The Birdland Images
first image generators - BASIC programs
| lieschen mueller | left: 1992/193 ‘generated self portrait’. right: ‘1990/17’ black rabbit |
 |
 |
The first version(s) the generator used combinations of differnt shapes: cricles, triangles, irregular quadrangles and their negatives as a cut out on a base of the 16 available colours. This has been a very good mix- and reproduceable colour palette. The images had to been photographed from the screen, the messurements had been printed, the whole image was reconstructed on an tracing paper first, photocopied to transfer and finally redone in self mixed accylic paint on MDF plate serrated with a piercing saw. Note: Early BASIC had have almost predictable random generators.
the birdland generator experiment
A more advanced program reconstruction of the lost source code of the basic program for semantic random image generators. The program language shifted from BASIC to javascript. Javascript comes with different interpretors in different flavors, which means each virtual machine has it’s own random generator.
The bridland generator uses a more accurate predictable set of shapes and preset of conditions to fix the problem with the mathematical more real-random random implementation.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
statistics
Final results, depending on the used underlying js-engines: There are five groups of images
- faces 10-15%;
- constellations 5-10% (not shown)
- interesting take outs ~5%
- complete failures <5% (not shown)
- birds 70-80%
With emulated random the settings will cause a complete failure of always similar results of one or two groups.
The semantics used do follow a similar approach to join the concrete shapes other investigations has used in the 80ties to create random music and music pattern. It is also inspired by Blues and Jazz ‘rules’, espacially by the music of John Coltrane.
music circle by John Coltrane
Navigation Shapes With Rules Generator List and Thoughts