Am 23.04.2016 um 03:17 schrieb Bald Eagle:
> clipka <ano### [at] anonymousorg> wrote:
> 
>>     f(x,y,z) = f1(x,y,z) + f2(x,y,z) + ... fN(x,y,z)
>>
>> and where each individual component function would have an associated
>> bounding box or bounding sphere, outside of which that particular
>> component would be presumed to be zero.
>> Any suggestions?
> 
> Um,  The only thing I can think of right now would be to define something along
> the lines of a color map or an interpolated spline, and apply that to f(0) ....
> f(n)
> 
> I never studied them, but it sounds like you're decomposing everything into
> parts like a Fourier transform, or synthesizing the final function from
> something like one of those series...   Taylor series, etc.

I think you misunderstood.

My goal is _not_ to speed up generic user-defined functions by
approximating them with functions satisfying the above form.

Instead, I want to provide users with a dedicated syntax for cases where
they know that the supplied function _already_ satisfies that form, so
that the parser and render engine can exploit this property. You know,
something like:

    #declare MyFn1 = function(x,y,z) { ... }
    #declare MyFn2 = function(x,y,z) { ... }
    ...
    #declare MyFnN = function(x,y,z) { ... }

    #declare MyFn = function {
        MyFn1(x,y,z bounded_by{box{ ... }}) +
        MyFn2(x,y,z bounded_by{box{ ... }}) +
        ...
        MyFnN(x,y,z bounded_by{box{ ... }})
    }

except that this particular syntax would neither be reasonably easy to
parse, nor do I consider it reasonably pretty.

Post a reply to this message