@Megalomaniak said:
Yeah, lights of this nature do need to be further along to the front. 2 reasons, first the floating point precision in positioning might be a bit of a problem, and second cascade shadows just aren't all that great. Someday raytraced shadows will be cheap enough and by their nature they will be essentially perfect. But that will never be the case for cascade shadows.
Well, seems I'll have to make the ray casting account for shadow position. I want to have some crystals emit a shadow onto a zone of an object, and only when the shadow fits the hole it would activate a portal to next level. I'm considering adding 5 rays, 1 to hit the crystal, and 4 to hit the 4 corners of the hole, which will have invisible targets to confirm that the shadow shaded the hole.
EDIT: It seems like the bias problem only affects spotlights. Using an omnilight and putting it inside of a spotlight-shaped object is way better and requires lower bias to look good even on distant objects.
