> The optimal value of the depth threshold is very dependant on the size of
> the scene geometry.  For example, in this scene, the geometry is rather
> large.  Because of the size, two neighboring pixels that are on the same
> surface may differ in depth by more than 1.0, causing an incorrect edge
> detection.  This perticular scene contains rather large objects.  A depth
> threshold of 6.0, for example, will detect the depth edges but not cause
> those 'shadows'.
> 
> -Nathan

Thanx for the explanation. 2.5 is a quite good value here. There's still
same little pieces of edges missing, but without shadows.

If I'm changing the scales of the objects, do I've to change the
depth_thresh too?

DEPerere

Post a reply to this message