Ken wrote:
> 
> Nieminen Mika wrote:
> >
> > Ken <tyl### [at] pacbellnet> wrote:
> > :  Having answered this question on another newsgroup recently
> > : it reminded me how often it is asked. I know that this pretty
> > : much the same answer as given for the video card question but
> > : also expands upon it to cover other areas. I didn't see it on
> > : the VFAQ ( I might have missed it) so here is the response I
> > : gave.
> >
> >   I added your response to the VFAQ and wrote also another one. My response
> > is far from complete; any suggestions are very welcome.
> > http://www.cs.tut.fi/~warp/povVFAQ.html#renderingspeed
> 
> Looks good to me.
> 
> --
> Ken Tyler
> 
> tyl### [at] pacbellnet

FYI here's the results of a little test I did the other day.
I made an array of spheres like this

union{ // switch this between merge and union
  #declare Z=-2;#while(Z<=2)
    #declare Y=-2;#while(Y<=2)
      #declare X=-2;#while(X<=2)
        sphere{<X,Y,Z>,0.9}
        #declare X=X+1;#end
      #declare Y=Y+1;#end
    #declare Z=Z+1;#end
    pigment{rgbt<1,1,1,.0.9>}
    bounded_by{box{-3,3}}  // toggle this line on and off
}

so I got an array of 5x5x5 spheres, set the camera so the spheres took
up about a quarter of the width of the image.  Now I rendered it using
union and with merge and also with/without bounding.

Results:
               pixels/second  
merge bound      4452
merge unbound    1528
union bound       933
union unbound    1839

So at least in this case the merge with manual bounding was by far the
fastest.

Cheers, PoD.

Post a reply to this message