Mark Stock's 2nd Raytraced Animation Page
Older Radiance raytraced MPEG movies
All of the movies are in
standard MPEG-1 video compression format (using the Brekeley Encoder), and
should play normally on most
machines with an MPEG-1 player...you have one, right?
Jump back to My
Recent Raytraced Movies page.
Older raytraced animations:
||In one of 12,000 apartments inside one
of the habitation modules of the arcology seen below and on my
raytracing page, 256x256, 120 frames, two ambient
bounces, 40 hours on an old SGI Indigo, 460KB|
||Flying through a latticework arcology,
also seen on my raytracing page, 128x128, 100 frames, no ambient
bounces (i.e. no radiosity calculation), an hour or two on old
||Third rocket shot, at night, with
launcher heavy gear, 256x256, 50 frames, no ambient bounces, several
jitter options set, 15 hours on same Indigo, 280KB|
||Second rocket shot, more scenic
test of sh scripts controlling firing and trajectory. 256x256,
80 frames, no ambient bounces, 20 hours on old SGI Indigo,
||Rocket shot, simply a test of
whether a light source can act as a good exhaust plume. See for
yourself. 304x200, 100 frames, 3 ambient bounces, less than 1 hour
rendering time on an HP 715/64! 311KB|
||Soleri's Linear city, only a small
section of it, though. This movie is 240x160, 200 frames, 2 ambient
bounces, but only took 6 hours or so on an unloaded HP715/64.
||Spinning CAEN logo, went back to action
scenes, a most cool one...watch for the Enterprise! 10 hours on
HP735/99, 160x120, 180 frames, 163KB, 511 KB version|
||EECS movie, take 3, first movie using
radiosity techniques, 240x160, 200 frames, 3 ambient bounces, took
all damn weekend! 415 KB|
||EECS movie, take 2, 200 frames, 160x120,
daylight calculation, 16 hours on HP715/64, 167 KB|
A little history...
I started using Radiance in March 95, I believe. It took a while for me to
get to the level of basic understanding necessary to render images. That
kept me going for a few months. Then, I thought, I could try to make an
animation out of my raytracings.
So, I took my map of North Campus, computed 36 viewpoints by hand, manually
rendered 36 frames, FTPd them to Mac, and made a movie with Sparkle. I
found out this was the long way to make a movie, but I did get a movie.
Here is is:
I realized I needed some sort of automation tool, though. I remembered a
little C from EECS 284 last term. Late May I decided to crunch some code
and see what I could do. While it isn't the prettiest tool, I wrote a
C program, that writes a C-shell script, which does all the rendering
calls and computes moving viewpoints. 10 Minutes of work setting up the
script, and I could load a machine for 15 hours. Happy was I.
I also found an MPEG encoder, precompiled for SunOS. Its from Berkeley,
and here's the FTP
Here are the first three movies made using this method:
Not satisfied with merely the viewpoint moving, I found a way to allow
objects in the scene move on their own. In the scene description file,
a number can be replaced by a numerical expression with the letters NUM
in it. The program I wrote replaces all instances of NUM with the
current frame number, and uses that scene file to make the rendering.
Here are two movies I have made with moving objects in them:
And, just recently, I figured out the daylight calculations. You need
to run the .pic through pfilt -e -1 or something to get a decent-looking
daylight rendering. So, I varied the time of day to make this next movie.
Then, to test how well this rendering technique would work for something
like movie credits, I made the following movie, with both a moving light
source and a moving viewpoint:
In this one, I made a script that generates a random forest. The
reflective arcology is courtesy of markusn. This is my
first multiple path movie. It is also my longest. I will soon code
the capability for multiple bezier paths and multiple moving objects
into my anim8.8 program. (note: yeah right, when I get some free time)
OK, I've finally got back into it. I'm looking forward to rendering movies
based on ADAMS simulations, that should be cool, but for now, there's a few
movies that are relevant to the RoboTech campaign I'm in. The first one
is a rotating view from the underside of a Zentraedi landing pod.
So, then, in March 1996, I decided I could make the EECS movie a little
better, so I did. I rendered it at 3 times the pixel size of the final
movie, leading to smoother edges. I also MPEG encoded it better. See
the above thumbnail links for these two movies!
Mark J. Stock, Graduate Student,
The University of Michigan