Subject:
|
Re: Ideas for next MOTM contest
|
Newsgroups:
|
lugnet.cad.dev.org.ldraw
|
Date:
|
Wed, 26 May 2004 17:09:01 GMT
|
Viewed:
|
823 times
|
| |
|
|
In lugnet.cad.dev.org.ldraw, Don Heyse wrote:
> In lugnet.cad.dev.org.ldraw, Travis Cobbs wrote:
> > The technical detail is that LDView requires the video card/video driver to
> > support the WGL_ARB_multisample OpenGL extension. The MX cards don't support
> > that. (I think this is because they only support supersample antialiasing.)
> > However, if the card's control panel allows you to set the FSAA mode used by
> > OpenGL cards, it might make the snapshots come out antialiased.
>
> What's the difference between supersample and multisample antialiasing?
> There was a setting in the control panel about some sort of ***sample
> antialiasing, but I was afraid to change it because it gave a sharp
> warning when I selected it.
The difference is that multisample AA only samples the texture once per pixel,
while supersample samples the texture once per sample. This makes multi-sample
AA significantly faster any time you're in a fill-rate limited situation.
Additionally, it appears that since the OpenGL extension is the multisample
extension, nVidia decided not to support it on cards that don't actually
implement multisample AA.
I suspect the warning in the control panel is there basically becuase FSAA will
be really slow on that card.
> > As far as I know, the main things that would differentiate an image
> > created by LDView from one created by LDGLite are smooth-shaded
> > curves and stud logos in LDView's favor, and smooth patterns in
> > LDGLite's favor (antialiased outlines of patterns).
>
> Oh, I don't know. I actually like the effect of Paul Easter's
> ldrawn logo studs better than the texture mapped stud logos in LDView.
>
> http://ldglite.sourceforge.net/minilogo.png
Just try not to use any mirrored parts or mirrored sub-models ;-).
> Now if you were bump mapping the logos, that might be a different
> story. ;^)
It's on my list of possible future features, but I've never gotten around to
trying it. Unfortunately, I think the surface normal for every stud has to be
re-calculated each frame in order to make them work right. I could be wrong
about that, though.
> > On another technical note, I think
> > we have different polygon offset settings, which can make either
> > one or the other look better, depending on the scale and viewing angle.
>
> That's a tough oneto get right. I did notice you also have a problem
> with z-bleeding sometimes. You can see it on the BattleDuck pics.
>
> http://www.brickshelf.com/cgi-bin/gallery.cgi?i=770502
Yeah. Unfortunately, tweaking the numbers can mess up other models. On the
other hand if you think the numbers you are using don't ever produce
bleed-through, I'd love to know what you're using.
--Travis
|
|
Message has 1 Reply:
 | | Re: Ideas for next MOTM contest
|
| (...) I thought we fixed that. (2 URLs) (...) Nope. I have the same problem. Possibly even worse with antialiased lines. I give them an extra bump forward beyond the polygon offset. More bits in your Z buffer reduces the symptoms somewhat. (20 years ago, 26-May-04, to lugnet.cad.dev.org.ldraw)
|
Message is in Reply To:
| | Re: Ideas for next MOTM contest
|
| (...) What's the difference between supersample and multisample antialiasing? There was a setting in the control panel about some sort of ***sample antialiasing, but I was afraid to change it because it gave a sharp warning when I selected it. (...) (...) (20 years ago, 26-May-04, to lugnet.cad.dev.org.ldraw)
|
17 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
