John Wright wrote:
>
> Oh gawd Justin, we don't seriously need to explain why floats have
> precision problems do we?

Yes and no:

1. I was bored and didn't feel like doing anything else at the time (ie
trying to avoid the other big piles of work I have to do)

2. You didn't seems to know the basic concepts of Zbuffer handling so I
assumed you probably wouldn't know why floats cause so much problems.

3. General interest for the others on the list. Plenty of non electrical
engineers on the list. I only learnt about float representation and
low-level handling in my engineering degree. The comp-sci degree never
even mentioned it so I assume a lot of other people don't know these
basic principles - particularly here on this list were we tend to get a
lot of people from the content side of life rather than the programmer
that can compute matrix transforms in thier sleep.

> I'm more interested and concerned about how these techniques can lead to
> really bizarre images.  Take your explanation for example, if we have
> one large polygon that has each of it's vertices obscured by small
> polygons then we make the decision that it is "not visible" however
> we've failed to account for the HUGE gap between the obscuring little
> polygons.

Exactly, and that's why I said is was a very simplified algorithm to
illustrate the points. You will find that no-one implements hardware
like that.

> Hence we are either at risk for some really ugly effects or we need a
> more sophisticated method for determining if an entire polygon is not
> visible or not.  In my opinion the moral of the story is that if your
> scene is made up of lots of large polygons you'll have worse Zbuffer
> problems than small polygon objects.

That's a fairly rough summary. It is more the problem of intersecting
and lots of nearly parallel polygons that are really close together that
are the problem.

--
Justin Couch                         http://www.vlc.com.au/~justin/
Freelance Java Consultant                  http://www.yumetech.com/
Author, Java 3D FAQ Maintainer                  http://www.j3d.org/
-------------------------------------------------------------------
"Humanism is dead. Animals think, feel; so do machines now.
Neither man nor woman is the measure of all things. Every organism
processes data according to its domain, its environment; you, with
all your brains, would be useless in a mouse's universe..."
                                              - Greg Bear, Slant
-------------------------------------------------------------------

===========================================================================
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".

Reply via email to