FlamePython
Another implementation of Flames Fracals in Python
All this work comes from the idea of Draves and Recase (http://flam3.com/flame_draves.pdf)
The main idea is to randomly sample a large number of points (500k will do) in the [-1:1] square. Each point is colored, starting as white (RBG [255,255,255]). Then, each point is "attracted" with an attractor function (starting with an injective function). The function modifies the coordinates of the point, but also its colour. The attractor is applied several times to insure an approximated convergence) (10-15 times works fine on Serpiensky's triangle attractor, it is set on 30 iterations so that you can see the useless iterations). The result is then a fractal flame.
So far, the "serpiensky_togif.py" file illustrates the convergence of such an algorithm, as it generates a gif file where the first frame is an image of 500k white points randomly disposed in the square, and each following frame is an iteration of the attraction. Frame after frame, the (coloured) Sepriensky's triangle appears. NB: for a further work, it is not advised to generate a png image of each step of the convergence, as it is quite long.
The "timed_blob.py" file generates a suite of images (Serp's triangle) on which a final filter (blob filter) is applied. A log file to write the operations timing is written. So far, the longuest operations is the "classePIL.toImage" method, that transforms a list of points into a bitmap. It is not surprising as it deals with 2 tabs of size sizeImage*sizeImage, and number_of_points accesses. Another costly operation is the deep copy of an image.