EMAIL: root@greenhybrid.net
NAME: Sebastian Mach
TOPIC: Complexity
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
TITLE: Twisted Cube
COUNTRY: GERMANY
WEBPAGE: http://greenhybrid.net
RENDERER USED: gladius (selfmade ray tracer)
TOOLS USED: none except for the gimp for the text in the image
RENDER TIME: 3 hours 23 minutes
HARDWARE USED: Athlon XP 1800+, 512 MB DDR RAM

IMAGE DESCRIPTION:

This is my second (and last) contribution. I could not decide which one to submit: "Twisted Spirals" or the cube edition, so, excusez moi, I submitted both. Since the images are derivatives of each other, I won't write a whole new text to make things shorter:

I thought about the topic "complexity". I have coded iterated function systems fractals some months ago, by plotting millions and millions of pixels (thanks to Thomas Ludwig alias lycium [http://lyc.deviantart.com] for teaching me). Those so called ifs are kinda complex in the mathematic sense: Match! (you can find a fractal gallery on my homepage at http://greenhybrid.net)

So my decision was to do something similar with Ray Tracing. I played with some formulas, especially trigonometric ones (e.g. sinus) and the power functions. What came up is this contribution: Twisted Spirals.

They are not real ifs, but you can do similar things with them.

Another reason for the complexity was simply the number of spheres rendered: 70^3=343,000. I think it's fascinating to see the skilit bunch of transmissive spheres getting darker in the center and beeing highly lighty transmissive at the borders.

All I want to say now: Thanks for reading and I hope you enjoy the image!


DESCRIPTION OF HOW THIS IMAGE WAS CREATED:

I've rendered the image with my selfmade ray tracer "gladius", which is "work-in-progress". Some key features can be seen in the image:

 - Standard Path Tracing Algorithm
 - for path termination I am using russian roulette
 - 150 samples per pixels
 - Realistic Skylight based on the paper by Preetham (the aerial perspective model has not been implemented yet)
 - kd-tree acceleration for the couple of sphere (I used the naive compiler for the image because the SAH one would bring no significant speed gain in this specific scene, disregarding that the SAH compiler would be quite slow regarding the number of spheres)
 - q/kd-tree for the water (a selfmade acceleration structure which I am currently writing a paper on). The q/kd-tree allows to crunch the 134,000,000 triangles (0.134G Tris) of the water into my 512 MB of memory
 - the spheres use a material mixture of diffuse and glossy (via the Phong model) reflections and transmissions


No textures have been used for the rendering, everything is pure geometry.