Re: Alpha texture problem

New Message Reply Date view Thread view Subject view Author view

From: Yair Kurzion (yair++at++polygon.engr.sgi.com)
Date: 04/02/2002 17:49:38


Hello Andreas !

By default, IR/IR2/IR3 use a different transparency method than Octane/Octane2.
For correct semi-transparency, Octane/Octane2 require sorting of objects
from farthest to nearest.

It looks like your trees are not sorted back-to-front. Do you assign
special bins to the tree geometry ?

Trees are a special case of transparency. You normally don't want to see
semi-transparent pixels. Each pixel is either completely transparent or
completely opaque (the tree texture defines a cut-out of the polygon).
If all you want is cut-out trees, read on ...

You can fix the tree edge problem without sorting back-to-front: Take a look
at pfGeoState modes: PFSTATE_ALPHAREF and PFSTATE_ALPHAFUNC. They let you
control what alpha values of a polygon are skipped while rendering.
Take a look at man pfAlphaFunc for more info on this.

-yair

> I have noticed that polygons with alpha textures (e.g. flat trees or
> bushes) become somewhat strangely rendered when viewed in front of each
> other. If you're viewing such a polygon in front of another, you can
> actually see through the second polygon through the edges between the
> alpha and non-alpha texels of the first polygon.
>
> I know, this might sound a bit confusing, but I'm sure this is a common
> problem and has been noticed by many Performers out there. It doesn't
> happen on IR/IR2/IR3, but it does with Octane/Octane2 as well as with
> Geforce2/Geforce3. I have attached an image that shows the problem quite
> clearly. Could anyone describe why this is happening and if you could
> avoid it?

-- 
\_________  \_____  \__    \__  \_____        
\_________  \_____   \__   \__  \_____         Yair Kurzion
       \__     \__   \____\__      \__         yair++at++sgi.com
       \__          \__  \__                  (650) 933-6502
       \__          \__   \__          
       \__          \__    \__             


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue Apr 02 2002 - 17:49:57 PST

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.