'Spheres' was created by Turner Whitted of Bell Laboratories, and appeared in Communications of the ACM, June 1980, vol. 23, no. 6, Copyright 1980 Association for Computing Machinery, Inc. I scanned this image from Page 21 of Art and the Computer by Melvin L. Prueitt.
I decided that I really liked the image. It was quite simple, yet showed the capability of powerful computers to depict realistic scenes using complex ray-tracing algorithms. When analysing the scene to re-create it with today's technology (AMD K6-200MHz, 64MB RAM), I became aware of several flaws in the image, that were not obvious when simply admired as art. It appears that the scene is lit by a single non-fading point lightsource. The shadows cast by the spheres could not possibly be created by such a lightsource, either individually, and especially not together. After further examination, I realised that there did appear to be a fixed point related to the shadows:
How can the lightsource by under the grid, yet cast shadows from the spheres? Turner obviously has some strange things happening in his shadowing algorithm.
Raytracing has been a hobby of mine for many years, and I decided that I wanted to render my own version of this image, aptly 'iSpheres'.
All the images below were rendered using Imagine for Windows 1.3.4 on an AMD K6-200MHz processor with 64MB of RAM. The most complex part of the scene is the transparent glass sphere. Many rendering packages allow rendering of perfect spheres (not made from triangular faces), as well as adjusting the Index of Diffraction to simulate almost any substance. However, Imagine does not support CSG (Constructive Solid Geometry) objects (think of boolean-algebra with objects: solid-Sphere-A AND NOT slightly-smaller-solid-Sphere-B = hollow-Sphere-whose-walls-have-a-thickness), therefore I had to find a way of making a hollow sphere that was not actually a hollow sphere (I ended up making two solid spheres, one slightly smaller than the other, and using Imagine's 'slice' operation to 'drill' a pin-sized hole through the bottom of both spheres, thereby making an object with a single continuous surface for both the inside and outside of the sphere). You can see the evidence of this 'hole' in the early renderings due to the phong smoothing applied to the entire object. I think Imagine has a way around this, but I haven't figured it out yet.
After several iterations of surface attributes and object and lightsource placement, I was finally satisfied with the final image below. I think the glass sphere may be a little 'too' transparent, and the jade sphere could use a bit more tweaking, but I'm still pretty happy with the results, and it looks wicked on my workstation's desktop.
Here are two full-sized renderings of the final image: 1024x768 and 1280x1024.
