Tuesday, September 1, 2009

Dull statistical analysis!

This is my antidote to too much art-making: I wasted half a day writing down data about the 2007 Fractal Art Contest, and did a little bit of number-crunching.

The first thing I found was that it's nearly impossible to accurately figure out which fractal-generating program has been used to make which image. There are certain kinds of effects that seem typical of Ultra Fractal, and Xenodream is pretty recognizable because it makes 3-D things (but then what about POV-Ray and other raytracers?), and flames might be Apophysis or they might be any one of half-a-dozen other options. It's entirely possible that there are people out there who are writing their own custom code, which isn't available to the public at all. No information at all is provided with the entry images, so if bias is going to be introduced on the basis of which program was used, the judges are all going to have to be clairvoyant so they can know which pictures to be biased in favor of.

So my statistics are highly speculative, and are based purely on guesswork and my own familiarity with the programs in question.

Out of 344 total entries, I found anywhere from 175 to 244 that looked like they could have been made with Ultra Fractal. Apophysis and other flames accounted for 51 images. 8 pictures looked like Xenodream, leaving 41 unidentifiable. There were a handful that could have been made in Fractint, and a few that had coloring I associate with the Tiera-Zon/Sterlingware/Flarium family of programs. Several appeared to have been made or significantly altered in some other graphics program, like Photoshop.

Total Entries
Ultra Fractal 71%
Apophysis 15%
Xenodream 2%
Other 12%

There were 15 winning entries, of which (again, guessing) 10 were Ultra Fractal, 3 were Apophysis, and 2 were Xenodream.

Ultra Fractal 67%
Apophysis 20%
Xenodream 13%

Possible conclusions:
  1. There is no way to be sure of which program was used, short of contacting all the entrants individually and pestering them for information.
  2. Winning spots are being stolen from users of less-popular fractal generators because of bias in favor of Apo and XD!!!1!
  3. All the images are being made in UF, because it's flexible enough to imitate all competing programs.
  4. Actually, the numbers look pretty normal, and there's no reason to suspect major bias or conspiracy or anything like that.

Free bonus conclusion:
Possibly, the people who have paid money for their fractal generators are more motivated to learn how to use them to make pleasing and effective images, such as might win contests. The more casual user, who downloads a free program and doesn't use it as often, has less experience and therefore is less likely to make a really impressive image. Better tools don't guarantee better results, but they do make some aspects of the process easier. I know for an absolute fact that I would have given up on fractal art years ago if I hadn't had access to UF's precise color controls.

Free bonus fruitcake conclusion:
If you think there's bias against non-UF images, you should see the bias against anything with mirror or kaleidoscopic symmetry. Seriously. It's crazy, man.

No comments: