“Different sexps, same project.”

Orlando is the codename for a port of Minara to Common Lisp. Minara was originally written in Scheme with some C code for rendering. A port to Common Lisp has the following benefits:

• Better development and debugging support.
• A richer programming language with more expressive potential.
• Stronger object system.
• A language that I (the developer) am more familiar with.
• Library support for OpenGL.
• Hopefully a speed increase, or at least no appreciable speed loss.

The target Common Lisp is SBCL, which runs on all major platforms. There will be very little implementation-specific code included and this will be carefully factored out.

I’m currently aiming for an April 2008 pre-alpha release.

Install sbcl

sudo apt-get install sbcl

Install flexichain using asdf-install

sudo sbcl

(require :asdf)
(require :asdf-install)
(asdf-install:install :flexichain)
(quit)

Install cl-opengl from its darcs repository

cd ~/.sbcl
mkdir site
cd site
darcs get http://common-lisp.net/project/cffi/darcs/cffi/
darcs get http://www.common-lisp.net/project/cl-opengl/darcs/cl-opengl/
cd ..
mkdir systems
cd systems
ln -s ../site/cl-opengl/*asd .
ln -s ../site/cffi/*asd .

Test loading the libraries

sbcl

(require ‘asdf)
(asdf:operate ‘asdf:load-op ‘flexichain)
(asdf:operate ‘asdf:load-op ‘cl-opengl)

I’m very pleased with how draw-something is performing at the moment. The debugging that resulted from the improved logging and the pen parameters system means that it is now at least not working in the way it is meant to, rather than doing some things twice and others not at all.

The current look is due to the colouring system based on AARON’s new colouring system, the varied drawing styles enabled by pen-parameters, and the compositional tightness enabled by all forms being created from an initial, shared, set of points.

The colouring system needs more debugging and the composition system needs tuning up. In particular; colour application, particularly to the background, needs fixing. I want to add stars, non-self-intersecting random walk lines and circles to the repertoire of shapes. I want to make some shapes specifically smaller and larger, either growing to include nearest points or using only the points in a small area. And I want to limit kinds of shapes to various areas of the composition.

One of the things that is sorely lacking in draw-something is decision making. The actual drawing of outlines is a continuous process of deciding where the pen should go next, but the forms that the pen is drawing around are laid out randomly. In the long term I want draw-something to start with a few basic shapes and then build up from them, relating them and adorning them. But that is in the long term.

The turn amounts are now randomized and the tolerance for deviation from the path is lower.

I’ve switched back to Debian.

Click on thumbnails for full size images:

Images from the start, middle and end of the run. There shouldn’t be any mathematical differences. I’ve been selective for variety more than quality.

Technorati Tags: art computing, artificial intelligence, draw-something, generative art