Aesthetic image synthesis using genetic programming
| The automatic synthesis of aesthetically pleasing images is investigated. Genetic programming with multi-objective fitness evaluation is used to evolve procedural texture formulae. With multi-objective fitness testing, candidate textures are evaluated according to multiple criteria. Each criteria designates a dimension of a multi-dimensional fitness space. The main feature test uses Ralph's model of aesthetics. This aesthetic model is based on empirical analyses of fine art, in which analyzed art work exhibits bell curve distributions of color gradients. Subjectively speaking, this bell-curve gradient measurement tends to favor images that have harmonious and balanced visual characteristics. Another feature test is color histogram scoring. This test permits some control of the color composition, by matching a candidate texture's color composition with the color histogram of a target image. This target image may be a digital image of another artwork. We found that the use of the bell curve model often resulted in images that were harmonious and easy-on-the-eyes. Without the use of the model, generated images were often too chaotic or boring. Although our approach does not guarantee aesthetically pleasing results, it does increase the likelihood that generated textures are visually interesting. |
Examples
All images copyright (C) Brian J. Ross.
With a couple of exceptions in the Sea 25 gallery, the images have not been retouched or adjusted.
In all the galleries, the image file name has embedded within it the generation at which that image arose. For example, "best_large25b18.jpg" means that the image evolved at generation 25. It is often the case that images evolved during earlier generations (< 25) are wilder in colours and less refined than those in later generations (50 and 75).