Tuesday, July 12, 2011

New concept of GMP.

Until now, I feel the great ability of generative modeling, but also I know this code has one problem. That is, this code doesn't have 'Gene succession' system. This code could generate tons of shape variation by using many parameters, but each object has no relation each other. And now, I got an idea for new system.

As you know, 'image variation' on ADOBE Photoshop is easy color manipulation system.

The center is origin(now), and six color variations are around it. If any color variation are selected, this will be center=next origin(now). If GMP accept the system like this, the new concept is below.

The eight simple operations are around the origin (Of cource, this is conceptual figure).
The center is simple object at first. Next, apply each operation for copied object. I'll get eight new shapes. Select no.'2' shape, this will be next original shape.
Next, apply each operation again. I'll get another new shape. The operation number that I selected is 'Gene' data. [2, 6, 5, 5, 8, 7, 1...]
So, I could trace the line of operations perfectly by this numbers. 

This is next concept of GMP.


  1. Nice. So you will interactively choose successive, related parameters, right?

  2. Yes. It's a very hard to try for me, but worth of challenge. :-)