TITLE: "Tools of the Trade"
NAME: Michael J. Kolb
COUNTRY: United States of America, Illinois
EMAIL: m-kolb@uiuc.edu
WEBPAGE: http://www.cen.uiuc.edu/~m-kolb/

TOPIC: Glass
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
JPGFILE: mjktools.jpg
ZIPFILE: mjktools.zip
RENDERER USED: POV-Ray 3.0 for Windows,
               POV-Ray 3.01 for SunOS

TOOLS USED: 
xv - reconstitution of image and conversion to JPEG
            Microsoft Visual Paint - creation of graduated cyl. texmap
            LViewPro - creation of graduated cyl. texmap


RENDER TIME: Approx. 120 CPU-hours

HARDWARE USED: 
75 MHz Pentium w/ 24 MB RAM running Windows 95
               A lab full of Sun SparcStations, config and SunOS version
unknown
               100 MHz Pentium w/ 32 MB RAM running Linux
               SGI O2 with a kick-ass monitor, config and IRIX version unknown
               HP ScanJet 2C color scanner


IMAGE DESCRIPTION: 
The image is of a chemistry experiment taking place, making use of various
types of chemical GLASSWARE.


DESCRIPTION OF HOW THIS IMAGE WAS CREATED: 
All objects were created using CSG - I'm not skilled enough at modellers to 
produce anything worthwhile, and I REFUSE to use someone else's models. Test
tubes, beaker, graduated cylinder, and liquids therein contained are all
fully refractive. The notesheets are flat boxes image_map'ped with an example
of my own illegible handwriting. The graduated cylinder has been image_map'ped
with a GIF created using MS Paint and LViewPro. Boiling bubbles created using
POV-Ray 3.0 preprocessor directives. Image contains 3 light sources: a white
point_light up-left-behind, a yellow point_light up-right-behind (for 
highlights), and an area_light within the bunsen burner flame. Image contains
4 halos: 2 in the bunsen burner flame, and 2 in the steam cloud. Image makes
use of a black atmosphere to provide depth-cueing (note the gradual fade-out
with distance). Most difficult part of the image was the steam cloud.

The image was repeatedly test-rendered on my puny little P75. For the
final render, I used the SunOS version of POV-Ray and renderfarm'ed the 
image overnight in a SparcLab (courtesy University of Illinois College of
Engineering) at 30 lines per machine, and recombined the results using the 
O2 (courtesy of SGI and the UIUC student chapter of ACM). My roommate's Linux
box served as a central control station for monitoring the render process
remotely. The render time is a reasonable estimate; the least complex 
segment required 26 minutes, and the most complex needed over 36 hours. 

Special thanks go to:
   Michael J. Drobick - creative input and hardware usage
   SGI - the O2, on loan to ACM @ UIUC
   ACM @ UIUC - hardware, creative input, advice
   UIUC COE - hardware, not kicking me off
   the IRTC-L mailing list - for listening to me whine, and answering my
                             questions
   everybody else I may have missed

