Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!mips!sgi!shinobu!odin!cashew.asd.sgi.com!kurt
From: kurt@cashew.asd.sgi.com (Kurt Akeley)
Newsgroups: comp.sys.sgi
Subject: Re: Blending across a polygon
Message-ID: <1991Jun12.001329.14386@odin.corp.sgi.com>
Date: 12 Jun 91 00:13:29 GMT
References: <83407@bu.edu>
Sender: news@odin.corp.sgi.com (Net News)
Reply-To: kurt@cashew.asd.sgi.com (Kurt Akeley)
Organization: sgi
Lines: 16

In article <83407@bu.edu>, tjh@bucrf11.bu.edu (Tim Hall) writes:
|> On our GTX-B, (no destination alpha planes) specifing a different alpha
|> value at each vertex of a polygon doesn't seem to work.  (This is, of course, using an apprpreate "blendfunction")  For example, the polygon should go from being completely transparent at one vertex to completely opaque at the other vertex.  The GL seems |> to want to interpolate the alpha values when the
|> alpha values of the first two verticies differ.  It's close, but no cigar.  Changing the alpha values on subsequent verticies has no effect. So is it the 
|> GL or should I get my eyes checked?
|> 
|> Oh, the machine is running 3.3.2.

Iris GT and early model GTX machines do not interpolate alpha correctly
when filling smooth shaded polygons.  Alpha is interpolated along polygon
edges, but not within vertical spans.  The error is in hardware, not
microcode, so it will not be corrected by future software releases.

Sorry.

-- Kurt