POV-Ray : Newsgroups : povray.newusers : Very big time difference of a render : Re: Very big time difference of a render Server Time
19 May 2024 10:30:04 EDT (-0400)
  Re: Very big time difference of a render  
From: Alain
Date: 19 Jun 2013 16:18:48
Message: <51c21228@news.povray.org>


> About the pigment_pattern with image map slowdown, how much does it really slow
> things down? Because that's really needed to get to more complex texture
> influence export  corner cases like specular mapping with a texture. To
> implement scripted conditions that detect more cleverly the complexity of the
> scene in original file and export only with necessary features, I need to know
> how much speed could be gained, and if it's really worth it.
>
>


This can vary greatly. It all depends on the complexity of the 
controling pattern and what you do with it.
Whenever you have any ..._map, there is some blending and averaging 
done. Everywhere outside the values that are explicitely provided, you 
need to evaluate at least 2 pigments, textures or normals, and ponderate 
each according to it's contribution. The worst case been an average_map 
of full textures with normals, reflection and transparency.

A best case:
pigment_map{[0.5 Pigment1][0.5 Pigment2]}
pigment_map{[0.25 Pigment1][0.25 Pigment2][0.5 Pigment2][0.5 
Pigment3][0.75 Pigment3][0.75 Pigment4]}

No notable slowdown, from 0 to 0.5, you use only Pigment1, and from 0.5 
to 1, it's only Pigment2. Here, it don't mather how many entries there 
are, as long as each transitions are abrupt.

Worst case:
texture{average{[1 texture Texture1][1 texture Texture2]...[1 texture 
Texture255]]

About 255 times slower as all 255 textures need to be evaluated and 
averaged.


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.