Angus Dorbie (dorbie++at++bitch.reading.sgi.com)
Sat, 6 Apr 1996 17:26:04 +0100
This old favourite again :).
Well you've just explained what the problem is, this isn't a bug. As you say
the first polygon drawn writes to the zbuffer & hides more distant polygons
even where the first polygon is transparent.
The easiest fix is to use
pfAlphaFunc()
and
pfAlphaRef()
calls (or modify the geostates) to cull most of the transparent pixels.
The best quality solution is to depth sort all transparent objects in the
scene, unless they overlap.
>
> I have the same effect if I load the database in perfly.ogl.
> On the RE2 it looks the way it should be.
This is because the RE2 will be using multisample transparency which stores
multiple values for z and colour at a blended pixel rather than a single value
(like the IMPACT) which cannot hold enough information to resolve a subsequent
non-depth-ordered pixel write.
>
> The graphics info of the MaxImpact is:
> Graphics board 0 is "IMPACT" graphics.
> Managed (":0.0") 1280x1024
> 2 GEs, 2 REs, 4 TRAMs,
> HQ rev A, GE11 rev B, RE4 rev A, PP1 rev A, TE1 rev A
> MGRAS revision 3, RA revision 5, RB revision 4
> VC3 rev A, Cmap rev D, MC rev C
>
> Can anybody tell me if this because of the graphics hardware in the MaxImpact
> or due to Performer 2.0?
This effect will happen on everything except ONYX & IR when multisampling, it's
due to zbuffered rendering with blended transparency.
>
> Mario
>
>-- End of excerpt from Veraart
Rgds,
Angus.
-- Angus Dorbie, The Reality Centre, Silicon Graphics Ltd, UK dorbie++at++reading.sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:41 PDT