|
 |
POVrayers:
I wrote a program that outputs a zillion POVray primitives - spheres,
cylinders, and triangles. My program, via the miracle of brute-force
programming, duplicates texture information into each one, so entire batches
of textures are the same.
Internally, does POVray discover logically identical textures and merge them
into the same data objects
Put another way, if I worked to merge the textures at the script level
before submitting the script, how much optimization, or pessimization, would
the resulting render experience? I'm looking for simple ways to raise the
ceiling of the number of primitives I can submit before WinXP goes into
terminal thrash mode. After discovering the simpler ones, I'l start on the
harder ones.
(I'm aware the primary way to optimize is to benchmark - I'm just asking
about internal object model goodies.)
--
Phlip
http://industrialxp.org/community/bin/view/Main/TestFirstUserInterfaces
Post a reply to this message
|
|
|
|
> POVrayers:
>
> I wrote a program that outputs a zillion POVray primitives - spheres,
> cylinders, and triangles. My program, via the miracle of brute-force
> programming, duplicates texture information into each one, so entire
> batches of textures are the same.
>
> Internally, does POVray discover logically identical textures and
> merge them into the same data objects
As far as I know, no. (at least, it wasn't so in 3.1).
In fact, povRay does this kind of thing only inside a mesh, for the
vertices.
>
> Put another way, if I worked to merge the textures at the script level
> before submitting the script, how much optimization, or pessimization,
> would the resulting render experience? I'm looking for simple ways to
> raise the ceiling of the number of primitives I can submit before
> WinXP goes into terminal thrash mode. After discovering the simpler
> ones, I'l start on the harder ones.
>
Me think you would gain most enlightement by studying the source code
instead of using a black-box experiment. But it requires personal work, a
value rarely found nowadays...
> (I'm aware the primary way to optimize is to benchmark - I'm just
> asking about internal object model goodies.)
>
Use the source, Luke...
--
l'habillement, les chaussures que le maquillage et les accessoires.
Post a reply to this message
|
|
