>From: "Oliver Schoenborn" <[EMAIL PROTECTED]> > >> What happens when going from 90 degree FOV to 180 degree FOV? >> As extreme case. >> >> With narrow FOVs, the factor 2 seems to be true. >> With larger FOVs, I'm not sure. >> Perhaps this is not that important. > >I think you are right, however, it's good enough :)
I have the formula now: R = tan(alpha)/tan(r*alpha) alpha = original half-FOV, measured from center of screen to, e.g., right; alpha < Pi/2 r*alpha = half-FOV after zoom-in, 0 < r <= 1 R = LOD scale With very small original FOV, we have R = alpha/(r*alpha) = 1/r. E.g., r = 0.5 gives R = 2.0. With alpha = Pi/8 and r = 0.5, we have R = 2.1. With alpha = Pi/4 and r = 0.5, we have R = 2.4. With alpha = Pi/3 and r = 0.5, we have R = 3.0. With r = 0.25 we get R = 4.0, R = 4.2, R = 5.0, and R = 6.5. I measured that alpha = Pi/8 is what I have in my setup. So, the simplification R = 1/r would work fine for me. The LOD scale may have a second problem: With good sized FOV, the spheres at the center of screen are circles, but the spheres near the edges of screen are ellipses. Right? How this affects the LOD scaling? When viewed at the correct viewpoint, the ellipses does look circles, but because the ellipses covers more pixels, more details can be seen. Juhana -- http://music.columbia.edu/mailman/listinfo/linux-graphics-dev for developers of open source graphics software _______________________________________________ osg-users mailing list [email protected] http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
