|
 |
As I was helping out Josh in povray.general with his "Object Pattern in
Media" question, I noticed a strange behavior with media method 3 and shadow
rays.
This media has the following sample parameters:
method 3
intervals 1
samples 3
aa_level 20
The intended behavior is that 3 samples should be taken, and then more
samples between them if they differ. This produces a very smooth result as
you can see in the image, with only a few artifacts near the edges of the
density shape.
The strange thing is that the shadow is not smooth. Neither the shadow on
the ground nor the self-shadowing of the media is smooth. It appears that
the recursive sampling is not happening for shadow rays! You can see that
there are only 4 levels of brightness on the ground, probably corresponding
to the amount of the 3 samples which are taken that have density.
Is this intended behavior? Although it may be convenient sometimes for
shadow rays to have less accuracy, there should be a way to specify how
accurate they should be, shouldn't there?
This is the scene code for anyone who wants to try this out themself:
#include "colors.inc"
camera {
location <0, 5, -40>*.8
look_at <0, 0, 0>
}
light_source {<-200, 1000, -400> White}
plane {
y, -20
pigment { color rgb <.8,.5,.5> }
}
#declare myObject = union {
cylinder {
<8,0,0>,
<-8,0,0>,
3
}
sphere {
<0,0,0>
,5
}
}
sphere {
<0,0,0>,
10
texture { pigment { Clear } }
interior {
media {
scattering {1, 0.1}
method 3
intervals 1
samples 3
aa_level 20
density {
pigment_pattern {object {myObject 0, 1}}
color_map {[0 rgb 0][1 rgb 1]}
}
}
}
hollow
}
- Slime
[ http://www.slimeland.com/ ]
Post a reply to this message
Attachments:
Download 'temp.jpg' (21 KB)
Preview of image 'temp.jpg'
|
|
|
|
> Slime, I've knocked up a sketch animation, and like a [DELETED] I've gone
a
> posted it in p.a., instead of p.b.a.
Looks cool. You can cancel the message (in outlook express select the
message, go to message -> cancel message) and repost it in p.b.a. Anyone who
comes along after you cancel it won't see it there.
- Slime
[ http://www.slimeland.com/ ]
Post a reply to this message
|
|
