I remember when I got my first computer, and I downloaded Fractint and would sit and watch as it rendered images, one slow line at a time. And I remember my early days of using Ultra Fractal, when new coloring methods would be written, some of them were ridiculously complicated and would render one slow line at a time.
Over the years, technology has advanced, processors have gotten faster, and I can open up some of those old parameters and have them appear in a few quick seconds. But somehow, I always still seem to end up staring at my monitor, transfixed, as an image appears one slow pixel at a time. Sometimes it's because I'm doing deep zooms, that get into arbitrary precision. Sometimes it's lots of layers. Sometimes it's ever-more elaborate coloring methods.
This tiny snippet is probably the slowest combination of things I've tried so far. It's a Mandelbrot zoom, with Extended Precision, fairly close to the boundary (so needing lots of iterations to avoid blank gaps), and with my own parametric coloring in a particularly slow configuration. At 400 x 400 pixels, this took four hours to render. I wanted to do a test of whether it would look like I expected, so as to be used in a larger, several-layered image. Unrendered, it dissolves into a mass of crunchy pixels.
So now I'm wondering, is it worth doing the somewhat larger render of the image that this is a component of? It looks like the guilloche-pattern effect is working the way it's supposed to. If I go by my on-screen working version, the colors and layers are okay. Render time estimate is somewhere between 850 and 900 hours. That's more than a month, assuming it doesn't slow down a lot when it gets near the minibrot in the middle.
Well, I've started it. It can mutter away in the background while I'm thinking of other things.
Saturday, August 29, 2009
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment