Toggle search
Toggle menu
5.1K
5.9K
229
48K
VR & AR Wiki
Toggle preferences menu
Toggle personal menu
Not logged in
Your IP address will be publicly visible if you make any edits.

Motion capture suit

From VR & AR Wiki
More actions

A motion capture suit (often shortened to mocap suit) is a garment worn by a performer or user so that the movement of the body can be recorded by a motion capture system. The suit either carries the sensing hardware itself or gives a camera system something to track. Close-fitting suits with hook-and-loop (Velcro) surfaces hold the reflective markers used in camera-based optical capture,[1] while inertial suits have inertial measurement units (IMUs) built into the fabric and calculate the wearer's pose without any cameras.[2] Earlier body-worn systems used mechanical exoskeletons with potentiometers at the joints, or electromagnetic trackers attached to a suit.[3]

Mocap suits are used to animate characters for film and video games, including virtual reality games, and to drive full-body avatars in real time. In VR research, a marker suit or an inertial suit is one of the standard ways to animate a self-avatar that follows the participant's body, which is used to study body ownership and embodiment.[4] Lighter alternatives for full-body tracking use a handful of separate body-worn sensors instead of a complete suit.[5]

This article covers the worn garment and its variants. The wider field, including markerless capture and the major studio camera systems, is covered in the motion capture article; suits whose main purpose is touch feedback are covered in Haptic suit.

Reviewed 6 October 2026. Re-checked the claims against their cited sources: Xsens, Rokoko, Sony mocopi, Teslasuit and OptiTrack pages, Sturman 1994 history, Game Developer, Machine Design, AWE, HCI Museum and Science and Media Museum pages, Spanlang 2014 and Choo 2022 full texts, and Caserman, Al-Amri, Robert-Lachaine, SIP, DIP and TransPose abstracts and metadata. About review dates.

Types

Motion capture suits are usually classified by the tracking technology they work with.

Type What the suit does Examples
Optical, passive markers A close-fitting suit with a hook-and-loop surface holds retroreflective markers at anatomical landmarks; infrared cameras around the capture volume track the markers Suits used with OptiTrack, Vicon, PhaseSpace and ART camera systems[1][4]
Optical, active markers Light-emitting diodes on the suit are sequenced and located by camera sensors The LED body suit of MIT's Graphical Marionette (1983), tracked with the Op-Eye system[3]
Electromechanical exoskeleton Rigid linkages follow the limbs, and potentiometers at the joints measure joint angles Tom Calvert's exoskeleton at Simon Fraser University (early 1980s), Animazoo Gypsy[3][6]
Electromagnetic Electromagnetic trackers are attached to a body suit The suit used to perform Mat the Ghost (1991), with Polhemus trackers on the torso, arms and head[3][4]
Inertial IMUs combining gyroscopes, accelerometers and usually magnetometers are built into a suit or worn on straps Xsens MVN, Rokoko Smartsuit Pro, Perception Neuron, Teslasuit[2][7][8][9]

How it works

Marker suits

In marker-based optical capture the suit is passive: it gives the markers a stable place to sit. OptiTrack's documentation notes that attaching markers directly to a person's skin is difficult because of hair, oil and sweat, and for that reason recommends "mocap suits that allow Velcro marker bases"; where markers must go on the skin, skin adhesives are used instead.[1] Markers are placed at the landmarks that the chosen marker set specifies, and the software builds a skeleton from the labelled markers. Misplaced markers can stop the skeleton from being created or cause labelling errors that have to be fixed in post-processing.[1] OptiTrack's Motive software, for example, offers four full-body marker sets: Conventional (39 markers), Baseline (41), Core (50) and Biomech (57).[10]

Because the cameras have to see the markers, optical capture is sensitive to occlusion, which makes it hard to track people in cluttered spaces or in close contact with one another, and passive-marker systems need tightly controlled lighting.[4]

Inertial suits

An inertial suit carries its own sensors. The 2009 technical description of the Xsens MVN suit by Daniel Roetenberg, Henk Luinge and Per Slycke gives a detailed example. The system used 17 MTx sensor modules, each containing 3D gyroscopes, accelerometers and magnetometers in a 38 by 53 by 21 mm housing weighing 30 g. The modules were daisy-chained to two Xbus Master units that synchronized sampling, supplied power and handled the wireless link to a PC. The sensors and cables were integrated into a Lycra suit, with the Xbus Masters mounted on the back, and the whole system weighed 1.9 kg including eight AA batteries. Sensors sat on the feet, lower legs, upper legs, pelvis, shoulders, sternum, head, upper arms, forearms and hands.[2]

Orientation can in principle be found by integrating gyroscope data and position by double-integrating accelerometer data, but the authors note that such uncorrected estimates are only accurate for a few seconds because of integration drift.[2] MVN therefore uses sensor fusion with a prediction step and a correction step. After a calibration in which the wearer stands in a known pose (the T-pose), the software maps each sensor onto a biomechanical model of 23 body segments. It then corrects drift using the fact that segments are joined at joints and by detecting points where the body touches the outside world, such as feet on a flat floor. The accelerometers give a gravity reference for tilt, and the magnetometers give a heading reference. Because the Earth's magnetic field is easily disturbed by metal in buildings or vehicles, the filter estimates the disturbance at each time step and warns the user when it is present. The system ran in real time at up to 120 Hz.[2]

Current inertial suits follow the same principle. Xsens's current Link system uses an "eSuit" with integrated, washable cabling and 17 click-in sensors plus one prop sensor, with an update rate of 240 Hz, a battery life of 4 hours internal or 8 hours external, and a range of 150 m. Xsens states that the eSuit cuts setup time by up to 40 percent compared with the previous generation and is qualified for up to 100 wash cycles. The company's Awinda system replaces the suit with adjustable straps carrying 17 wireless sensors plus a prop sensor, at 60 Hz.[11][12] Rokoko's Smartsuit Pro II has 9-degree-of-freedom IMUs connected to a Wi-Fi hub; Rokoko lists a sensor frame rate of 200 fps, a 3D orientation accuracy of plus or minus 1 degree and a 100 m tracking range, and says that a 10,000 mAh power bank gives roughly six or more hours of capture. The electronics can be removed so the textile can be washed.[13]

Exoskeleton and electromagnetic suits

Electromechanical suits measure joint angles directly. In David Sturman's 1994 history, Tom Calvert's early-1980s system at Simon Fraser University strapped a kind of exoskeleton to each leg with a potentiometer alongside the knee, and Pacific Data Images later built a lightweight plastic upper-body exoskeleton with potentiometers. PDI found it encumbering and electrically noisy.[3] The commercial Gypsy line from the British company Animazoo used the same approach: Gypsy5, launched in 2006, built its sensors into two exoskeleton frames fitted around the body to record the rotation of the actor's bones.[6] Electromagnetic suits instead attach magnetic trackers to the body, as in the Polhemus-equipped suit worn by the actor who performed Mat the Ghost.[3]

History

Marked garments predate computers. In his chronophotographic studies of movement, Etienne-Jules Marey dressed his models in black and marked out a "skeleton" on them with shiny buttons for the joints and metal bands between them, so that the photograph isolated the motion of the limbs.[14]

Sturman's history traces computer-based body capture to the early 1980s. Besides Calvert's exoskeleton, Carol Ginsberg and Delle Maxwell at MIT presented the Graphical Marionette in 1983, a "scripting-by-enactment" system. They wired a body suit with sequenced LEDs on the joints and other landmarks; two cameras with special photodetectors reported the 2D position of each LED, and the computer combined the two views into 3D coordinates that drove a stick figure.[3] In 1989 Kleiser-Walczak made the music-video animation Dozo with an optical system from Motion Analysis that triangulated small pieces of reflective tape placed on the body. Tracking markers through occlusions was then a slow post-process.[3] In 1991 the French producer Videosystem began performing the daily television character Mat the Ghost in real time, with an actor miming the upper-body motion "while wearing a suit with electromagnetic trackers (Polhemus) on the torso, arms, and head".[3]

Virtual reality had its own full-body garment in the same period. At VPL Research, Ann Lasko-Harvill was chief designer of data suits; she led development of the company's "Reality Built for Two" (RB2) whole-body VR suit, developed around 1989 for multi-user interaction in virtual spaces, and designed the DataSuit, which the AWE Hall of Fame describes as a full-body suit with sensors for tracking arm and leg movement.[15] The HCI Museum's VPL exhibit, which focuses on the EyePhone and the DataGlove, notes that the DataSuit extended VPL's approach to full-body motion capture.[16]

Inertial suits became commercial products in the 2000s. In 2007 Machine Design described Animazoo's IGS-190, which placed 19 small inertial gyroscopes on an actor's body suit and radioed the data to a receiver for real-time mapping onto a skeleton; prices started at US$39,600, and the magazine listed Rockstar North, the BBC and RAI among its users.[17] Xsens published the technical description of its MVN suit in 2009.[2] Lower-priced suits for independent creators followed in the 2010s. When TechCrunch covered the Rokoko Smartsuit Pro in 2017, it cost US$2,500 and had 19 sensors embedded in the suit, and the article contrasted it with the "industry-standard green man MoCap suit" used by Hollywood studios.[7] Rokoko announced the Smartsuit Pro II on 9 November 2021 at US$2,745 (or 2,745 euros), with deliveries expected in January 2022. The company listed elevation tracking for stairs and ladders, better handling of high-impact motion, reduced drift, and native connection to its Smartgloves as the main changes.[18]

Commercial systems

The table lists representative suits and strap systems with manufacturer-stated figures.

System Maker Technology Stated configuration
Xsens Link (next generation) Xsens Inertial, wired eSuit 17 click-in sensors plus 1 prop sensor, 240 Hz, range 150 m[11]
Xsens Awinda Xsens Inertial, wireless straps 17 wireless sensors plus 1 prop sensor, 60 Hz, range 50 m[11]
Smartsuit Pro II Rokoko Inertial suit with Wi-Fi hub 9-DoF IMUs, 200 fps, range 100 m[13]
Perception Neuron Noitom Inertial 17 IMU sensors in the full-body configuration used in a 2022 validation study[8]
Teslasuit Teslasuit Inertial, combined with haptics and biometrics 14 IMU sensors (6-axis or 9-axis modes), 100 frames per second, 80 haptic channels[9]
OptiTrack suits OptiTrack Passive optical markers Velcro-compatible suit; Motive full-body marker sets of 39 to 57 markers[1][10]

Applications in VR and AR

VR content production

Suits are used to animate characters for VR games. In a 2019 Xsens case study, Survios animation director John Kim described using an Xsens MVN suit "a few times a month" while making the boxing game Creed: Rise to Glory, dropping captured performances into Unreal Engine to view in VR the same day: "Xsens allows us to capture an idea and put it in virtual reality in the same day."[19] Xsens markets its software for streaming suit data directly into Unreal Engine and Unity, lists a latency of 20 ms, and offers "3D positional aiding with HTC Vive" in its Link bundle, in which HTC Vive tracking aids the suit's position estimate.[20] The same page describes live uses in which actors in Xsens suits drove animated characters in real time, including two cartoon characters produced by Psyop to answer live questions online.[20]

Live avatars and VTubing

Real-time suit data can drive a streamer's avatar. The VTuber CodeMiko, a 3D character built in Unreal Engine, is controlled by its creator through an Xsens motion capture suit; designboom reported in 2021 that such a suit "might cost upwards of $30,000".[21] Lower-cost alternatives skip the suit entirely: Sony's mocopi uses six 8 g sensors attached with Velcro bands and a clip to the head, hip, wrists and ankles, pairs with a smartphone over Bluetooth, and works with VRChat, Unity, MotionBuilder and Unreal Engine; in the United States it was offered for pre-order at about US$449 from 29 June 2023, with shipments from 14 July 2023.[5]

Embodiment research

In their 2014 guide to building a VR embodiment lab, Spanlang and colleagues at Mel Slater's group explain that body movements can animate the virtual character that represents the self, giving visuo-motor feedback that is "a powerful tool" for inducing body ownership illusions. For experiments in which the participant wears a head-mounted display, they recommend either a marker-based optical system (OptiTrack, PhaseSpace, ART or Vicon) or "an inertial suit (such as the Xsens MVN)", and they note that putting on and removing the suit or markers and calibrating the participant "can add up to 20 min" to an experiment.[4] They contrast the two families: inertial suits work indoors and outdoors, need little computing power and cover large areas, but have lower positional accuracy, drift that compounds over time, and drift correction that relies on magnetic north and can be upset by metal structures; the suit is also more cumbersome to put on.[4]

A 2020 systematic review in IEEE Transactions on Visualization and Computer Graphics by Polona Caserman, Augusto Garcia-Agundez and Stefan Göbel examined 53 studies that track a head-mounted-display user and show a full-body avatar, grouped into markerless, marker-based and IMU-based systems. It found that a full-body avatar can enhance the sense of embodiment and immersion, that the Kinect was still the most frequently used sensor (27 of 53), and that many multiplayer studies (7 of 17) used marker-based systems because they track several users' full-body movements more robustly.[22]

Combined haptic and capture suits

Some VR suits combine capture with feedback. The Teslasuit uses 14 IMUs to track the wearer's movement alongside 80 channels of electrical stimulation and a photoplethysmography sensor for pulse, and Teslasuit lists animation, sports performance monitoring, ergonomics testing and XR training as uses.[9] Suits built mainly for touch feedback are described in the haptic suit article.

Accuracy

Validation studies compare inertial suits with optical marker systems, which are treated as the reference. Xavier Robert-Lachaine and colleagues fixed an Optotrak marker cluster on every IMU of a full-body Xsens system worn by 12 participants. They reported a mean joint-angle RMSE of 1.2 degrees during short functional movements and 2.8 degrees during long manual-handling tasks. Differences in biomechanical model definition caused more of the discrepancy than the sensors themselves, and the technological error stayed under 5 degrees during handling tasks.[23] A 2018 study of the Xsens MVN BIOMECH system with 26 participants found excellent validity in the sagittal plane for hip, knee and ankle angles during walking, squatting and jumping, and acceptable validity in the frontal and transverse planes for squats and jumps.[24]

Lower-cost suits show larger errors on some joints. Choo, Chow and Komar compared a 17-sensor Perception Neuron setup with an eight-camera Vicon system recording at 120 Hz during walking, jogging and floorball shots. Most joint angles had an RMSE below 4 degrees, with an average Pearson correlation of 0.85, but shoulder abduction and adduction errors ranged from 5.36 to 15.15 degrees. The authors concluded that the system may not replace traditional motion analysis when raw joint angles must be reproduced.[8]

Research

Much research aims to reduce the number of sensors a wearer needs. Daniel Vlasic and colleagues at MIT, Mitsubishi Electric Research Laboratories and ETH Zurich presented a wearable system in 2007 that combined ultrasonic time-of-flight and inertial measurements from inexpensive sensors worn on the garment, fused with an extended Kalman filter. They reported that the acoustic data reduced the drift seen in purely inertial systems, though the prototype did not reliably recover the body's global position and orientation, and suggested it could become an input device for augmented reality applications.[25]

Later work replaced the full suit with six IMUs. Sparse Inertial Poser (von Marcard, Rosenhahn, Black and Pons-Moll, 2017) fitted a statistical body model to six sensors on the wrists, lower legs, back and head using multi-frame optimization.[26] Deep Inertial Poser (Huang and colleagues, 2018) used a recurrent neural network to reconstruct full-body pose from six IMUs in real time; its DIP-IMU evaluation dataset was recorded with ten subjects wearing 17 IMUs.[27] TransPose (Yi, Zhou and Xu, 2021) estimated both body pose and global translation from six IMUs at over 90 fps.[28] Sony's consumer mocopi system, described above, also uses six body-worn sensors.[5]

See also

References

  1. ↑ 1.0 1.1 1.2 1.3 1.4 "Skeleton Tracking". OptiTrack Documentation. NaturalPoint. https://docs.optitrack.com/motive/skeleton-tracking. Retrieved 2026-10-06.
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 Daniel Roetenberg, Henk Luinge, Per Slycke (2009-04-08). "Xsens MVN: Full 6DOF Human Motion Tracking Using Miniature Inertial Sensors". Xsens Technologies technical paper (version 8 April 2009). Xsens Technologies B.V.. https://www.semanticscholar.org/paper/Xsens-MVN:-Full-6DOF-Human-Motion-Tracking-Using-Roetenberg-Luinge/ffd2416057dc96d69ab6ed047910606f270039be. Retrieved 2026-10-06.
  3. ↑ 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 David J. Sturman (1994). "A Brief History of Motion Capture for Computer Character Animation". SIGGRAPH 94, Course 9: Character Motion Systems (ACM SIGGRAPH Education HyperGraph). https://education.siggraph.org/static/HyperGraph/animation/character_animation/motion_capture/history1.htm. Retrieved 2026-10-06.
  4. ↑ 4.0 4.1 4.2 4.3 4.4 4.5 Bernhard Spanlang, Jean-Marie Normand, David Borland, Konstantina Kilteni, Elias Giannopoulos, Ausiàs Pomés, Mar González-Franco, Daniel Perez-Marcos, Jorge Arroyo-Palacios, Xavi Navarro Muncunill, Mel Slater (2014-11-27). "How to Build an Embodiment Lab: Achieving Body Representation Illusions in Virtual Reality". Frontiers in Robotics and AI, vol. 1, article 9. doi:10.3389/frobt.2014.00009. https://www.frontiersin.org/journals/robotics-and-ai/articles/10.3389/frobt.2014.00009/full. Retrieved 2026-10-06.
  5. ↑ 5.0 5.1 5.2 "Sony Electronics Announces Mobile Motion Capture System "mocopi"". Alpha Universe. Sony Electronics. 2023-06-29. https://alphauniverse.com/stories/sony-electronics-announces-mobile-motion-capture-system--mocopi-/. Retrieved 2026-10-06.
  6. ↑ 6.0 6.1 "Animazoo Launches Gypsy5 Motion Capture Systems". Game Developer. 2006-03-10. https://www.gamedeveloper.com/game-platforms/animazoo-launches-gypsy5-motion-capture-systems. Retrieved 2026-10-06.
  7. ↑ 7.0 7.1 Sarah Buhr (2017-07-05). "Make Hollywood-quality animations at low-budget prices with this motion capture suit". TechCrunch. https://techcrunch.com/2017/07/05/make-hollywood-quality-animations-at-low-budget-prices-with-this-motion-capture-suit/. Retrieved 2026-10-06.
  8. ↑ 8.0 8.1 8.2 Corliss Zhi Yi Choo, Jia Yi Chow, John Komar (2022). "Validation of the Perception Neuron system for full-body motion capture". PLOS ONE, vol. 17, no. 1, e0262730. doi:10.1371/journal.pone.0262730. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0262730. Retrieved 2026-10-06.
  9. ↑ 9.0 9.1 9.2 "Full Body VR Haptic Suit with Motion Capture". Teslasuit. https://teslasuit.io/products/teslasuit-4/. Retrieved 2026-10-06.
  10. ↑ 10.0 10.1 "Full Body Marker Sets". OptiTrack Documentation. NaturalPoint. https://docs.optitrack.com/markersets/full-body. Retrieved 2026-10-06.
  11. ↑ 11.0 11.1 11.2 "Motion Capture". Xsens. https://www.xsens.com/products/motion-capture. Retrieved 2026-10-06.
  12. ↑ 13.0 13.1 "Smartsuit Pro II - Professional body motion capture in one mobile mocap suit". Rokoko. https://www.rokoko.com/products/smartsuit-pro. Retrieved 2026-10-06.
  13. ↑ "Art imitates life: The surprising origins of motion capture". National Science and Media Museum. 2023-11-15. https://www.scienceandmediamuseum.org.uk/objects-and-stories/surprising-origins-motion-capture. Retrieved 2026-10-06.
  14. ↑ "Ann Lasko-Harvill - Chief Designer of Data Suits at VPL Research". AWE XR Hall of Fame. Augmented World Expo. https://www.awexr.com/hall-of-fame/24-ann-lasko-harvill. Retrieved 2026-10-06.
  15. ↑ "VPL EyePhone & DataGlove". HCI Museum. https://interfacemuseum.com/exhibits/vpl-research/. Retrieved 2026-10-06.
  16. ↑ "Gyro suit captures motion". Machine Design. 2007-12-01. https://www.machinedesign.com/archive/article/21828834/gyro-suit-captures-motion. Retrieved 2026-10-06.
  17. ↑ "Rokoko Betting on the Creator Economy With Their New Smartsuit Pro II". GlobeNewswire. Rokoko. 2021-11-09. https://www.globenewswire.com/news-release/2021/11/09/2330983/0/en/Rokoko-Betting-on-the-Creator-Economy-With-Their-New-Smartsuit-Pro-II.html. Retrieved 2026-10-06.
  18. ↑ "Creed: Rise to Glory delivers a VR knockout with Xsens". Xsens. 2019-03-14. https://www.xsens.com/resources/cases/creed-rise-to-glory-delivers-a-vr-knockout-with-xsens. Retrieved 2026-10-06.
  19. ↑ 20.0 20.1 "VR, AR & XR". Xsens (Movella). https://www.xsens.com/entertainment/vr-ar-xr-solution. Retrieved 2026-10-06.
  20. ↑ Kat Barandy (2021-01-12). "technician and streamer codemiko is revolutionizing the digital space". designboom. https://www.designboom.com/technology/codemiko-virtual-streamer-twitch-01-12-2021/. Retrieved 2026-10-06.
  21. ↑ Polona Caserman, Augusto Garcia-Agundez, Stefan Göbel (2020). "A Survey of Full-Body Motion Reconstruction in Immersive Virtual Reality Applications". IEEE Transactions on Visualization and Computer Graphics, vol. 26, no. 10, pp. 3089-3108. doi:10.1109/TVCG.2019.2912607. https://doi.org/10.1109/TVCG.2019.2912607. Retrieved 2026-10-06.
  22. ↑ Xavier Robert-Lachaine, Hakim Mecheri, Christian Larue, André Plamondon (2017). "Validation of inertial measurement units with an optoelectronic system for whole-body motion analysis". Medical & Biological Engineering & Computing, vol. 55, no. 4, pp. 609-619. doi:10.1007/s11517-016-1537-2. https://doi.org/10.1007/s11517-016-1537-2. Retrieved 2026-10-06.
  23. ↑ Mohammad Al-Amri, Kevin Nicholas, Kate Button, Valerie Sparkes, Liba Sheeran, Jennifer L. Davies (2018). "Inertial Measurement Units for Clinical Movement Analysis: Reliability and Concurrent Validity". Sensors, vol. 18, no. 3, 719. doi:10.3390/s18030719. https://www.mdpi.com/1424-8220/18/3/719. Retrieved 2026-10-06.
  24. ↑ Daniel Vlasic, Rolf Adelsberger, Giovanni Vannucci, John Barnwell, Markus Gross, Wojciech Matusik, Jovan Popović (2007). "Practical Motion Capture in Everyday Surroundings". ACM Transactions on Graphics, vol. 26, no. 3, article 35 (SIGGRAPH 2007). doi:10.1145/1276377.1276421. https://people.csail.mit.edu/drdaniel/research/vlasic-2007-pmc.pdf. Retrieved 2026-10-06.
  25. ↑ Timo von Marcard, Bodo Rosenhahn, Michael J. Black, Gerard Pons-Moll (2017). "Sparse Inertial Poser: Automatic 3D Human Pose Estimation from Sparse IMUs". Computer Graphics Forum, vol. 36, no. 2, pp. 349-360. doi:10.1111/cgf.13131. https://arxiv.org/abs/1703.08014. Retrieved 2026-10-06.
  26. ↑ Yinghao Huang, Manuel Kaufmann, Emre Aksan, Michael J. Black, Otmar Hilliges, Gerard Pons-Moll (2018). "Deep Inertial Poser: Learning to Reconstruct Human Pose from Sparse Inertial Measurements in Real Time". ACM Transactions on Graphics, vol. 37, no. 6. doi:10.1145/3272127.3275108. https://arxiv.org/abs/1810.04703. Retrieved 2026-10-06.
  27. ↑ Xinyu Yi, Yuxiao Zhou, Feng Xu (2021). "TransPose: Real-time 3D Human Translation and Pose Estimation with Six Inertial Sensors". ACM Transactions on Graphics, vol. 40, no. 4. doi:10.1145/3450626.3459786. https://arxiv.org/abs/2105.04605. Retrieved 2026-10-06.
Retrieved from "https://vrarwiki.com/index.php?title=Motion_capture_suit&oldid=47486"