IDeATe Optitrack Motion Capture System

The OptiTrack motion capture system in the Hunt A10A Media Lab can be used for measuring the movement of multiple rigid objects and human bodies. It is typically used to record data files for analysis but also supports real-time streaming output.

Users of the Media Lab should be aware of the Media Lab Policies. The room can be reserved outside of class time using the Media Lab Reservations web interface.

Optitrack Hardware

The IDeATe equipment desk can check out the software license dongle, calibration wand, ground plane reference triangle, tracking suits, and various individual tracking markers.

The media lab has eight OptiTrack Flex 13 cameras connected via two OptiHub USB hubs. (There is one spare OptiHub behind the equipment desk.) The OptiHubs are connected to each other via a synchronization cable, and each connects to the control computer via USB extension cables. The cables are labeled Optitrack; you may attach them to different host computers but please leave them connected to the camera hubs at all times. Please leave the hubs connected to AC power via the existing extension cord and outlet.

The cameras identify as vendor ID 0x131d, Product ID 0x012a. The hubs identify as vendor ID 0x131d, Product ID 0x0117.

Motive Software with Mac Pro

The Mac Pro in the media lab can dual-boot OS X or Windows 8. To use it with the Optitrack, it must be running Windows. If it is running OS X, you will need to restart it with the Option key held down and select the Windows partition. The default username under Windows is ‘media lab’, please ask your instructor for the password.

The software needs a hardware key to even launch. The key is a USB dongle (key serial number 104516) which may be checked out from the IDeATe equipment desk.

The lab has Motive version 1.9 with a license for both rigid bodies and human skeleton tracking. This version writes Optitrack CSV Version 1.21 format files, and uses NatNet 2.9.0.0 streaming format.

Motive Software with Cluster MacBook Pro

The cluster MacBook Pros do not have the Motive software installed by default. The software is Windows-only and must run on Windows, which is available via pBootcamp on the cluster laptops.

A copy of the software is cached in the 16-375 Google Drive software folder. You might need an instructor with administrator permissions to authorize the installation. Registering the software requires the hardware dongle and a separate license serial number; see your instructor.

You will need a USB hub to attach one camera hub, license dongle, and external mouse to one USB port; the other camera hub should attach directly to the other USB port. The Motive software doesn’t work well at scaled resolution on the Retina, so you may wish to turn the font size down even if all the menus are tiny. It isn’t as fast or pleasant as the Mac Pro, but it works.

Quick Start Guide

The best reference for getting the system running are the instructions in the OptiTrack Quick Start Guide. The following sections provide some auxiliary notes which may also help.

Workspace Interface

The Motive interface includes a 3D workspace view for visualizing the cameras, coordinates, markers, and bodies. Right-mouse drag rotates the view. Scroll-wheel drag pans the view, and scroll-wheel zooms. Left-mouse drag selects marker points.

Camera Check and Masking

The first thing to do it to make sure all the cameras are turned on and numbered 1 to 8. If they are, then you can begin the calibration process. You will want to make sure you mask out any unwanted makers/interference before doing the calibration. This is a pretty simple process where you can select a masking option from the menu and then mouse over any white dots you see in the viewport. The white dots should turn to red, which indicate they shouldn’t affect they calibration. You can then click the “Block Visible” button in the upper left menu to complete the process.

Calibration

Choose the Layout/Calibrate interface layout, click “Start Wanding”, and then move the calibration wand throughout the capture volume until the interface indicates enough samples have been recorded. A good practice is to move in diagonal lines with your swing and try to cover as much ground as you can. It’s good if every space is seen by at least three cameras.

Click Calculate to finish. You should get an ‘excellent’ or ‘great’ for best results. You can then hit “Apply” and it will ask you to save the calibration to a file. If the cameras do not get touched or moved after this point, you can just load this saved version and not have to calibrate again.

Setting Ground Plane

The last step of calibration is to establish the ground plane so the cameras have a correct reference to the world space. For this, you can take the item that looks like a traingle/right angle, place it on the ground in the middle of the space, and then press “Set Ground Plane.” You should then see the cameras reposition themselves to the room’s space.

Rigid Body Definition

The Media Lab Motive license supports both skeletal and rigid body tracking. To define a rigid body for tool tracking, first select “Layout/Create”. On the right, select the “Rigid Bodies” tab. Select the markers in the scene corresponding to the tool and click “Create From Selection”.

It is recommended to name each body; if you right-click on the assigned name under Assets on the left, you can Rename Asset.

Human Body Tracking

If you wish to do full body tracking, there are several standard marker configurations. There is a different window from the calibration one that shows you a 3D model/skeleton with markers placed in the proper locations. After your actor/actress is suited up, you can select the type of marker set up you want and then apply the markers to the body according to the skeleton reference of your selection. Once finished with the marker application, you can have your actor/actress stand in the middle of the room in a T-Pose, select and drag the mouse over all the body markers you see in the viewport, and then press the “Create” button under the 3D skeleton reference. A stick figure avatar should then appear in the viewport to represent your actor.

When doing full-body capture, it is a good idea to always start and finish in a T-Pose and have one take that is a range of motion. This range of motion take helps with character skinning when transferring the skeleton to the character mesh.

Recording a Trajectory

Select Layout/Capture to enter the recording interface. Click the red button at bottom to start and top recording. Each segment recorded will be saved as a separate Take.

The Timeline slider allows reviewing the most recent (or current) Take as individual poses.

Cleaning Data

Rigid body capture is frequently clean enough to be used as-is. Human body data usually requires manually fixing the tracks. A separate window in Motive allows for manual adjustment based off the marker’s tangent curve/path. You can scroll through the timeline and notice parts where the marker labels disappear. You will want to essentially close these gaps so that the markers do not get lost. This video gives a more in depth visual on how to go about doing this - https://www.youtube.com/watch?v=iMPh7JhlCqc

Saving Data

Right-click on a Take and select Export Tracking Data. There are various formats; the CSV format is a legible ASCII format which can be processed in Python.

FBX is generally what we use when working in Maya because you can easily import the data into there. BVH and C3D are popular mocap files but are harder to bring into Maya without a plug-in. Motion Builder will take any of the mocap file types if you need to make any further modifications or conversions.