On Fri, May 23, 2008 at 12:18 PM, Alejandro Segovia <[EMAIL PROTECTED]> wrote:
> > > On Fri, May 23, 2008 at 11:02 AM, Robert Osfield <[EMAIL PROTECTED]> > wrote: > >> Hi Alejandro, >> >> On Fri, May 23, 2008 at 2:31 PM, Alejandro Segovia >> <[EMAIL PROTECTED]> wrote: >> > Actually, I am using a TrackballManipulator and replacing its ref_ptr >> with a >> > new NodeTrackerManupulator when I have to enter "node tracking mode". >> I'm >> > not using the KeySwitchMatrixManipulator, but simply changing back and >> > forth, overwriting the ref_ptr... Maybe sometimes I get one handler and >> > sometimes the other? It shouldn't be the case... >> >> I'm afraid I can't really help, you. There is plenty of stuff you are >> doing at your end that I am not party to. You'll need to step through >> with a debugger and work out yourself what you are doing wrong. >> > > Yes, you are absolutely right, I'm sorry if the topic went too much into my > own application. I'm trying to understand how the NodeTrackerManipulator > works so I can adjust the way I'm using it. > > Following you advice, I added a couple of printf's to see what the > computeNodeCenterAndRotation call was returning (had to make it public in > order to call it from my app), and got the following for an non-moving node > at 0,0,0: > > Tracked object is at: (0.000000,0.000000,7.660254) with rotation: > (0.000000,0.000000,0.000000,1.000000) > Tracked object is at: (0.000000,0.000000,-8.660254) with rotation: > (0.000000,0.000000,0.000000,1.000000) > > for two consecutive frames. > > Now, thinking about it, I wonder, if the world coordinates are indeed > affected by the camera position, then it kind of makes sense to have the > object's center changing. > > The first time the camera is at certain position and the nodetracker > determines it has to be moved away in the Z+ direction, then the next frame > the center is recalculated and now , since the camera is further away, the Z > of the object changed and the tracker determines it has to move closer in > the Z- direction. The consecutive application of this steps makes the camera > ping-pong back and forth. > > I guess my question is, does the camera position affect the calculation of > the node's center? If so, how can I prevent it? And also, how come the > center's Z is never 0? can the tracking "step" be adjusted somehow? > Okay as I was suspecting, the problem had to do with the camera's coordinates affecting the "world" coordinates of the node I was tracking. Since calculating the world coordinates for the node is done in the getInverseMatrix() method, what I did was just set the camera's view matrix to an identity right before calling the getInverseMatrix() method. This seems more of a hack than a solution, but I guess it's the cost of using SceneView instead of osgViewer, which does not seem to have this problem as the osgsimulation example runs exemplary. Best regards, Alejandro.- -- [EMAIL PROTECTED] http://varrojo.linuxuruguay.org
_______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
