Subject:
|
Re: Could someone run a perspective mode test on ldglite?
|
Newsgroups:
|
lugnet.cad.dev.mac
|
Date:
|
Thu, 6 Jun 2002 17:37:41 GMT
|
Viewed:
|
2186 times
|
| |
|
|
In lugnet.cad.dev.mac, Travis Cobbs writes:
> In lugnet.cad.dev.mac, Don Heyse writes:
> > Neat pics, but while viewing them I realized that all of the
> > perspective (non-orthographic) views displayed a lot of z-fighting
> > (underlying surfaces and lines bleeding through). This also
> > used to happen in Mesa because of the default 16bit zbuffer.
> > I fixed it there by moving the znear and zfar clip planes closer
> > together if GL_DEPTH_BITS reports a value less than 24. Perhaps
> > I need to do it for values less than 32 instead. If I recall,
> > someone posted that ATI rage 128 used on the Macs posted a value
> > of GL_DEPTH_BITS = 24 when ldglite is run. Can someone verify
> > this?
>
> This isn't a direct answer to your question, but one thing I noticed in
> LDView is that setting the polygon offset too high will result in some of
> the observed problems. In particular, the problem on inverse slope bricks
> of the pillar showing through the slope is aggravated by polygon offset.
> This can be seen in this picture from that site:
>
> http://wind.prohosting.com/anoved/legopics/hatchet/5.jpg
>
> Having said that, there may be a more general Z buffer problem causing the
> display problems in the images.
Yes, I've also seen an occasional small amount of bleeding on on the
slope bricks which can probably be attributed to my maxing out the
polygon offset setting. But this goes way beyond that. Edge lines
from way back in the depth buffer are bleeding through. You can see
the same sort of effect setting the far clip plane in ldglite to some
huge number like -Z100000000. However, having just tried that, I
remmber the other z-fighting symptom is a sawtooth effect where different
colored shaded polygons meet. The Mac pictures don't seem to exhibit
that particular symptom. They only seem to have the problem with
the edge lines, so perhaps the polygon offset is somehow involved.
It *does* look a bit like a polygon offset run completely amok.
Hmmm, I remember the polygon offset function didn't work at all in
the so-called "hardware accelerated" driver of my old ATI Rage-Pro
card, so I made up the -eN.N command line switch to overcome this
limitation. The Macs use a newer version of the ATI card, and
perhaps it has other problems with the polygon offset function. I
wonder if a -e-1.0 or even -e-10.0 would undo some of the odd effects?
Unfortunately I don't have a Mac to test it on...
Don
|
|
Message is in Reply To:
66 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
