TITLE: How many rendering hours does it take to... NAME: Seppo Halonen COUNTRY: Finland EMAIL: seppo.halonen@hut.fi WEBPAGE: N/A TOPIC: Glass COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT. JPGFILE: litebulb.jpg ZIPFILE: litebulb.zip RENDERER USED: TOOLS USED: RENDER TIME: HARDWARE USED: P60 / 24M memory, pocket calculator IMAGE DESCRIPTION: ...screw in a lightbulb? Yep, an old joke, it just hit me and I thought it was funny. After a few days of intense creative work and plenty of previews and variants, I ended up with this scene. DESCRIPTION OF HOW THIS IMAGE WAS CREATED: Lots of hard work, like my other scene, probably even more so. The first step of the modeling was to acquire a lightbulb, and since I couldn't find a clear one, just pearl-glass bulbs, I had to break one to see the guts. However, the main items of the work were the internal glass, (a merge of cylinders, torii, one superellipsoid and blobs,) the base (1380 spheres and a few cones and cylinders) and the screw ( almost as bad CSG of spheres and cylinders/cones, and one box for the gouge ). Splitting them into smaller parts or using union instead of merge might have given some speed improvement. The scene was rendered with antialias 0.3 and depth 3, max_trace_level at 35, which is very high. I don't know really how when the depth increase ceased being visible, though. Trace level used was 35/35, tho. There still is some 'black glass', but I don't know if it is because of this. Most of the parts of the made separately, with quite a bit of trial and error to get them look right. Since I had little idea as to how to align the objects to the scene in the beginning, there is again much more detail than what is actually shown. Most of it contributes, though, via refractions and reflections, and adds to the rendering time. The text part uses two fonts, arialbd.ttf and courbd.ttf, which are standard MS-Windows fonts. The lack of a function giving the width of an individual font character prevented using a proportional font for the lower text. Textures: The glassy effects are all mine. They are quite similar, except for a couple of things - the bulb has very low IOR ( 1.02 or so ) to simulate the hollowness. For some reason, POV doesn't quite dig the ray coming out of the glass. It also has no reflection.The inner thing is a bit brigher, and also has some reflection and caustics thrown in. Maybe adding it was a mistake, but it doesn't look too bad. There also is an Air texture for the inner part of the bulb, but it didn't change the appearance at all, compared to using an inverse bulb for the inside. POV's supplied Silver1 was used for the screw, and Metal finish was utilized too for the metal parts. The plate is just a white crackle beams atop a smaller, brick/feldspar/yellow crackle - I originally used pink/slate but that was too surreal. The table is leopard with high turbulence and low octaves and omega, and the 'floor plane' is just a dumb bozo - it looks dull, but I got it from the floor of my room. Top 3: sphere 16.1 * 10^9 light buffer 0.75 * 10^9 blob bound 0.72 * 10^9 16 billion sounds much, but viewing the histogram, the time used for the spheres was negligible compared to the time used for the blob/superellipsoid glass. Probably most of the intersection tests were for the numerous refracted rays.