Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!mintaka!bloom-beacon!eru!luth!sunic!mcsun!ukc!tcdcs!swift.cs.tcd.ie!maths.tcd.ie!jorice
From: jorice@maths.tcd.ie (Jonathan Rice)
Newsgroups: comp.graphics
Subject: A-Buffer source (and plotting lots of triangles)?
Message-ID: <1990Jul16.225553.2623@maths.tcd.ie>
Date: 16 Jul 90 22:55:53 GMT
Organization: Dept. of Maths, Trinity College, Dublin, Ireland.
Lines: 37

I'm writing this program to draw perspective pictures of landscapes and I could
do with some help on a couple of fronts. What I start out with is a gridded
height map and a satellite photo overlay image. To render the thing I use just
about the simplest method possible - I treat neighbouring heights in the height
map as the corners of quadrilateral surface tiles. I perspectively transform
these corner points to give me a 2-D quadrilateral in screen space, which I
fill with a colour read out of the photo overlay. Hidden surface stuff is done
with back-to-front overplotting along scan-lines of the height map as in
Coquillart and Gangnet's "The Shaded Display of Digital Maps". I've got a sort
of a prototype going in an X environment and it's okay as far as it goes, but
in order to improve image quality, I'm going to have to renounce the graphics
functions of X and do my own anti-aliased 24-bit polygon or triangle filling.
What I think I need is something along the lines of Carpenter's A-Buffer - has
this algorithm been improved on in any substantial way since it was published
in '84? One paper I've read claims better quality images by anti-aliasing on an
area greater than just the current pixel ("Efficient Alias-free Rendering by
using Bit-masks and Look-up Tables" - Abram, Westover & Whitted, '85).  How
much does that matter? I'm going to be trying to put my images together into a
short video, so what I'm particularly interested in is removal of
frame-to-frame aliases rather than absolute rendering accuracy. Also, I have a
HELL of a lot of very small triangles or quadrilaterals to draw - are there any
optimisations that can be made to the scan-conversion process when no polygon
ever gets larger than about five pixels? Any information that anyone wishes to
grace me with would be appreciated. I'd particularly love some code that
implements this stuff, particularly an A-buffer or such-like. I am no
masochist.

-- Jonathan

o----------------------o----------------------------o--------------------------o
|    Jonathan Rice     | Email: jorice@cs.tcd.ie    | He was a common fly      |
|----------------------| Tel: 353.1.772941 x2156 (w)| With a taste for fashion |
|Computer Science Dept.|      353.1.6245415      (h)| They were thrown together|
|  Trinity College     | Fax: 353.1.772204          | In a heat of passion     |
|     Dublin 2,        |               woof /\___/  |         - "Human Fly",   |
|     Ireland.         |                     /| |\  |          The Horseflies  |
o----------------------o----------------------------o--------------------------o