Dave,
I'd agree in general but I have observed some strange, what I believe to
be ZBuffer issues, effects. In a simple case you should be able to take
two intersecting polygons and as you say the pixel by pixel depths will
be compared on the card to determine which color to display. I still
find it difficult to believe that a graphics card could calculate all
pixel locations for potentially tens of thousands of polygons for nearly
a 1 million pixel display (1024x768) all in 1/100th of a second? I'm
guessing that they somehow optimize this to reduce the raw power needed.
Perhaps the large polygon issues are a result of some sort of culling
Java 3D or OpenGL is doing? Or perhaps calculating the location of a
pixel on a large polygon (slanted almost perpendicular to your view) is
victimized by floating point round off errors?
- John
David wrote:
>
> I am pretty sure ZBuffer detection happens during rasterization. Thus it is
> not until you get to the pixel level that the depth value is checked. I am
> pretty sure this is done on the card and not in Java3d since they just hand
> geometry to the card for processing. I am also pretty sure that a single
> polygon spanning a large number of pixels and depth values will be rendered
> correctly assuming it is within tolerance of the ZBuffer.
>
> Dave
>
> ----- Original Message -----
> From: John Wright <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Tuesday, May 08, 2001 7:21 AM
> Subject: Re: [JAVA3D] Grafic card question
>
> Desiree,
>
> That's my understanding too... for WinNT 4.0 your only choice is OpenGL.
>
> So it sounds like an issue that existed but changing the video card made
> the issue more noticable. Is there any chance you changed the color
> depth you are using too?
>
> Yes, I would consider your geometry large. Looking at your points there
> are differences of over 800. Perhaps Sun could comment on this more, but
> 800 is likely to be greater than the distance from front to back clip
> and from my own observations I'd question exactly how Java 3D (or
> OpenGL) calculates the "Z" value for each pixel on a surface like that.
> Does Java 3D calculate a single depth value for a polygon? It seems to
> me that the calculations would be astronomically complex if every pixel
> on a polygon actually had a depth calculated. Thus two long flat
> surfaces that are separated by a very small distance become difficult to
> display properly. I'm just guessing at this from my observations, it
> would be nice if a Sun Engineer could explain more of what problems
> there might be in using two very large surfaces (single polygon)
> separated by a small distance.
>
> - John Wright
> Starfire Research
>
> Desiree Hilbring wrote:
> >
> > >
> > > You might want to clarify if you are using Java 3D OpenGL or DirectX.
> >
> > I am using OpenGL (as far as I know DirectX is not supported for NT?)
> >
> > > Also did you just swap the video card or did you re-install the video
> > > drivers?
> >
> > I uninstalled the old driver and installed the driver shipped with the new
> > card.
> >
> > Also it's not clear from your e-mail if the problem is worse
> > > or better than before... sounds like it wasn't perfect originally.
> >
> > Ok it was not perfect before, but it is worse now.
> >
> > >
> > > All that being said... I'm wondering if you are seeing a ZBuffer issue.
> > > Are your ground terrain and water table polygons very big?
> >
> > What are you considering as very big?
> > These are some sample points, from which my layers are created:
> >
> > x[m] y[m] z[m]
> > 3566390 5474900 around 300m (between 300 and 305m)
> > 3566610 5474565
> > 3566120 5474500
> > 3566930 5474450
> > 3566950 5474340
> >
> > Thanks Desiree
> >
> > >
> > > - John
> > >
> > > Desiree Hilbring wrote:
> > > >
> > > > Hi everybody,
> > > >
> > > > my Elsa Erazor X (Nvidia Geforce Chip) crashed this morning, now I
> have a
> > > > Elsa Gladiac Geforce 2 MX. The performance is okay.
> > > > My problem is the following:
> > > > I am creating several layers, say one terrain layer and a second
> ground
> > > > water table label. Without scaling they height difference between them
> is
> > > > very small, and then both layers a rendered it looks like one triangle
> > > > terrain the other water, while the terrain is really above the water,
> I
> > > > only want to see the terrain, I did not use transparency or anything
> else.
> > > > With the Elsa Erazor X I had sometimes the effect the a whole part of
> the
> > > > object was rendered wrong, but not any triangle different.
> > > >
> > > > Any hints?
> > > >
> > > > Thanks for your help in advance
> > > >
> > > > Desiree
> > > >
> > > > PS: I am working on NT4 Service Pack 6 with JDK1.3 and Java3D 1.2
> > > >
> > > >
> o------------------------------------------------------------------------o
> > > > | Desiree Hilbring Institut fuer Photogrammetrie und
> Fernerkundung |
> > > > | Universitaet Karlsruhe, Germany
> |
> > > > |
> |
> > > > | email: [EMAIL PROTECTED]
> |
> > > > | # 0721 6083676
> |
> > > >
> o------------------------------------------------------------------------o
> > > >
> > > >
> ===========================================================================
> > > > To unsubscribe, send email to [EMAIL PROTECTED] and include in the
> body
> > > > of the message "signoff JAVA3D-INTEREST". For general help, send
> email to
> > > > [EMAIL PROTECTED] and include in the body of the message "help".
> > >
> > >
> ===========================================================================
> > > To unsubscribe, send email to [EMAIL PROTECTED] and include in the
> body
> > > of the message "signoff JAVA3D-INTEREST". For general help, send email
> to
> > > [EMAIL PROTECTED] and include in the body of the message "help".
> > >
> >
> >
> ===========================================================================
> > To unsubscribe, send email to [EMAIL PROTECTED] and include in the
> body
> > of the message "signoff JAVA3D-INTEREST". For general help, send email to
> > [EMAIL PROTECTED] and include in the body of the message "help".
>
> ===========================================================================
> To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
> of the message "signoff JAVA3D-INTEREST". For general help, send email to
> [EMAIL PROTECTED] and include in the body of the message "help".
>
> ===========================================================================
> To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
> of the message "signoff JAVA3D-INTEREST". For general help, send email to
> [EMAIL PROTECTED] and include in the body of the message "help".
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
