Motion Capture Hardware Craig Maki and Ned Kahvo.

18
Motion Capture Hardware Craig Maki and Ned Kahvo

Transcript of Motion Capture Hardware Craig Maki and Ned Kahvo.

Motion Capture Hardware

Craig Maki and Ned Kahvo

Motion Capture, Craig Maki and Ned Kahvo 2

Motion Capture: Overview

What it is and what it does.How it works:

Optical Inertial Magnetic Electro Mechanical

Motion Capture in Sports: Hockey Tennis

Motion Capture, Craig Maki and Ned Kahvo 3

What it is and what it does

Process of converting physical motion to digital representation

Motion is measured and the position in 3D space is calculated

Hardware used to measure changesSoftware used to calculate position

Motion Capture, Craig Maki and Ned Kahvo 4

How it works: Optical

Markers are placed on the actors Number of markers varies by

application/accuracy Cameras used to record movement

Generally 8 or more cameras used (depends on size of scene; high-end studios use ~70)

Software detects the markers and triangulates the position of each visible marker relative to camera position

Motion Capture, Craig Maki and Ned Kahvo 5

Optical: Markers

Passive Markers Retroreflective coating

Shines back the light sent from alight source beside the camera(like reflectors on a bike)

Active Markers LEDs used instead of reflectors Camera filters all light except for the LEDs (IR) Synchronized LEDs flash one at a time (at high

speed) this makes identifying LEDs easier

Motion Capture, Craig Maki and Ned Kahvo 6

Optical: Active Markers

RF receiver is worn on the actor

Used to synchronize LED flashes with cameras’ frame rate

1 LED visible per frame(s) Allows faster processing Requires high speed

cameras

Motion Capture, Craig Maki and Ned Kahvo 7

Optical: Cameras

Special cameras High speed

480 FPS High resolution

3600px X 3600px (12-13MP) Large Field-of-View (FOV)

60º FOV

Motion Capture, Craig Maki and Ned Kahvo 8

How it works: Inertial

Uses accelerometers and gyroscopes to measure movement Think Wii Remote with higher accuracy

The more sensors used, the better (more human) the results

Inertial provides 6 Degreed of Freedom (6DoF)

Sensitive to within 1º of rotation 6DoF:-translate X, Y, Z-pitch, yaw, roll

Motion Capture, Craig Maki and Ned Kahvo 9

How it works: Magnetic

Sensors on the actor measure low-frequency magnetic field created by the source

Control Unit correlates the locations of the sensors and source within the field

Benefits: Markers can not be occluded Fewer markers required

Drawbacks: Interference caused by steel (in

building) or other electronic devices (like monitors)

Motion Capture, Craig Maki and Ned Kahvo 10

How it works: Electro Mechanical

Exoskeletal suit is worn by actor Aluminum or plastic rods

connect multiple potentiometers which simulate joints

Gyroscopes on hips and upper back used to measure rotation

Change in voltage is measured (analog to digital) and position is calculated

Motion Capture, Craig Maki and Ned Kahvo 11

Motion Capture in Sports

FoxTrax Hockey Puck

Americans cant follow a hockey puck, so a streak was added.

Motion Capture, Craig Maki and Ned Kahvo 12

FoxTrax Hockey Puck

Standard Hockey Puck Hollowed out and circuit board placed inside Shock Sensor Infrared Emitters Internal Battery (10-30 minute duration)

Puck emitted infrared pulses that were detected by 20 different pulse detectors and 10 modified IR cameras located in the rafters.

Motion Capture, Craig Maki and Ned Kahvo 13

FoxTrax Hockey Puck

Shuttering of IR cameras synchronized with pulse detectors

Each IR camera had a 486-66 PC to process the video locally

These computers then transmitted the coordinates of the puck to an on site production trailer.

This trailer (called the ‘Puck Truck’) used the coordinates to apply a CG trail on the puck, which was then broadcast to the viewers at home.

Motion Capture, Craig Maki and Ned Kahvo 14

Hawk-Eye in Tennis

Does not need to modify the ball to track it.Uses the principles of triangulation using

visual images and timing data from 4+ high speed cameras (60 fps or higher).

Motion Capture, Craig Maki and Ned Kahvo 15

Hawk-Eye in Tennis

On each frame System identifies the ball in every image sent

from each camera Calculates the position of the ball in 3D space

by comparing each image. This is done for every instant of time to create a

path of the ball in 3D space.System can predict the path of the ball and

where the ball will strike the ground using physics.

Motion Capture, Craig Maki and Ned Kahvo 16

Hawk-Eye in Tennis

System is set up before each match with the bounds of the playing area and the rules of the game, thus allowing it to make accurate calls (within 2-3mm)

It uses 4-6 high speed cameras set at different locations around the court.

Each camera feeds into a high speed video processor which runs the ball tracking analysis.

3D coordinates are then passed to a central computer which runs the physics algorithms.

Data is then stored into a database and can be used in recall situations or to track player stats (shot percentages, etc.)

Motion Capture, Craig Maki and Ned Kahvo 17

Hawk-Eye in Tennis

Motion Capture, Craig Maki and Ned Kahvo 18

References

http://www.metamotion.com/motion-capture/motion-capture.htm http://entertainment.howstuffworks.com/king-kong2.htm http://www.metamotion.com/PhaseSpace/optical-motion-capture-

cameras.html http://www.intersense.com/InertiaCube_Sensors.aspx Johnny Lee’s Wiimote Projects: http://johnnylee.net/projects/wii/ http://www.metamotion.com/motion-capture/magnetic-motion-capture-1.htm http://www.metamotion.com/gypsy/gypsy-motion-capture-system-

mocap.htm http://en.wikipedia.org/wiki/FoxTrax http://en.wikipedia.org/wiki/Hawk-Eye