Dear James

Here is the calibrationdata.tar.gz you asked for : 
http://fly.yozora-irc.net/shadow/
These data come from my last try of calibrating the stereo camera, which
went a lot better than the other ones, but which is not really good
though.
Moreover, when I tried to view the cloud points in rviz, I could only
see one color-changing point, because all the positions published on the
topic are at 0.0, 0.0, 0.0, even after the calibration. The good point
is that I no longer have any infinite values in camera infos.

My other question was about calibrating using a 30 fps / 320*240 camera,
which would be the most interesting for me, and still crashes the
calibration program.

Cheers
Sacha

Le vendredi 28 mai 2010 à 10:57 -0700, James Bowman a écrit :
> Hi Sacha, yes I can help with for these problems.  It would be most
> useful if you can send me the attempted calibration inputs, so I can
> add this case to the package tests.
> 
> To do this, click on the 'Calibrate' button, then click on the 'Save'
> button.  Then email me the file /tmp/calibrationdata.tar.gz, and we
> can look at why epipolar error is so high.
> 
> On Fri, May 28, 2010 at 8:11 AM, Sacha Aury <sacha@shadowrobot.com>
> wrote:
>         Hi,
>         
>         I already sent a mail to the mailing list for that subject,
>         but with a
>         wrong topic, so I ask again. Sorry for the double post.
>         
>         I am trying to use a custom stereo camera. I did not calibrate
>         it, but I
>         get a pretty good disparity map :
>         http://fly.yozora-irc.net/shadow/Capture-2.png
>         The problem is that I am getting bad values for some
>         parameters in
>         camera_info, such as K and P, which are infinite. So I try to
>         calibrate
>         using image_pipeline/camera_calibration. If I do so using 30
>         fps/320*240
>         resolution, I get a crash when I click on calibrate :
>         
>         >>> epipolar error: 0.857935751957
>         >>> OpenCV Error: Assertion failed (src.type() == dst.type()&&
>         dst.size()
>         >>> == mapx.size()) in cvRemap,
>         >>>
>         file /opt/ros/boxturtle/stacks/vision_opencv/opencv2/build/opencv-svn/src/cv/cvimgwarp.cpp, line 3083
>         >>> Exception in thread Thread-9:
>         >>> Traceback (most recent call last):
>         >>>     File "/usr/lib/python2.6/threading.py", line 532, in
>         __bootstrap_inner
>         >>>       self.run()
>         >>>     File
>         >>>
>         "/opt/ros/boxturtle/stacks/image_pipeline_copy/camera_calibration/nodes/cameracalibrator.py", line 59, in run
>         >>>       self.function(m)
>         >>>     File
>         >>>
>         "/opt/ros/boxturtle/stacks/image_pipeline_copy/camera_calibration/nodes/cameracalibrator.py", line 233, in handle_stereo
>         >>>       lscrib = self.c.lremap(lrgb)
>         >>>     File
>         >>>
>         "/opt/ros/boxturtle/stacks/image_pipeline_copy/camera_calibration/src/camera_calibration/calibrator.py", line 407, in lremap
>         >>>       cv.Remap(src, r, self.lmapx, self.lmapy)
>         >>> error: src.type() == dst.type()&&   dst.size() ==
>         mapx.size()
>         
>         So I try using 15fps/640*480, which is better. The only
>         problem is that
>         I can't get any good calibration parameters : epi is about 6-8
>         and the
>         size is totally wrong. Moreover, I have a strange problem, the
>         camera
>         seems to be zooming after clicking on 'calibrate' :
>         http://fly.yozora-irc.net/shadow/Capture-4.png
>         http://fly.yozora-irc.net/shadow/Capture-5.png
>         
>         I hope someone can help me.
>         
>         Thanks
>         
>         Sacha
>         
>         
>         _______________________________________________
>         ros-users mailing list
>         ros-users@code.ros.org
>         https://code.ros.org/mailman/listinfo/ros-users
> 
> 
> 
> -- 
> J.
> 
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users