>
>
> Launch Genetish. Click on the image you think is most beautiful or
> interesting. This one will move to the middle of the screen, and a
> new set of images will be generated, based on this one. Repeat.
> Enjoy!
>
>
### Non-mathematical description:
Each of the nine 'segments' in the main window contains a mathematical
formula. This formula generates an image which is displayed in that
segment. When you click an image, all other images are discarded and
replaced by the clicked one. Then the formula in each segment - except
the middle one - is subject to a small random modification, and the
modified images are drawn. The modification is usually big enough to
cause a visible difference, but small enough so that one can see that
all images are closely related.
In other words, there is a form of 'genetic' development, not entirely
unlike the one found in living organisms. You provide the natural
selection by selecting the image which is most beautiful or
interesting. This image gets to reproduce, and each new generation of
children has random mutations.
### Mathematical description:
Each image is based on a mathematical formula in the form of a tree
structure. A node in the tree accepts one input value and returns
another value. The return value is calculated by adding, subtracting,
multiplying or dividing two values, each of which may be the input
value itself, the return value of a sub-node, or the result when
stuffing the input value or a return value into one of a number of
functions. The possible functions are sin, cos, arctan, square root,
and the absolute value. Approximations of other transcendental
functions will slowly develop as you go along, but the set which is
already there proved to give the most interesting results.
Random modifications are made one at a time, and can be made anywhere
in the tree. A modification can mean adding a subnode, removing a
subnode (and thus the entire part of the tree starting with this
subnode), changing a function in a node or chaning the way the two
sides of a node are applied to each other. The probability of a change
occuring in a particular node is always exactly as large as for any
other node. Thus, as the tree grows larger, modifications near the
beginning of the tree (which would generally have a very high impact)
grow more rare, and differences between images get more subtle.
The actual image is radial in nature and generated by sweeping the
tree with angles ranging from 0 to 360 degrees. The return values from
the tree represent magnitudes, and every point in the graph is
specified by the inserted angle and this returned magnitude. In a
spiral, all these points are connected to each other; in a star, each
point is connected to the origin and in the scatter plot, each point
is drawn isolated.
Say, there's just one thing...
> How come my system looks odd when I use Genetish?
If you have a monitor which can display only 256 colors, your system
will look very odd in the background. Fear not; when you switch back
to the Finder, the original colors will be restored.
And what about the name?:
Genetish implements a very simple form of genetic algorithm.
