From: Marcin Romaszewicz (marcin++at++asmodean.engr.sgi.com)
Date: 08/09/2000 12:25:02
On Wed, 9 Aug 2000, [iso-8859-1] Juan R. Sáenz-Diez (Yuri) wrote:
> Thanks Marcin,
>
> I had already set the right vector, but the calls to pfSpotLSourceDir were
> missing. Now they are in place and it works. I am enclosing the corrected
> file for future reference should somebody check back on the archives.
>
> Regarding the red bleeding I managed to avoid it by commenting out the color
> mode call since there was already a material there, and you were right. I
> saw somebody reporting red squares and I thought this could be related. Is
> it legal for a color or material from a geostate to bleed to the upcoming
> geometry?
>
It is legal if the geometry it bleeds onto does not have its own
pfGeoState. If you do not specify a geostate on a geoset, performer thinks
that you do not care about that geosets appearance, so it inherits the
appearance of whatever was drawn before it.
>
> Now I have this working but during the cycle of the light there is
> some area where the shadow screws up by drawing a big black polygon.
> Is there some discontinuity in some parameter during the trajectory,
> or could there be something wrong with the pipe?
>
This could be one of many things. First, make sure that there are no
windows overlapping your window and that it is not clipped against the
edges of the screen - this will copy undefined values into the shadowmap
when performer does a glTexSubImage2D to move depth buffer values into
texture memory.
Second, it the shadow casting object does not entirely fit into the shadow
light's frustum, you will get strange blocks along the edges of the shadow
map, to solve this make your shadow casting light's frustum larger.
If the problem still happens, please email me an example which exhibits
it and I'll look at it.
-- Marcin
This archive was generated by hypermail 2b29 : Wed Aug 09 2000 - 12:25:05 PDT