New ARENA software 1.3.035 beta available

Birch
Posts: 1139
Joined: Thu Jan 30, 2003 5:00 am
Location: Corvallis, Oregon

New ARENA software 1.3.035 beta available

Post by Birch »

We are pleased to announce a new release of the ARENA motion capture software. Version 1.3.035 beta 1 is now available for download.

The product page is here : http://www.naturalpoint.com/optitrack/p ... n-capture/
The release can be downloaded here : http://www.naturalpoint.com/optitrack/s ... loads.html

ARENA 1.3.035 Beta 01 - June 13, 2008

Changes :
* Added V100 in-camera masking
* Added v100 precision greyscale mode
* Improved V100 camera-sync and numbering
* Improved camera view windows
* Added double-click 2d camera border to zoom
* Added 3D scene light controls in display tab
Birch
Posts: 1139
Joined: Thu Jan 30, 2003 5:00 am
Location: Corvallis, Oregon

Re: New ARENA software 1.3.035 beta available

Post by Birch »

Update :


* The installer originally posted was missing the USB drivers.
* An exception was being triggered when software news updates were available on startup.

A new installer is now posted which includes the previously missing files and a fix for the exception. We apologize for the inconvenience.
rock soderstrom
Posts: 62
Joined: Fri Jun 20, 2008 12:01 am

Re: New ARENA software 1.3.035 beta available

Post by rock soderstrom »

Hi,

I am new at the ARENA universum, so hello @ all!

I want to give you a short feedback about your last Arena beta release. We got strong problems with it.

- Cameras do not sync anymore
- the numbers of the camera are very high
- we have pretty much crashes, without any error message

Our system:

Win XP/ SP2, Intel Quad Core, 8 GB, nVidea 8800 GTX

We are now back at the last release ( 1.2.034 ) and everyhing works fine!

best regards

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

Re: New ARENA software 1.3.035 beta available

Post by VincentG »

What kind of sync issues were you having?

Could you define high numbers?
rock soderstrom
Posts: 62
Joined: Fri Jun 20, 2008 12:01 am

Re: New ARENA software 1.3.035 beta available

Post by rock soderstrom »

Hello,

the numbers on the cameras and in ARENA were high, like 62, 79 or 92. Some cameras had identical numbers, everything was confusing.

The error message was something like:"...can not sync cameras, do you want to proceed?"

The same setup with no changes works pretty good with the former release of ARENA. We are now back to that release.

bye

rock
Birch
Posts: 1139
Joined: Thu Jan 30, 2003 5:00 am
Location: Corvallis, Oregon

Re: New ARENA software 1.3.035 beta available

Post by Birch »

Can you describe how you have the sync wired for the cameras?

Is it in a chain with one camera at the start, followed by cameras in the middle, and one camera at the end?

Or is it in a loop, so that the last camera connects to the first camera? (hopefully not a loop - however this was the only way I was able to reproduce the issue with erratic camera numbers)
rock soderstrom
Posts: 62
Joined: Fri Jun 20, 2008 12:01 am

Re: New ARENA software 1.3.035 beta available

Post by rock soderstrom »

Hi Birch,

we had a loop!!! ;-)

Now, with the cameras synced as a chain our calibration results are all excellent! ( before we had just "greats" and "goods" )

Thanks a lot for this information. That issue was clearly our mistake. Probably, a good point for the FAQ.

thanks again

rock
leith
Posts: 194
Joined: Tue Jan 02, 2007 2:17 pm

Re: New ARENA software 1.3.035 beta available

Post by leith »

impressions:

in camera masking is unclear. Can't tell if its working or not. Does it replace the existing masking or is there a specific way to turn it on?

I now have all 8 cameras using 60 degree fov lenses. The calibration is running quite smoothly and the calib does seem to be doing a good job of figuring out the lense distortion.

As I pour over the ray for ray reconstruction of the markers, I'm finding a lot of situations where a ray that could be used for two different marker reconstructions is being used for an obviously poor marker, removing it from contention for other markers. I'd almost rather have the system use it for both than have it making mistakes and therefore cull data from a potentially better marker. I'd rather cull the coincidental bogus marker later than have it steal data from another valid marker with no opportunity for recourse.

bugs:

I had repeated crashes in the skeleton wizard hitting the "next button" on the marker assignment page. A reboot fixed the issue. not repeatable anymore.

feature requests:

ability to create multiple rigid bodies rapidly by selecting marker groups and hitting a button. Its frustrating to have to go through the wizard fully every time I create a rigid.

ability for rigids to share markers, in the case that a marker sits on a rotation axis between two related rigids (on an elbow, knee or wrist).

ability to create two marker rigids, simply for the purposes of ID-ing markers rather than giving an actual 6DOF transform.

when in 2d reconstruction mode (prior to trajectorization) the ability to specify a set of selected markers as "correctly id'd for this frame" for any frame at any time. The solver should be forced to adhere to these authoritative manual labels when trajectorizing and tracking.

unlabeled data should live through the trajectorization even if its not editable or exportable.

Data should be able to be "unlabeled" after it is trajectorized and the data should be kept.

a trajectorized take should be able to be "re run" through tracking. Labeled data should be considered authoritative during a rerun. However, unlabeled data should be used to try to fill in missing markers.
Jim
Posts: 1395
Joined: Mon Oct 14, 2002 5:00 am
Location: Corvallis, Oregon
Contact:

Re: New ARENA software 1.3.035 beta available

Post by Jim »

Brad:

Thanks for the detailed feedback, as usual.

I am surprised that the 60 degree lenses are working that well for you. Did you have to adjust your residual value at all? What are the average calibration values, like .3's?

I agree about the rigid body wizard, it needs the help you suggest, that is on our list.

We will also be making it possible to trajectorize unlabeled markers, they will survive from now on, and can be output in C3D.

We will discuss the authoritative marker labeling.

The new camera masking is there by default, no way to turn it on/off. It is working if it blocks markers and there is no more data coming in for a given camera, is that the behavior you expect?
leith
Posts: 194
Joined: Tue Jan 02, 2007 2:17 pm

Re: New ARENA software 1.3.035 beta available

Post by leith »

residual is up to about 1.5 - 2 cm. But its worth noting that my cameras are close around my volume. Closer than you'd have them if the lenses were longer. Which is the intent. More compact volume. Not larger volume.

The calib values are more around .2. Some are sub .1.

The masking "just working" is about what it should do. I guess I'd expect that when I have it automatically block, I'd see it generate a thresholded raster image as a hold out mask and really see that image and work with it. But that assumes there is enough memory in the device to hold a full image and there's enough processing power to work with it. If the in camera blocking is vector based rather than raster based, then the boxes still make sense. Its just confusing is all. Don't know what the cameras can and can't do. So I don't know exactly what to expect as far as UI and higher level features. Vicon cameras for example, have a really neat feature in that they can read and store the background as a grayscale and then run a subtraction live. So the system can knock out a gray splashed area, but when a bright marker moves in front of it, it can still be seen. They also can store a certain amount of raster data when they capture and store 2d data to disk. In that way, working with the masks after the session is complete is still somewhat possible. You could still in theory, dial back the mask and reveal data. Anyhow, the result of that capability is that you expose to knock out most of the garbage. You threshold low to knock out noise. You mask out to flatten the grays and kill the overblowns. And the result is as much grayscale information on the markers as possible, for grayscale centroid fitting. The threshold does not flatten out the grayscale precision and the overall exposure is a little more sensitive (open) than it could be otherwise due to the ability to depend on the mask as a spatial threshold. So you get yourself maybe half a stop to a stop more of exposure.
Post Reply