Features that might optimize the calibration process

Post Reply
Nacho_ITBA
Posts: 5
Joined: Fri Mar 26, 2010 1:15 pm

Features that might optimize the calibration process

Post by Nacho_ITBA » Fri May 28, 2010 12:40 pm

Hi there! We've recently purchased an OptiTrack system (12-camera configuration) and we're having some issues with the calibration process. More precisely, the calculation process is taking quite a lot of time (~45 minutes in a Core2 Duo 1.86GHz with 2GB RAM). We're kind of worried about this because, due to space constraints, we need to set up and calibrate the equipment every time we use it.

We've been thinking about how can we solve this and we'd like to ask some questions:

1) We've opened up the task manager and, apparently, Arena executes in a single core. Is there any way to speed up the calculation process by using more than one core? Can we offload some of the calculation process to the GPU in order to make it faster?
2) Is there a 64 bit version of arena in the works?

Many thanks for your help.

Cheers,

--Nacho

VincentG
Posts: 7728
Joined: Mon Jul 17, 2006 5:00 am
Location: Corvallis, Oregon

Re: Features that might optimize the calibration process

Post by VincentG » Fri May 28, 2010 3:06 pm

What version of the rena software are you using?

Is the software taking 45min. to get to a point where the option to apply the results is available?

In general the software should not take that long to calibrate.

Are you aiming all of the cameras toward a relatively central point?

Are you using 3.5, or 4.5mm lens, on the cameras?

How large is your setup area?

Jim
NaturalPoint Employee
NaturalPoint Employee
Posts: 1395
Joined: Mon Oct 14, 2002 5:00 am
Location: Corvallis, Oregon
Contact:

Re: Features that might optimize the calibration process

Post by Jim » Mon May 31, 2010 3:26 pm

This sounds like a calibration issue, it should only take 2-5Min to get a good calibration with 12 cameras. Perhaps there is a calibration setting that is causing an issue, or some other problem Vince can help with.

To answer your other questions, longer term there will be a 64 bit version, and all of the camera systems and tracking is multi-threaded, just calibration runs on one core.

Nacho_ITBA
Posts: 5
Joined: Fri Mar 26, 2010 1:15 pm

Re: Features that might optimize the calibration process

Post by Nacho_ITBA » Tue Jun 01, 2010 12:16 pm

Vincent and Jim,

Thanks for your prompt replies! Regarding your questions:

* I'm using Arena 1.5 (#build 6676).
* The software takes approximately 45 mins to calibrate, but sometimes it might take less (eg. the last calibration took ~30 minutes and gave excellent results).
* Regarding the cameras, I'm using the ones that come by default with the FLEX:V100R2 ones (I'm not sure whether they are 3.5 or 4.5 mm).
* I'm aiming all the cameras towards a relatively central point but, unfortunately, my setup area is not that large.

Do you think using a much larger capture space would help? Do you have any other suggestion?

Thanks again for your replies.

Cheers,

--Nacho

VincentG
Posts: 7728
Joined: Mon Jul 17, 2006 5:00 am
Location: Corvallis, Oregon

Re: Features that might optimize the calibration process

Post by VincentG » Tue Jun 01, 2010 1:20 pm

Do you have access to the 1.6 version of Arena, with the 3 point calibration wand?

How large is your camera setup?

What speed do you have the calibration set to? (ideal is medium)

How many data points, you should have it set to 100-150.

What other applications do you have running in the background?

During the wanding process, do you have as much of the camera preview screens covered as possible?

Nacho_ITBA
Posts: 5
Joined: Fri Mar 26, 2010 1:15 pm

Re: Features that might optimize the calibration process

Post by Nacho_ITBA » Thu Jun 03, 2010 10:40 am

Hello Vincent,

Thanks for your prompt reply.

I might download the 1.6 version of Arena but, unfortunately, we don't own the 3 point calibration wand (the whole motion capture system was purchased before that wand became available). Do you think it's worthy to upgrade nonetheless?

The camera setup for full-body capture is a circle with an approximate diameter of 3.5 mts. Bare in mind, though, that the roof is not very high (~2.4 mts) and that there's a beam that runs across the middle of the room that partially occludes some of the higher cameras (there's nothing we can do about that beam since it's part of the building).

In the calibration processes for both full-body and facial capture we've done so far we've set the calibration speed to "Slow" and captured 500 points. Nonetheless we followed your advice when calibrating a six camera setup for facial capture and found out that a "Fast" speed and 175 points provided much better results in a matter of minutes. We were quite shocked with these results though, since we've always thought that the more points the better. In which cases do you recommend using a higher number of points?

Regarding your last questions: we have no other applications running in the background and we always try to cover as much of the camera preview screens as possible.

Once again, thanks for your reply.

Cheers,

--Ignacio

VincentG
Posts: 7728
Joined: Mon Jul 17, 2006 5:00 am
Location: Corvallis, Oregon

Re: Features that might optimize the calibration process

Post by VincentG » Thu Jun 03, 2010 12:29 pm

There are some features, added with the 1.6 software, and you can still use the single point calibration with it as well (3 point wand is only required for the 3 point calibration).

As far as settings go, I would recommend a medium speed @ 100-150 points.

VincentG
Posts: 7728
Joined: Mon Jul 17, 2006 5:00 am
Location: Corvallis, Oregon

Re: Features that might optimize the calibration process

Post by VincentG » Thu Jun 03, 2010 2:09 pm

Can you also send in a copy of your single point tracking data?

Post Reply