> I stumbled onto an interesting and simple trick concerning normals on an object.
> I suspect it's well-known, although new to me. (The only mention of it seems to
> be at 2.4.8 in the docs, "The shadow line artifact", but that deals exclusively
> with smooth_triangles.)
>
> I'll use a smooth sphere as an example--which is about the only kind of object
> that seems to work well with this set-up. If the object is made no_shadow (or
> its light is made shadowless), an applied normal does not show the typical
> 'sharp terminator line' from the light source. Instead, the normal effect looks
> far more natural on the surface, more '3-dimensional.'
>
> It even works if the sphere is animated (revolving, for example.) I expected
> some flaws or inconsistencies to show up, but I don't see any.
>
> The only little problem is that the effect extends *past* the terminator line
> somewhat (more than seems 'natural.') But with an appropriately-chosen lighting
> angle, it fools the eye.
>
> I also tried it on a POV-Ray-generated spherical height_field; but there are
> some flaws (most likely due to the typical situation encountered when lighting
> smooth_triangles, as mentioned in 2.4.8)
>
> As it says there, the object's self-shadow is the main reason for a normal's
> typical sharp-terminator-line appearance. (I found this out myself by overlaying
> an identical sphere--made no_image but *with* a shadow--and the terminator line
> re-appeared.) But by scaling down this 2nd sphere a little bit--which reduces
> its shadow size--the nice effect comes back.
>
> So the trick has a rather limited use; but it's good for planets and such
> things.
>

As long as the planet don't cast any shadow on another one, or anything 
else. Also, if you use any media for the athmosphere, it will get 
wrongly illuminated on the night side.




Alain

Post a reply to this message