hi i’m alessandro,

we are doing a MT circular table using the fantastic ccv tool… and vvvv
we have a problem configuring the camera because off course there is no way to configure a circular table…
also tha camera is very wide angle.
does any one of you has experience with that?
is there some one who can help us?

thank you in advance

alessandro

i met similar problem

aha… yeah, we’re experiencing the same problem. we have a hexagonal table, and when we try to calibrate, the points on the bounding box lie outside of the table.

we found a tricky workaround. we built an mtmini (which is rectangular obviously) and we switch cameras before calibration to the mini. we calibrate with the mini, then my application does its own calibration. our program uses a very coarse grid, so my very cheap calibration routine seems to work. if you need fine accuracy then this will probably not do the trick.

i’d really like a more elegant solution though, like the ability to disable certain points of the calibration grid, instead using an extrapolated value in its place.

hi bluetshirt,

thank you very much for your answer… this trick you found sound very interesting but unfortunatly i need a lot of accuracy…
i solve the problem using vvvv… there’s a plugin called warp point that do exhactly the same of the calibration of ccv… but is easy for me change the shape of the reference point…

i think that the solution it will be change the reference point inside in the ccv calibration from a quad to a circular shape… but unfortulatly i’m not enough inside openframework to do it.

any way i’m close to finish my MT talbe and as soon as i finish it i will post the documentation on how i do it…

thank you again

ales

hi,

is there a non vvvv solution for circular touch screens? the vvvv wrapper thingi is far too inaccurate!

by the way!
ccv1.2 mac has got a bug on adding calibration points i think. Can’t add points on X axis.
[+] is adding points on Y axis. That’s right…
[-] is doing nothing at all (Should remove Y points...)
[shift]+[+] is doing nothing at all.
[shift]+[-] is removing Points on the Y axis instead of removing points on X axis (if there were any points )

kind of confusing!!

sebastian

dear sebastian,

you are right the ccv calibration for mac also seams me bugged…

about the vvvv calibration i did a patch that do tha calibration for circular table using warppoint plugin… and it is extreamly accurate…

i’m close to finish my project and as soon i finish it i will post a little documentation about it

ciao

alessandro

great alessandro! I’d be interested in checking that out…

cheers,
J

Hi,

has anyone found a decent way to calibrate circular screens in CCV?

We are thinking about a workaround that is kind of terrible and no fun at all so I was hoping that someone had found a decent workaround.  Our method:

- Precalculate x & y of all the points on the grid. So if the grid is 5 x 5 we just use a spreadsheet to calculate the 25 points in a perfect grid
- Start the calibration process and then for each point that is “off stage” we click just anywhere and obviously get bad values
- For each point on stage we just calibrate as usually
- After calibration we open the XML file and manually fix the wrong points with the values from our spreadsheet.

This would be very tedious on a large grid like 10 x 10.

So - what do you say?  Who will save my mental health and tell me how I could do this more easily?

Unfortunately I can’t replace my circular surface with a square one while calibrating. That would have been ideal.

Best,
Sammi

Maybe it would be a good idea to add a keystroke to the calibration process to “Skip point”.  What CCV would actually do then is to set the point’s x & y to precalculated values.

Wouldn’t that be an easy way to have a decent workaround for circular calibration?

Best,
Sammi

Is there a update on the calibration process for circular tables or the possibility to skip points in the calibration process and interpolate them?

Would be very interested in using CCV on a circular table.

Sambuco

Hi,

For my circular table I just created a tiny flash software that “corrected” the xml by calculating a likely position for points outside the circle using an average of the points inside it.

After creating it I actually thought of a better idea but since this worked for me I never had the time to implement that idea.

Best,
Sammi

Hi Sammi,

to understand it correct, you used the normal calibration process and corrected the wrong points afterwards. How did you calibrate the points that were not visible with CCV and determine the wrong points? I’m not that much into the logic of calibration.

Would you share your software?

Sambuco

Hi,

Yes of course I am willing to share it.  It was not made to be shared but rather to quickly solve our problem so don’t expect a fancy UI or a help file.  But it is very simple to use so don’t worry either.

So what you do is:

- place this little Flash application in the data folder of CCV next to the calibration.xml file
- start your normal calibration process
- you must make sure to set the bounding box so that at least one cross is clickable on the all the edges
- do you calibration in CCV but whenever a cross it not clickable, just click somewhere ( it will be corrected later )
- after your have completed the calibration launch the CircularCalibrationWorkaround application
- you can use the arrow keys to adjust what I call a safe zone.  Experiment with it until you have a pretty good looking grid
- now you system clipboard contains the contents of a “fixed” calibration file
- open calibration.xml and override it’s contents with your system’s clipboard.
- done!

What happened is that the application tried to calculate the positions of all points outside the safe zone using the points inside it.

I hope you can use it.

Best,
Sammi
CTO @ gagarin.is

Hi Sammi,

thank you very much. I will try this next week and will give you feedback.

Sambuco

No problem,

You can check out our circular table on our website if you like.  There are a few shots in the video.  We also have it on our Vimeo channel

Best,
Sammi

Did you ever test it?

Best,
Sammi