.2 Template:Cite news
- ↑ 8.0 8.1 8.2 8.3 8.4 8.5 8.6 Template:Cite news
- ↑ 9.0 9.1 9.2 9.3 Template:Cite news
- ↑ 10.0 10.1 10.2 10.3 10.4 10.5 10.6 (2015). "Kinect Range Sensing: Structured-Light versus Time-of-Flight Kinect".{Template:Journal. https://arxiv.org/pdf/1505.05459. Retrieved 2026-06-15.
- ↑ 11.0 11.1 Template:Cite news
- ↑ "Evaluation of Kinect skeletal tracking in a virtual reality rehabilitation system for upper limb hemiparesis". 2013. https://ieeexplore.ieee.org/document/6662084/.
- ↑ Template:Cite news