Is it hard to calculate the coordinates and zoom factor that was used to generate a fractal image of, say, the Mandelbrot set? If you know the rest of the parameters, like how many iterations where used, in generating the image.
Asked
Active
Viewed 84 times
2
-
2I'm not certain I understand your question. Is this the kind of thing you're talking about? – Mark McClure Oct 10 '15 at 10:31
-
My initial guess is that it would be difficult, but I have not thought about it too much. Very nice link, Mark. – mathreadler Oct 10 '15 at 10:34
-
@MarkMcClure Yes, very interesting. They are discussing many of the things I was wondering about. – Lars Holm Jensen Oct 10 '15 at 10:47
1 Answers
1
Sketch of an approach commonly used in image processing / computer vision.
Say we want to do a systematic computational approach ( not using the algebraic properties of the fractal ) to do this and we know the mapping between color and iteration count. To be time efficient we would want to use clues from the fast-to-compute points first, i.e. the points escaping the set after a low number of iterations.
- Start finding sets of points of fast divergence in the image we are searching for. For most well-behaved fractals these will be connected sets.
- Analyze the shapes of their boundaries. This can be done in multiple resolutions to give a hint of where to zoom in to produce a fit.
- Start at a zoomed-out state, try and match the descriptors we have built from our zoomed-in-image. At first it will probably suffice to use stuff like mean iteration count over areas to find reasonable areas to search in. But as we go further in matching shape will get more and more important.
mathreadler
- 25,824