TITLE: oillamp
NAME: Bernd Sieker
COUNTRY: Germany
EMAIL: bsieker@techfak.uni-bielefeld.de
WEBPAGE: http://psybert.uni-bielefeld.de/~bsieker/
TOPIC: glass
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
JPGFILE: oillamp.jpg
ZIPFILE: oillamp.zip
RENDERER USED: Real3D V3.5
TOOLS USED: ArtEffect Paint Program, NetPBM
RENDER TIME: 57 Hours
HARDWARE USED: Amiga 4000, WarpEngine MC68040-40, 30 MB RAM.

IMAGE DESCRIPTION: 

The image shows a setting of some items, most of them made of glass, on a
table whith a wooden rim and tiles on top.


DESCRIPTION OF HOW THIS IMAGE WAS CREATED: 

(although this is not POVray some of it may be helpful to POV users, which
is what this competition is all about, I guess.)

The image was entirely created with Real3D Version 3.5 on an Amiga 4000
with a WarpEngine 4040 with 30 MB RAM.  All objects were created especially
for this image and modelled by me alone.

Some textures are part of the Real3D V3.5 package, others were found at
various sites in the Internet, some I have painted myself using ArtEffect.

Some of the objects are CSG:  the room, the table, the blue oil lamp body
and the flames.  Others are B-spline meshes:  The clear glass oil lamp, the
wicks, the dish, the fruits.

The table consists of a hexagonal polyhedron used as a base for something
to be seen between the tiles and the wooden rim. The tiles also consist of
a smaller hexagonal polyhedron, color and bump mapped with appropriate
textures. The wooden rim was made of six identical "pies", that were
created of a complex CSG object to form the rounded edge. Each pie was
then rotated an additional 60 degrees.

The triangular blue oil lamp is a very simple CSG object, consisting of
only two subtractedd cut pyramids and a cylinder for the wick hole.  The
wick was made by "sweeping" a circular b-spline curve along another
b-spline curve.  the oil was modelled using another cut pyramid a tiny bit
smaller than the interiour of the lamp body, colouring it red and giving
appropriate refractive index to it. The lamp body has an algorithmical
"noise" bump map attached to it.

The other oil lamp was made rotating a b-spline curve around an axis,
creating the complete shape in one step, including the opening.  For the
oil itself, the spline mesh of the body was duplicated, its u and v
coordinates swapped and the shape was divided into a number of circular
b-spline curves.  The curves not needed were deleted, the remaining ones
corrected to represent the oil shape and then made into a mesh again.  A
top plate was added (using a primitive rectangle and a b-spline "trim
curve", since now the oil was just an open dish.  It was then shrunk a very
little bit to avoid accuracy problems, and coloured red.

The star shaped glass dish stand was originally created by rotating a
simple profile around an axis.  then every second "knot" curve was selected
and shrunk towards the center, creating the star like shape.  I had used an
object similar to this one in earlier (unpublished) scenes, but recreated
this one from scratch. (It's quite easy with Real3D once you know how to do
it.)

The fruit are very simple B-spline objects, the apple has a hand-painted
but algorithmically distorted colour map and a slight noise bump map, the
Orange has no special color map and a slightly stronger noise bump map than
the apple.

The flames consist of some very simple CSG.  They also contain the only
light sources in the scene.  This is the reason why I used such prominent
lens flare effects.  (Although they are called post effects in Real3D they
are really created in the rendering process by the 3D program and thus
cannot be considered illegal post processing.) In real photography taking a
picture of a scene including the only illuminating light sources will
produce a very strong glow around the source.  (I think partially caused by
grease on the lens and partially be the film and the camera back wall
themselves diffusing some of the light to the surrounding area).  The main
streaks, however, are usually created by some special effect filter.  The
smaller random streaks are caused by small scratches on the lens.  The sub
flares are caused by internal reflections between the lenses.

Real3D simulates caustics for glass shadows by brightening the central
areas of them.  While this is fine when using simple glass objects, when
using multiple glass objects between the light and shadowed areas, this
produces unwanted light amplification.  To avoid this I had to insert a
very dense shadow fog object into the left clear glass lamp.  This object
is only visible to the shadow calculations, not to the primary rays,
neither to reflected/refracted rays.  Real3D offers special attributes for
this (called "scene" and "not reflected", the latter including 'not visible
through transparent objects').

For simple glass objects, however, Real3D's method produces a good caustic
simulation that looks a lot better than "flat" glass shadows.

I used the default adaptive antialiasing and focal blur (called DOF, depth
of field) of Real3D.

You need at least version 3.5 of Real3D to load the scenes. There is a 3.5
demo CD containing the demo version and most of the textures I used. If you
have any problems examining the scene, contact me by private email. I am
also on the IRTC mailing list. It should be loadable with the Amiga and the
Windows versions, although the Interface layout was designed for my Amiga
screen and may look strange under Windows.

I did no post-processing to the image except converting it to a
98-quality-JPEG. The lens flares were created by the raytracer.

