Motion Capture History, Technologies and Applications
Advanced Computing Center for the Arts and Design Ohio State University
Vita Berezina-Blackburn
©2003- 2016, The Ohio State University Motion Capture
• motion capture (mocap) is sampling and recording motion of humans, animals and inanimate objects as 3d data for analysis, playback and remapping
• performance capture is acting with motion capture in film and games • motion tracking is real-time processing of motion capture data
©2003- 2016, The Ohio State University History of Motion Capture
• Eadweard Muybridge (1830-1904)
• Etienne-Jules Marey (1830-1904)
• Nikolai Bernstein (1896-1966)
• Harold Edgerton (1903-1990)
• Gunnar Johansson (1911- 1998)
©2003- 2016, The Ohio State University Eadward Muybridge
• the flying horse • 20,000 photos of animal and human locomotion • UK-USA, 1872
© Kingston Museum
©2003- 2016, The Ohio State University Eadward Muybridge
• zoopraxiscope
© Kingston Museum ©2003- 2016, The Ohio State University Etienne-Jules Marey
• first person to analyze human and animal motion with film • created chronophotographic gun and fixed plate camera • France, 1880s
©2003- 2016, The Ohio State University Modern Art
• Futurism (Boccionni, Balla and others)
• Marcel Duchamp
©2003- 2016, The Ohio State University Rotoscoping
• allowed animators to trace cartoon characters over photographed frames of live performances.
• invented in 1915 by Max Fleischer
• Koko the Clown
• Snow White
© Walt Disney
©2003- 2016, The Ohio State University Nikolai Bernstein
• General Biomechanics – 1924, Central Institute of Labor, Moscow
• physiology of sport and labor activities, foundations of ergonomics • cyclography • concepts of degrees of freedom and hierarchical structure of motion control
©2003- 2016, The Ohio State University Harold Edgerton
• electronic stroboscope and flash • exposures of 1/1000th to 1/1000000 sec • MIT, 1930s-1960
© Palm Press Inc. ©2003- 2016, The Ohio State University GUNNAR JOHANSSON
• Visual perception of biological motion, experimental psychology, 1970s, University of Uppsala, Sweden • Retro-reflective patches on joints • Video recording instead of film, search light mounted very closely to the camera lens, light reflects from patches into the lens • Computer modeling of motion variations
©2003- 2016, The Ohio State University 1980’s Computer Graphics
• military and medical research purposes
• first computer graphics use in research labs
• first production use o Brilliance by Robert Abel , brute force animation technique(1985 Superbowl ad) o Waldo C. Graphic (1988) PDI for Jim Henson tour o Mike the Talking Head (Siggraph 88) ’ o Don t Touch Me (1989)
©2003- 2016, The Ohio State University Mocap Technologies
ACTIVE PASSIVE
•electromechanical •optical: retroreflective markers •optical fiber •acoustic •optical: strobing LEDs •optical markerless (video based) •acoustic •inertial •optical markerless based on structured light •optical markerless based on video
©2003- 2016, The Ohio State University Optical motion capture systems
• light weight, variable size, retro- reflective markers
• VGA to16 megapixel resolution cameras with strobing LEDs digitize different views of performance
• up to 5000fps
• under 1mm accuracy
• marker occlusion
• capture volume limits
VICON NATURAL POINT MOTION ANALYSIS QUALISYS
©2003- 2016, The Ohio State University Strobing LED marker system
• red or Infrared LEDs
• unique strobing frequency for each marker
• no marker swapping
• limited volume
• limited capture time due to battery life for LED
• wires running up and down capture subject
PHASESPACE
©2003- 2016, The Ohio State University Electromechanical suits
• linked structures • potentiometers determine degree of rotation for each link • no occlusion • no magnetic or electrical interference • unlimited capture volume • low cost
• no global translation • restricted movement • fixed configuration of sensors • low sampling rate • inaccurate joints
GYPSY MOCAP SYSTEM
©2003- 2016, The Ohio State University Inertial systems
• inertial trackers placed on joints • measures orientation and position with accelerometers, gyroscopes, magnetometers on each segment • UWB RF for position tracking • unlimited capture volume • no occlusion, multiple subjects
• positional drift • translational data needs to be collected separately • battery packs and wires on the performer’s body.
XSENS MOCAP SYSTEM
©2003- 2016, The Ohio State University Electromagnetic systems
• electromagnetic sensors placed on joints or other critical points • measures orientation and position of sensor relative to electromagnetic field generated by the transmitter • no sight line requirements
• no occlusion, multiple subjects • electromagnetic interference, small volume if body translation tracking is needed
ASCENSION-TECH NORTHERN DIGITAL
©2003- 2016, The Ohio State University Optical fiber system
• fiber-optic sensor
• bend and twist sensors measure transmitted light
• no occlusion
• flexible capture volume
• adjustment to individual proportions is limited
• less accurate data
CYBERGLOVE ©2003- 2016, The Ohio State University Acoustic system
• set of transducers/transcievers generate and evaluate high frequency sound wave
• other sounds in frequency range can disrupt capture
• accuracy not as high as other systems
INTERSENSE
©2003- 2016, The Ohio State University Markerless Motion Capture
Full Body o Max Plank Institute research (3d scanner + silhouette analysis from video) o Captury
©2003- 2016, The Ohio State University Markerless Motion Capture
Full Body
o Kinect and other RGB-d sensor development
ORGANIC MOTION ILM and ManhattanMocap Group’s Multitrack System (markers for computer vision)
©2003- 2016, The Ohio State University Markerless Motion Capture
Face
FACS/Paul Ekman
Video based:
Original R&D: Digital Emily Project Faceware
Medusa (Disney Zurich)
RGB-d based:
Faceshift
©2003- 2016, The Ohio State University Markerless Motion Capture
Hands
Leap Sensor
©2003- 2016, The Ohio State University Video-based Motion Analysis
Research Areas o equipment and subject calibration o motion tracking o 3D movement reconstruction (markerless motion capture) o skeletal solving o action recognition o 3D surface reconstruction (surface scanning)
Challenges: o complex environment variability o body segmentation o occlusion o data volume
©2003- 2016, The Ohio State University Typical Marker Based Optical Motion Capture Pipeline
• planning (performers and actions, props, space requirements)
• recording point data(Vicon Blade)
• data processing, realtime or post standard skeletal solving (Vicon Blade, MotionBuilder, Ikinema )
• skeleton creation (3d animation software)
• remapping standard skeletal motion to customized characters (MotionBuilder)
• binding skeleton to a model (3D animation software)
©2003- 2016, The Ohio State University Optical Marker Based 3D Motion Reconstruction
• Single camera o Model assumptions required
• Multiple cameras o Require at least 2 cameras, unique with 3 o Camera calibration
• Motion capture with markers o Use retroreflective markers to simplify video information
©2003- 2016, The Ohio State University Problems Related to Marker Occlusion
©2003- 2016, The Ohio State University Skeletal Solving (remapping mocap data to a character model) • how to make markers move a skeleton o photo reference or 3d scan of a performer o CG model o Motion Builder or Ikinema o Vicon Blade o other methods
• problems with detecting joint centers…
• organization of joint hierarchies
©2003- 2016, The Ohio State University Planning
• shot list • performance space dimensions • interactions in shot • shots to be blended or looped • length of shots • size and location of props • gross proportional differences for retargeting • camera motion
©2003- 2016, The Ohio State University Planning
• Character/Prop setup o target skeleton/character topology o ready stance considerations o space preparation/occlusion removal/ camera stability
• Marker setup o marker redundancy o three markers per segment o place markers close to bone o asymmetry o recognizable configuration
• output format • file naming conventions • frame rate • target software platform • database management • potential technical issues
©2003- 2016, The Ohio State University Virtual Production
• Pioneered for the production of James Cameron’s “Avatar” • virtual camera • simulcam
©2003- 2016, The Ohio State University Feature Films, Games and VR applications
• Avatar • Dawn of the Planet of the Apes
• The Force Awakens • Curious Case of Benjamin Button
• EA Sports football capture session • EA Sports soccer capture session
• ILMxLab • PrioVR • Sixense
• VR news
©2003- 2016, The Ohio State University Applications
• Biomedical and Physical Rehabilitation Mixed Reality Rehabilitation Markerless Gait Analysis Tongue Capture for Speech Therapy
©2003- 2016, The Ohio State University Applications
• Historical Preservation Native American Performance
• Arts Open Ended Group Walking City ACCAD Motion Lab Deakin University Motion Lab Projects Virtual Puppets in Landing Place Robotic Camera Choreography via Motion Capture
©2003- 2016, The Ohio State University Applications
• Life Sciences
• Engineering
• Military and Law Enforcement VR weapon training with acoustic tracking system Virtual Crime Scene Simulation
• Sports Golf Training Simulator Various Sports Analysis and Training
©2003- 2016, The Ohio State University Motion Technology and Integration Researchers
• Univesity of South California (Paul Debevec) • Chris Bregler (NYU, Stanford, Google) • Carnegie Mellon (Jessica Hodgins) • Max Planck Center (Christian Theobalt) • Stanford (Vladlen Koltun) • Synlab at Georgia Tech (Ali Mazalek)
©2003- 2016, The Ohio State University Mocap Studios
Giant Studios
Capture Lab
WETA
House of Moves
Imaginarium Studios
Jim Henson Digital Studio
ILMxLab
©2003- 2016, The Ohio State University References
1. Menache, Alberto, “Understanding Motion Capture for Computer Animation and Video Games” 2. http://www.kingston.ac.uk/Muybridge/ 3. http://www.anotherscene.com/cinema/firsts/marey.html 4. http://cmp1.ucr.edu/exhibitions/edgerton/edgerton.html 5. Siggraph 2001 Course 51: 6. http://www.metamotion.com 7. http://gvv.mpi-inf.mpg.de/files/pami2013/jgall_motioncapture_multiple_pami13.pdf 8. http://dl.acm.org/citation.cfm?id=2614176 9. http://www.utdallas.edu/~xxg061000/tongue.pdf 10.http://en.wikipedia.org/wiki/Nikolai_Bernstein 11.http://masgutovamethod.com/content/overlays/nikolai-bernstein.html 12. http://www.theartstory.org/movement-futurism.htm 13. Reinhard Klette and Garry Tee “Understanding Human Motion: A Historic Review” 14. Johansson, Gunnar. “Visual perception of biological motion and a model for its analysis”, Perception & Psychophysics, 1973. Vol. 14. No.2. 201·211
©2003- 2016, The Ohio State University