Hi, are there any updates for the releasedate of the next version? Could be a pleasure to test it

The camera setup went like a breeze - tried SONY miniDV camera with night mode on and a modded A4tech webcam… The problems began when I tried callibrating the multitouch table - I can’t change grid size to the active screen area as I could in earlier CCV versions, also after callibration is done the points seem to be a bit off to one direction while in say CCV 1.3 it’s very precise.

For now I’m sticking with CCV 1.3

I have two PS3 Eye cameras and i done callibration. Pictures from both cameras are smooth connected but picture has offset on x axes. I tried to make few times callibration and allways same problem.
Does anyone knows how to fix this problem?

By stitching 2 cameras as shown in the figure, the calibration of its camera is wrong (rotated by 90 degrees) as seen in the second image. Is any way to correct that?

CCV 1.3 have some extra configuration feuatures such as camera rotation (in degrees) and overlap width. In CCV 1.5 i dont see the same parameters. Is it possible to adjust them?

Except from the calibration, i have problem with the overlap width. It is not possible to adjust it so in my case CCV 1.5 fails to stitch the two cameras correctly. A small gap appears in the middle that disappears when i physically decrease the overlap (move the cameras apart).

Any ideas from the CCV guru’s ??

Thanks in advance!!

@echnos.com: I think you should change the camera grid! I think you have arranged the cameras below each other (1x2), but you need them parallel (2x1). If you change it the camera grid the calibration points should be in the correct place.
But i have a similar problem, my setup is nearly the same as yours but i need the camera images turned by 90° this was possible in CCV 1.3 but i can’t find this option 1.5.

3dfactor - 23 January 2012 04:08 PM

Hello,
Today I tried out the 1.5 (before worked on 1.4 and it was fine). After a successful install the app refused to start throwing an error. I’ve tried installing all vc redistribution packages until I found the one that saved the day, and isn’t mentioned in CCV 1.5 requirements: Microsoft Visual C++ 2005 Service Pack 1 Redistributable Package ATL Security Update (I had the VC redist 2005 x86 and x64, VC redist 2005 SP1 also both x86 and x64, and only “ATL Security” one made it work)

Hope this saves some time for people having issues.

Aside of this little incident CCV 1.5 is working flawlesly



Cheers!

Thanks, you saved me days of frustration… though I alread have spent half a day with this problem

Hi.
Has anybody problem with no working TUIO in CCV 1.5 then “objects” is on?
I switch “objects” off and TUIO working.

Daniel Heidorn - 04 February 2012 12:27 PM

@echnos.com: I think you should change the camera grid! I think you have arranged the cameras below each other (1x2), but you need them parallel (2x1). If you change it the camera grid the calibration points should be in the correct place.
But i have a similar problem, my setup is nearly the same as yours but i need the camera images turned by 90° this was possible in CCV 1.3 but i can’t find this option 1.5.

This is true. I arranged the cameras in 1x2 positions. It is usefull in case you want to increase the effective resolution on a table that have 1024x768 aspect ratio. Therefore, for the height of the table you have 640 pixels and for the length 960 pixel. If you put the cameras in a 2x1 setup you get 480pixels for the height and 1280 for the physical width of the table. This means that your table should be narrow and long!

I’m just wondering if anybody in here could help us/me to start ccv1.5 under Windows 8 Consumer Preview?

Anat are there also any news about the new version of CCV you could share with us?

Need few more time for running on Windows 8

Hello Anat & all

First of all thanks for all the great work on that 1.5 release! It’s a pleasure to work with it…especially with a multicam setup.
I just used it for setup with 4 cams recently and run into a strange problem when using fiducials.

Details of the setup:
- Windows7 x64
- Codelaberatories driver
- 4x activated eyefi (2x2 setup), cams running at 640x480px, 60fps each, multiplexer set to 1280x960px
- DSI surface

What I’m trying to do is to track fingers and fiducials at the same time. Everything with the setup works fine and finger tracking is great, but the fiducials don’t work reliable. I also tried with a fiducial of gigantic size, so it’s clearly well visible in the camera image and I also made sure it’s in the middle of one of the cams, so not a stitching problem either.
The fiducial would get recognised for some frames (green frame around it appearing in the preview), but then it will lose it for a few frames (green frame jumps to some position on the left top of the interface…outside the cam preview!!). So it’s constantly flickering….also the coordinates I get on the other end of the TUIO switch between: correct data, some wrong (random?) coordinate and a few lost frames.
Having experienced this, I went back to only using 1 cam, and Fiducial tracking was fine. Then I set up the 2x2 cams again and found out, that the fiducial flickering gets worse the higher I set the resolution for the multiplexer?!? If I leave the multiplexer setting at 640x480px it only flickers occasionally (and green reference frame does not appear anywhere else in the interface)…when setting back to 1280x960px the flickering started (plus green frame jumping to that position on the interface).

Is this a known bug? Anyone else experienced this before? Or does anyone have an idea about what I could have done wrong?

Any help appreciated…

Hi sander!
Sorry for late response - I was extremely busy these weeks. I will try to answer your question right now.

First of all - you should understand how fiducial (reactivision) tracking works - 1) vision system preprocesses source image (in CCV case source image is the image that comes from multiplexer) and we get b/w image
2) Then we build zones tree - b[www[bb[w]]] - that means that there is a 3 white blobs on the first level, last blob on this level has a two black blobs inside and last of them has one white blob inside. So there are 3 candidates to be recognized as a fiducial w w and w[bb[w]]. 3) if we got fiducial code represented by w[bb[w]] tree in out database - so we recognize last one as a fiducial.

Let’s return to life - sometime there can be a noise on the image and than bigger this image (resolution) than more pixels can has a problem. So it can be happened that w[bb[w]] can be turned by one pixel noise to w[b[w]bb[w]] and then on the next few frames returned to w[bb[w]] again. And for this tracking method there is no stable way to provide “true” stable continius tracking. Reactivision solves this problem by dropping few “wrong” frames - but that’s not a general solution of this problem.

So in the next version of CCV will be used little different way of fiducial tracking based on analizing of the square of zones. And that will solve this problem. Before that - everyone have to find “magic” settings for their setups.

@anatoly.

wouldnt a denoise/smoothing filter eliminating most of the noise causing the problem? theres allready a smoothing filter for finger blobs. ah would be so great to have a flexible filterchain and just drop a smoothing filter to before the fiducialtracker to try out.

mh seems also that the reactivision standalone version is much much more stable. wonder whats the difference to the ccv inplementation…

Hi there

I’m setting up a large MT table with two identical Ps3 Eye camera using Ccv 1.5.

I’m having trouble saving camera setting, since just for the first Ps3 all the settings are preserved, while for the second one everything is lost each time I quit ccv.

Every time I run Ccv I have to re-set the second camera and obviously the entire calibration.

Looking in the xml file multiplex.xml, I see that the GUID (the unique id of the camera) is recreated and is always different from the previous one. Maybe could it be the reason why Ccv doesn’t recognize the second camera?

Anyone had the same issue and a possible solution?

Thanks in advanced

@todo.to.it - few community members got this problem. And they solved it by granting write permission inside CCV folder.