Paris <par### [at] lycoscom> wrote:
> Physically-based rendering models of glass use a fresnel term.

  Could you please explain what 'fresnel' in POV-Ray (see the documentation
for details) does differently than what you want?

> Someone's reply to my take on pov-ray's global illumination was not only
> rude and uncooth, but it was wrong as well.   Pov-ray "gathers" global
> illumination by shooting out rays in a random way based loosely on a stored
> table of directions.  Every single piece of literature on 3d graphics
> rendering calls this method *DISTRIBUTED RAY TRACING*.  (simply google
> "distributed ray tracing")  Pov-ray does not use monte carlo based
> path-tracing to simulate global illumination.

  Could you please explain us what is the difference between a
monte carlo based global illumination technique and distributed
raytracing?

>  The person who replied to me
> (whose name I will leave out here) was just dead wrong on this issue.
> Distributed ray tracing suffers from shooting rays often into completely
> dark areas of the scene.  This is a wasted calculation since just
> rgb<0,0,0> rays don't contribute anything at all to the shaded point
> mathematically.

  And how does a monte carlo method help with this problem?

  Also, you wouldn't mind explaining us how can you determine a part
of the scene is so dark that it does not contribute in the illumination
of a certain part of the scene without actually looking what is the
illumination in that dark part of the scene?

>    There is no way to completely avoid this from happening,
> but there are methods to make it happen less, such as bi-directional
> path-tracing techniques that use at least some information about the light
> coming out of the light sources.

  But light sources are not the only elements affecting the lighting of
a scene. Some surfaces may have brightness on their own, and there can
be other illuminating effects in the scene as well (eg. media).

> Someone, in their reply, asked me if I "realized the cost" of implementing
> this.  I'm not sure what that meant.  It's only slightly slower in terms of
> CPU clock cycles needed, because the slowest parts of ray-tracing (in
> general) is calculating ray intersections with surfaces, not in the
> shading.  Anyone should realize that spectral integration is a different
> *SHADING* technique not a radical overhall of the geometry.  Dispersion in
> prisms is only ONE of the things this technique gives.

  Could you please explain us how to calculate the dispersion of a prism
so that the rendering becomes "only slightly slower in terms of CPU
clock cycles needed"?

  Btw, do you have any measurement data showing that the heaviest part of
raytracing is the ray intersection process in the average scene?

> Actually that picture does not
> look like brushed metal at all to me.

  Isn't that a question of opinion?
  Your opinion seems prejudiced in that you are probably thinking like
"POV-Ray does not use this 'physically accurate model' I am thinking of
and thus this image does not look like brushed metal".

> Already POV-ray
> can calculate photorealistic blooms around light sources using atmosphere,
> but not yet around bright surface highlights. For an idea of what a bloom
> is, look at the bell tower in this image:

> http://paris1839.tripod.com/computers/afterain.jpg

  You do realize that image has been rendered with POV-Ray?

> My point is that Pov-Ray is lagging behind commercial packages
> in terms of photorealism.

  POV-Ray is not a commercial package, so what do you want?
  If you are not happy with it, then either use one of your commercial
packages (which you will pay for, naturally, if you have even the
smallest amount of honesty), or change POV-Ray to suit your needs (the
source code is available for you to modify).

  If you really want for other people to implement what you can't or
have no time to, then you might better revise your tone of voice and
get some more specific information on things.

-- 
plane{-x+y,-1pigment{bozo color_map{[0rgb x][1rgb x+y]}turbulence 1}}
sphere{0,2pigment{rgbt 1}interior{media{emission 1density{spherical
density_map{[0rgb 0][.5rgb<1,.5>][1rgb 1]}turbulence.9}}}scale
<1,1,3>hollow}text{ttf"timrom""Warp".1,0translate<-1,-.1,2>}//  - Warp -

Post a reply to this message