Re: [osg-users] NodeTrackerManipulator precision ( [Was] Math problem relative to OSG)
Hi Vincent, The Quat is stored as doubles, as should the rest of the internal variables in NodeTracker. Which version of the OSG are you using? Older versions didn't use double maths. Robert. On Fri, Dec 5, 2008 at 10:51 AM, Vincent Bourdier [EMAIL PROTECTED] wrote: Hi, Using a nodeTrackerManipulator to make some tests, I think I get some bug in precision... I set the tracked node with a PAT. Each frame, this PAT is modified with two Quat and one translation. Each Quat one is a different rotation... they are added (Q = q1 * q2) and applied to the PAT. The nodeTracker seems to have a precision problem... the camera do not stay motionless relative to the PAT... a little relative translation is visible, when running more than 10s. So, Is the Quat accumulation responsible of that ? is there really a lack of precision in the manipulator ? Sorry I have no enough time to make a example to show that... but I think you can try easily. Thanks. Regards, Vincent. ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] NodeTrackerManipulator precision ( [Was] Math problem relative to OSG)
Hi Vincent, When I set the PAT as Tracked node... is PAT is rotated and translated... camera do not follow it... Does your PAT have geometry under it? Or is it just a PAT in space without any children? We had a problem like that before, where if a transform (for us it was an osgSim::DOFTransform) had no geometry under it, it could not be followed by a NodeTrackerManipulator. The reason is that the NTM uses the node's bounds to calculate where it should track, and a Transform without anything under it has invalid bounds (a default-constructed bounding sphere with radius = -1). What we did to fix it is to add a ComputeBoundingSphereCallback to the transform, which returns a bounding sphere of center = (transform.translation) and radius = 0.1. That makes the NTM behave properly. Not sure if that's your problem. If it isn't then I don't know, your code looks good. If you send me a complete example (perhaps a modified osgviewer.cpp with your code) I could have a look. J-S -- __ Jean-Sebastien Guay[EMAIL PROTECTED] http://www.cm-labs.com/ http://whitestar02.webhop.org/ ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] NodeTrackerManipulator precision ( [Was] Math problem relative to OSG)
Hi, It is not a PAT children problem... I'll have a look one day on reproducing the problem... but for now it is not a priority. Thanks for your help, and thanks a lot to everyone here, I've been very intrusive this week, sorry for that. Regards, Vincent. 2008/12/5 Jean-Sébastien Guay [EMAIL PROTECTED] Hi Vincent, When I set the PAT as Tracked node... is PAT is rotated and translated... camera do not follow it... Does your PAT have geometry under it? Or is it just a PAT in space without any children? We had a problem like that before, where if a transform (for us it was an osgSim::DOFTransform) had no geometry under it, it could not be followed by a NodeTrackerManipulator. The reason is that the NTM uses the node's bounds to calculate where it should track, and a Transform without anything under it has invalid bounds (a default-constructed bounding sphere with radius = -1). What we did to fix it is to add a ComputeBoundingSphereCallback to the transform, which returns a bounding sphere of center = (transform.translation) and radius = 0.1. That makes the NTM behave properly. Not sure if that's your problem. If it isn't then I don't know, your code looks good. If you send me a complete example (perhaps a modified osgviewer.cpp with your code) I could have a look. J-S -- __ Jean-Sebastien Guay[EMAIL PROTECTED] http://www.cm-labs.com/ http://whitestar02.webhop.org/ ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org