home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Collection of Tools & Utilities
/
Collection of Tools and Utilities.iso
/
graphic
/
rtnews.zip
/
RTNEWS3
< prev
next >
Wrap
Text File
|
1992-09-13
|
49KB
|
1,216 lines
_ __ ______ _ __
' ) ) / ' ) )
/--' __. __ , --/ __ __. _. o ____ _, / / _ , , , _
/ \_(_/|_/ (_/_ (_/ / (_(_/|_(__<_/ / <_(_)_ / (_</_(_(_/_/_)_
/ /|
' |/
"Light Makes Right"
April 6, 1988
Compiled by Eric Haines, 3D/Eye Inc, ...!hplabs!hpfcla!hpfcrs!eye!erich
All contents are US copyright (c) 1988 by the individual authors
New Distributor Address
-----------------------
I have been trying to switch over to something faster, cheaper, and more
reliable than uucp mail. As you (hopefully) know, I have sent you test
messages, to which about 2/3rds the mailing list has replied. If you receive
this issue of the RT News with a return address of "...!eye!erich", then I have
NOT received your reply. Write to:
saponara@tcgould.tn.cornell.edu
Those who have successfully replied will get it with the "saponara" return
address. Please consider "...!eye!erich" to still be my mailing address - the
other account is "on loan" and may disappear someday. By the way, the new
mailing list is attached at the end of this issue.
-- Eric
-----
Contents:
RT News, hardcopy form (Andrew Glassner)
node changes (Paul Heckbert & Brian Barsky, Darwyn Peachey)
new members (Cary Scofield, Michael Cohen, John Chapman)
question for the day (Rod Bogart)
Re: Linear-time voxel walking for BSP (Erik Jansen)
some thoughts on the theory of RT efficiency (Jim Arvo)
Automatic Creation of Object Hierarchies for Ray Tracing (Eric Haines)
Best of USENET (Tait Cyrus, Todd Elvins)
-----------------------------------------------------------------------------
from: Andrew Glassner
subject: RT News, hardcopy form
I was surprised that there are still some folks on the softcopy list
not on the hardcopy list. On the next mailing, please ask anyone on
the electronic list who isn't getting the hardcopy (but wants to) to
drop me a line, electronically (glassner@cs.unc.edu) or physically
at the Department of Computer Science, UNC-CH, Chapel Hill, NC 27599-3175.
-Andrew
-----------------------------------------------------------------------------
subject: node changes
[the gist: change the net addresses of Brian Barsky, Paul Heckbert, Don Marsh,
and Michael Hohmeyer's "degas" node to "miro". Change Darwyn Peachey to Pixar]
from: Paul Heckbert
My net address is changing from ph@degas.berkeley.edu to ph@miro.berkeley.edu
(we're switching from an impressionist to a more modern artist)
Although the machine "degas" is being phased out, mail to my old address
should continue to reach me, but the new address is faster and more reliable.
from: Brian Barsky
Degas is going away. Mail should get to me as "barsky@berkeley.edu" on
the ARPAnet and as "...ucbvax!barsky" on Usenet. My machine is "beta",
but that shouldn't be necessary in the address.
from: Darwyn Peachey
Change of address:
pixar!peachey@ucbvax.Berkeley.EDU
- or -
{sun,ucbvax}!pixar!peachey
-----------------------------------------------------------------------------
subject: new people
from: Cary Scofield
Please add my name to the Ray Tracing News mailing list. You can
use the same address(es) you use for Jim Arvo or Olin Lathrop or the
one I give you below. I really enjoy reading RTN -- very enlightening
stuff!
Thanks,
Cary Scofield
Apollo Computer Inc.
270 Billerica Rd.
Chelmsford MA 01824
apollo!scofield@eddie.mit.edu
> Could you please write up a paragraph about yourself which I
> could include next issue as an introduction?
I don't know that there is much about myself worth talking about, but here
goes ...
I've been working off-and-on with 3D graphics applications and systems
for about 9 years ... started at Apollo 4 1/2 years ago ... been
working on Apollo's 3dGMR and Raytracer packages ... presently in
part-time pursuit of an MS in Systems Engineering at Boston Univ. ...
my prevailing interests nowadays are in finding and developing a
suitable software architecture for an image sythesis system that is
network-distributable and user-extensible.
- Cary Scofield
-------
from: John Chapman
[John is presently doing grad. work in computer graphics]
Yes I'd like to be on the list - getting the back issues would be great,
thanks! The most stable email address for me is:
...!ubc-cs!fornax!sfu_cmpt!chapman
although this may change shortly nothing sent to it will be lost.
Fastest response address for short notes is still ...fornax!bby-bc!john
--------
from: Eric Haines
Michael Cohen has also been added to the list. He should be well known to
anyone who has read anything about radiosity, and has also done work in
ray tracing (his latest efforts with John Wallace resulting in an algorithm
for combining radiosity and ray tracing [Siggraph '87, last article]). In
1983 we both entered the Program of Computer Graphics at Cornell, and after
graduation he stayed on as a professor. Embarrassingly enough, I thought
that he was on the RT News mailing list (most of the PCG staff are) - anyway,
he's been added:
cohen@squid.tn.cornell.edu
-----------------------------------------------------------------------------
subject: question for the day
from: Rod Bogart
I actually do have a ray tracing question. There has been mucho
noise on the net about "point in poly". Around here, we raytrace
triangles. So, what is the fastest reliable way to intersect a
ray with a triangle and also get back barycentric coordinates?
We subdivide our B-spline patches into a hierarchy of bounding
volumes with triangles at the leaves. We preprocess the triangles
by calculating a matrix to aid the intersection process. The problem
is, the matrix must be inverted, and doesn't always cooperate (this
usually fails for triangles nearly aligned with a coordinate plane).
Is there a favorite way of storing a triangle (instead of our failing
matrix) so that the intersection returns barycentric coordinates
(r,s,t) all 0 to 1?
You don't need to bother the rest of the RT gang, if you have solution
of which you are confident.
Thanks,
RGB
-----------------------------------------------------------------------------
From: erik jansen
Subject: Re: Linear-time voxel walking for BSP
I have some experience with the 'recursive' ray traversal (as I called it)
described by Jim Arvo. The first implementation dates from fall'83 (as
mentioned in the previous RT news). I presented the idea during the
workshop 'Data structures for raster graphics' (June 1985) and I compared
it there with the 'sequential' traversal and did the suggestion that both
methods can also be applied to a hierarchical box structure. See:
F.W. Jansen, 'Data structures for ray tracing',
In: 'Data structures for raster graphics', Proceedings workshop,
Kessener, L.R.A, Peters, F.J., Lierop, M.L.P. van, eds., 57-73,
Eurographics Seminars, Springer Verlag, 1986.
In my thesis 'Solid modeling with faceted primitives', (Sept. 1987) it is
discussed on the pages 63-66.
I agree that these sources are rather obscure and not known to people
(let say) outside the Netherlands. Here follows a summary of the pages of
my thesis:
The (sequential) ray tracing algorithm for the (BSP) cell structure
proceeds by local access. First the intersection of the ray with the total
structure is calculated, then the cells are stepwise traversed by calculating
the intersection of the ray with the cell planes and determining which cell
is traversed next. The spatial index (directory) is used to access the data
in the cell. (..).
The index in the dir