On Fri, 26 Mar 1999, Brian Paul wrote:
> Sorry Steve, I just had to test this out. :-)
>
> I ran the mipmap.c program on my IMPACT, IR and O2 systems and saw
> straight transition lines.
Really?
I find that rather suprising. Do you agree with my explanation as
to why the lines should be *expected* to be curved?
> However, the the transition lines on the IR
> and O2 were actually a bit ragged - a few pixels getting different
> texels than their vertical neighbors.
Well - you've got to expect some roundoff error. That's not a
calculation that demands any great precision - and in hardware,
excess precision costs money.
> Nobody has complained so it's probably close enough.
Yes - sure.
The biggest problem I found was that the whole calculation is
specified in such a way as to guarantee a little aliasing at
map transitions. On 3Dfx hardware, I use a GLIDE call to add
a little bit to the computed map number to cut down on some
of the 'liveliness' in the scene.
There was some discussion of this when I found that out some
time ago - and IIRC, the conclusion was that the OpenGL spec
is somewhat 'off' in this respect.
Steve Baker (817)619-2657 (Vox/Vox-Mail)
Raytheon Systems Inc. (817)619-2466 (Fax)
Work: [EMAIL PROTECTED] http://www.hti.com
Home: [EMAIL PROTECTED] http://web2.airmail.net/sjbaker1