A few questions

Post Reply
moz
Posts: 2
Joined: Wed May 27, 2015 8:00 am

A few questions

Post by moz »

Hello,

I am using Motive:Body with 12 Prime13 cameras for two weeks now. I'm in the process of setting it up as means of acquiring motion data for researching movement/gestures in conjunction with speech. Several questions came up and I would like to pose them alltoghether here if that is Ok with you.
  • I understand there is no possibility to create a user defined skeleton. I found "Rigid Body Assisted Labelling" (RBAL) promising. However, documentation is thin. Are the undefined markers near a rigid body somehow labeled distinctively when this labeler "acquires" them? Having RBAL activated I see no difference between markers in the vicinity of a rigid body and other markers. Am I doing sth wrong? (I tried finger tracking just as in the wiki for that)
  • Is there a possibility to tweak the tolerance for rigid body markers? I.e. to assume a slightly "less rigid" body?
  • The Reconstruction Settings in the Reconstruction View essentially define trajectorization as in Left-Clicking on a Take and selecting "Trajectorize", is that right? I assume 2D Data from the cameras is always stored once, never touched and "Trajectorization" performs always on the original 2D data albeit with different settings depending on the "Reconstruction view", correct?
  • Does the use of a skeleton somehow affect the position (not the labelling) of markers in the 3D data during trajectorization? Or is the segment/limb data calculated "on top" of the marker data, and marker data thus not altered?
  • Is there any known limit for the duration of one take save the one by available hard disk space?
  • When we use one Prime13 camera as reference camera, should this one be included in the calibration process or not?
  • Is there some way to transfer a skeleton defined in one take to another take of the same subject?
  • What and when do the Move/Scale/Rotate 3D operations modify exactly?
  • Is it possible to record analog data fed into an eSync 2 device?
  • Using the Batch Processor, is it possible to extract avi-files from the takes as I can in the user interface? If not, is it possible to somehow else exctract 2D video data from the .tak files? The problem is, we want to further process the movie files, and exporting the avi's one by one is tedious.
  • I'm having trouble running the example c# batch script as well as my python interpretation of it (see below) the error message is with both scripts the same: "Exception executing script: Object reference not set to an instance of an object". Also, you seem to have a typo in your wiki. (ITakeProcessingScript vs. ItakeProcessingScript)

Code: Select all

#import sys and clr modules
import sys
import clr

# Add a reference to the NMotive assembly
clr.AddReference("NMotive")

# Import everything from sys and NMotive.
from System import *
from NMotive import *

# Define the ProcessTake function.
def ProcessTake(take, progress):
	exporter = C3DExporter()
	exporter.Export( take, take.FileName, True)
Thanks for clarifying and helping me understand your software! We were amazed btw, how fast we were able to set everything up and make first recordings. I hope it is allright to put all this issues in one thread, I mean the search function will find it, right?
steven.andrews
NaturalPoint Employee
NaturalPoint Employee
Posts: 738
Joined: Mon Jan 19, 2015 11:52 am

Re: A few questions

Post by steven.andrews »

Hello moz,

I would be happy to answer your questions inline, here, for you. If you have any followup after this, I highly encourage you to open a support ticket with us at help.naturalpoint.com, as this is a lot of information to communicate.


- I understand there is no possibility to create a user defined skeleton. I found "Rigid Body Assisted Labelling" (RBAL) promising. However, documentation is thin. Are the undefined markers near a rigid body somehow labeled distinctively when this labeler "acquires" them? Having RBAL activated I see no difference between markers in the vicinity of a rigid body and other markers. Am I doing sth wrong? (I tried finger tracking just as in the wiki for that)
- - - RBAL does reconstruct the markers close to the rigid body with a different set of parameters. The proximity and the RBAL reconstruction settings can be made stricter to help prevent the swapping of markers that move more slowly in proximity to each other. For example, the head can be made a rigid body and the face markers can be reconstructed with a lower prediction radius to prevent them from swapping as easily.

- Is there a possibility to tweak the tolerance for rigid body markers? I.e. to assume a slightly "less rigid" body?
- - - Yes. When you select the rigid body and open the Rigid Body Settings pane, this is the Deflection property.

- The Reconstruction Settings in the Reconstruction View essentially define trajectorization as in Left-Clicking on a Take and selecting "Trajectorize", is that right? I assume 2D Data from the cameras is always stored once, never touched and "Trajectorization" performs always on the original 2D data albeit with different settings depending on the "Reconstruction view", correct?
- - - Your understanding sounds correct.

- Does the use of a skeleton somehow affect the position (not the labelling) of markers in the 3D data during trajectorization? Or is the segment/limb data calculated "on top" of the marker data, and marker data thus not altered?
- - - The marker data affects the skeleton position and orientation, but the skeleton does not affect marker position.

- Is there any known limit for the duration of one take save the one by available hard disk space?
- - - I'm not sure what the file size limitation would be. On a 64 bit machine, however, users have been able to record takes as long as an hour. I believe this is only on simple tracks, such as a rigid body. While this is not a recommended workflow, it is possible.

- When we use one Prime13 camera as reference camera, should this one be included in the calibration process or not?
- - - The usual preference is to include it in the calibration. You can then overlay the 3D data on the 2D reference video.

- Is there some way to transfer a skeleton defined in one take to another take of the same subject?
- - - While the take containing the assets is selected, you can drag and drop the assets onto another take to make this association. Be sure to save the take so that the assets are then a part of the take file.

- What and when do the Move/Scale/Rotate 3D operations modify exactly?
- - - This is used to adjust the skeleton and how it solves within the marker cloud, but this is rarely done. This is a bit of a clunky workflow for the time being and should be done with care.

- Is it possible to record analog data fed into an eSync 2 device?
- - - Motive does not currently support this.

- Using the Batch Processor, is it possible to extract avi-files from the takes as I can in the user interface? If not, is it possible to somehow else exctract 2D video data from the .tak files? The problem is, we want to further process the movie files, and exporting the avi's one by one is tedious.
- - - Looking through the API documentation that installs with Motive1.8 Beta 1, it looks like this capability has been added. The documentation can be fond in the \OptiTrack\Motive\Help folder

- I'm having trouble running the example c# batch script as well as my python interpretation of it (see below) the error message is with both scripts the same: "Exception executing script: Object reference not set to an instance of an object". Also, you seem to have a typo in your wiki. (ITakeProcessingScript vs. ItakeProcessingScript)
- - - Is the script being run as it was provided, or have there been any changes made to it? I highly encourage you to open a ticket for this one so that we can properly troubleshoot this.

Cheers,
Steven
--
Steven Andrews
OptiTrack | Customer Support Engineer
Post Reply