RE: Perlin and simplex noise

It’s a bit premature, but I think you may enjoy Eric Elder's solution to the problem that we'll talk about in Zurich in two weeks. It doesn't deal with turbulence at all but rather with a way to invoke seeded values repeatedly and randomly, but consistently from invocation to invocation. The bad news is that it involves not only declarative randomness, but also <replicate>. We'll be sure to talk about this issue since it seems to be crucial to implementers for standards purposes. Scientists would like it too, and maybe even artists. Basically, the seeded generator is supplied, under replication, with different values stemming from a replicateModifier applied to the <random> element nested within the object being replicated. Perhaps some ideas can be gleaned from http://cs.sru.edu/~ddailey/svg/replicateRandom.html . None of those are dealing with seeds to fix the randomness from invocation to invocation, but some deal with replicateModifier, which is further discussed at http://svgopen.org/2010/papers/46-A_proposal_for_adding_declarative_drawing_to_SVG/index.html 

Regards
David

On Fri, Aug 31, 2012 at 1:49 PM, David Sheets <kosmo.zb@gmail.com> wrote:
> On Fri, Aug 31, 2012 at 10:05 AM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
>> (While we're at it, though, we should pay attention to how it would 
>> be possible to do declarative randomness in properties.  Marrying a 
>> stateful RNG with a nominally stateless language is hard. :/ )
>
> Procedural noise can be seeded from time alone. I believe that CSS 
> transitions should be able to provide custom shaders with a fractional 
> completeness value, no? Is there a means in New CSS to create 
> unbounded animation?

I may have been unclear about what the issue is.

If using randomness to, say, set the background-color of an element, you usually will want different elements to have different random values.

However, you also want these values to be stable across common types of changes, like temporarily setting the background-color to a non-random value on hover, then returning to the random color afterwards.

Handling this intelligently and with a minimum of fuss seems non-trivial, though I do have some ideas of how to possibly do it.

~TJ

Received on Saturday, 1 September 2012 01:39:48 UTC