Hi Tom, On Wed, Dec 17, 2008 at 6:58 PM, Robert Osfield <[email protected]> wrote: > Perhaps the particle rotation is the thing which is messing things up. > I'll will look at this aspect.
I've been doing testing and it does indeed look that that rotations are meant to be done in eye coordinates not local coordinates to get them to look "correct". This is how the original code most likely behaved. However, all the other parameters that control particles are in objects coordinates and the rotation certainly should be too for the effect to relate to kinematics of real particle systems. For instance if you have a particle rotating about its local Z axis then if you look from above it would say go clockwise, but look from below and it should go anti-clockwise, and looking wideway you shouldn't see any rotation. I can't see the original code doing this, it'll always rotate the particle one way w.r.t the screen. Your own suggested fix doesn't really fix things either, rather it fixes the look of it for certain types of particle systems, but it'll break the particle systems that changes made back in March address. Right now it looks like both the current code is broken and the original way was broken. Things are probably correct if you don't billboard, but once you introduce billboard I believe both are broken. To fix the case of billboards with local rotation I think we'll have to project the rotation into the local coords of the billboard once rotated onto eye plane, what we should effectively see with a billboard is the projection of a rotate objects. To do this properly one would really need some kinda of volumetric particle. I will do some work on paper to see if there is workaround solution for this projection problem based on billboards. Robert. _______________________________________________ osg-submissions mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
