Evolutionary systems & genetic algorithms
From GenerativeArt
(Difference between revisions)
(→Example of Classic Genetic Programming for Problem Solving - Lawrence Fogel: removed TSP example) |
Current revision (22:11, 9 September 2013) (view source) (→Further Evolutionary Art Development by Karl Sims: Added a link to NEvAr) |
||
<span style="font-size:larger">Image synthesis</span> | <span style="font-size:larger">Image synthesis</span> | ||
| - | Karl Sims uses an old idea to create new art. In this SigGraph paper he describes how mathematical expressions can be treated as an assembly of genes. | + | Karl Sims uses an old idea to create new art. In this SigGraph paper he describes how mathematical expressions can be treated as an assembly of genes. His paper can be found [http://www.karlsims.com/papers/siggraph91.html here.] |
Basically the expression is parsed into a tree structure, and then subjected to mutations, or a form of crossover by substituting entire branches of the tree. An equivalent implementation can be done by using text oriented regular expression processing techniques. The resulting phenotype is an image computed pixel by pixel by evaluating each expression substituting the specific (x,y) coordinate. | Basically the expression is parsed into a tree structure, and then subjected to mutations, or a form of crossover by substituting entire branches of the tree. An equivalent implementation can be done by using text oriented regular expression processing techniques. The resulting phenotype is an image computed pixel by pixel by evaluating each expression substituting the specific (x,y) coordinate. | ||
* Image processing or video processing can be done by calculating pixel values as a function of (x,y,i,t) where i is the value of the image channel and t is time. | * Image processing or video processing can be done by calculating pixel values as a function of (x,y,i,t) where i is the value of the image channel and t is time. | ||
| + | Worth noting in this realm is the NEvAr system that automates the fitness function by using (more or less) the ratio of two complexity measures, compressibility via JPEG versus compressibility via fractal methods. A paper documenting NEvAr can be found [https://estudogeral.sib.uc.pt/bitstream/10316/7641/1/obra.pdf here.] | ||
== Interactive Genetic Art Installation == | == Interactive Genetic Art Installation == | ||
