Mueen Nawaz wrote:

> 	I suppose there are two many parameters. But surely there should be
> some way to make sense of the standard patterns provided in the SDL
> other than just the description in the docs.
> 

Yes I am haunted by the sense that the available patterns reflect a 
collection of archetypes.  And that these archetypes derive from the 
mathematical techniques used to generate them.  And that, therefore, 
they possess distinguishing structural properties, that if understood, 
would aid in an intelligent application of them.  Some, for instance, 
seem to be a nested or fractal sort of thing, others more linearly 
repeated.

If such internal structure cannot be made understandable, perhaps, one 
wonders, a descriptive set of properties, relevent to texturing, might 
be cataloged. For instance some subset of patterns are what I call 
"omni-directional" meaning I can apply them to a surface of any shape 
and they will seem to "wrap" to the surface.  Their scaling properties 
might distinguish different patterns.

It might be that the best way to "map" this would be a kind of "best 
practices" hybred of the top down, bottom up approaches.  Like, we can 
group the patterns into these structural families, we can also group 
them into families exhibiting these common properties, and use that as a 
basis for a sort of guide. ie. "Do you want to do natural structures 
such as clouds?  A pattern from this family is the best bet because it 
has such and such a structure and it exhibits these useful properties." 
  The hope would be not that you index how to make clouds, but rather 
that you give a heuristic example of how to think about the patterns and 
their possible application.

Post a reply to this message