 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Perhaps this question is not about animations, per se, but I'm trying to
follow a select few of these povray newsgroups.
I am working on an animation. I had the characters running on a plane.
The render time for 50 frames was maybe 12 hours. I wanted to make them
running instead on one long narrow strip. I set up some CSG with the
plane, by looking at the intersection of three planes. The render time
increased to over 24 hours. I don't have the parsing vs. true render
time, etc., in front of me. Does CSG with infinite objects greatly
increase overall render time?
Would a 2000 unit long box take less render time than a CSG'ed plane?
________________
Greg M. Johnson
"Volunteer firefighters is fine, ma'am, but don't tell me
there ain't nobody burnin' in that buildin'," said the Fire Chief to the
Libertarian Lady.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Greg M. Johnson wrote:
>
> Perhaps this question is not about animations, per se, but I'm trying to
> follow a select few of these povray newsgroups.
>
> I am working on an animation. I had the characters running on a plane.
> The render time for 50 frames was maybe 12 hours. I wanted to make them
> running instead on one long narrow strip. I set up some CSG with the
> plane, by looking at the intersection of three planes. The render time
> increased to over 24 hours. I don't have the parsing vs. true render
> time, etc., in front of me. Does CSG with infinite objects greatly
> increase overall render time?
>
> Would a 2000 unit long box take less render time than a CSG'ed plane?
>
> ________________
> Greg M. Johnson
> "Volunteer firefighters is fine, ma'am, but don't tell me
> there ain't nobody burnin' in that buildin'," said the Fire Chief to the
> Libertarian Lady.
Yes.
To make it even worse true adding reflection or transparancy.
Remco
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
The box object, being one of the primary primitives
in POV, has been heavily optimized for rendering speed.
Since the box object can be scaled to any dimension it
certainly has it's advantages.
I have experienced some weird situations creating an
object using CSG's that experienced long render times.
In reality if a CSG is done correctly it should render
almost as fast as a box.
Look at shapes.old, shapes.inc, and shapes2.inc
for examples of using planes to createvarious shapes.
It could be you have one of your surface normals
pointing the wrong way and that would increase render
time.
You might also try a difference instead of an
intersection.
Ken Tyler
Greg M. Johnson wrote:
> Perhaps this question is not about animations, per se, but I'm trying to
> follow a select few of these povray newsgroups.
>
> I am working on an animation. I had the characters running on a plane.
> The render time for 50 frames was maybe 12 hours. I wanted to make them
> running instead on one long narrow strip. I set up some CSG with the
> plane, by looking at the intersection of three planes. The render time
> increased to over 24 hours. I don't have the parsing vs. true render
> time, etc., in front of me. Does CSG with infinite objects greatly
> increase overall render time?
>
> Would a 2000 unit long box take less render time than a CSG'ed plane?
>
> ________________
> Greg M. Johnson
> "Volunteer firefighters is fine, ma'am, but don't tell me
> there ain't nobody burnin' in that buildin'," said the Fire Chief to the
> Libertarian Lady.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> You might also try a difference instead of an
> intersection.
>
> Ken Tyler
>
I'm a relative newcomer with POV but I've had some experiences with CSG
and found that especially difference took a very long time to render. I
didn't expect that, nor did I expect blobs to be quite fast in
comparison. I must add that I'm working on some images using differences
with non-primitive forms. I guess that makes it even more complicated.
Is there any 'rule of thumb' in this?
Remco
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Remco de Korte wrote:
> > You might also try a difference instead of an
> > intersection.
> >
> > Ken Tyler
> >
> I'm a relative newcomer with POV but I've had some experiences with CSG
> and found that especially difference took a very long time to render. I
> didn't expect that, nor did I expect blobs to be quite fast in
> comparison. I must add that I'm working on some images using differences
> with non-primitive forms. I guess that makes it even more complicated.
>
> Is there any 'rule of thumb' in this?
>
> Remco
Be careful about which way the surface normal is
pointing. It took me long time to get the hang of
it and I still find myself taking a look at the
different shapes inc files on occasion for ideas
and to figure out why something isn't working.
Differences don't necessarily have to take a long
time to render.
Take the following as an example:
difference{
sphere{<0,0, 0.75>,1}
sphere{<0,0,-0.75>,1}}
This will produce a lens shaped object. Add
a glass texture and an ior value and it will
work just like a magnifying glass. Try rendering
it with a regular pigment attached and you will
find it no slower than a box. I was working with
this shape last night so I can verify this.
Also the pyramid objects in the includes render
with no speed penalty that I can detect.
I believe whenever there is a severe time increase
from a CSG it has something to do with not having your
inside and outside properly oriented. Similar to the
error: warning camera is inside a non hollow object
you get if you place your camera on the wrong side of
a plane's surface normal, or you put your camera in a
big sphere, that is your sky, and you forget to use
the inverse/hollow keywords, that generate the same
warning.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> I believe whenever there is a severe time increase
> from a CSG it has something to do with not having your
> inside and outside properly oriented. Similar to the
> error: warning camera is inside a non hollow object
> you get if you place your camera on the wrong side of
> a plane's surface normal, or you put your camera in a
> big sphere, that is your sky, and you forget to use
> the inverse/hollow keywords, that generate the same
> warning.
Oops! That might be something I have done wrong all along...
(goes to show how "green" I am ;-))
I'll look into that right away.
Thanks!
Remco
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I don't have my .POV infront of me, but I suspect that I did not commit the
kind of error that Ken is talking about.
I had one plane with a Y normal. I wanted to change in into an infinitely long
"runway." I CSG'ed it with two planes with normals parallel to the X line.
Now I know I had to try several iterations of +x, -x, and / or +/- distance
along normal before I got the strip to look that way I wanted it to look.
Is there a "wrong" way to get the right output? If I had made an error (and I
did several times), I would imagine I'd get something totally wrong looking,
as opposed to an extra-slow-rendering version of exactly what I wanted.
Again, the render time was 2X for a CSG'ed plane (runway) than for a simple
plane.
Ken wrote:
> Be careful about which way the surface normal is
> pointing. It took me long time to get the hang of
> it and I still find myself taking a look at the
> different shapes inc files on occasion for ideas
> and to figure out why something isn't working.
> Differences don't necessarily have to take a long
> time to render.
Post a reply to this message
Attachments:
Download 'us-ascii' (2 KB)
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Greg M. Johnson wrote:
> I don't have my .POV infront of me, but I suspect that I did not
> commit the kind of error that Ken is talking about.
>
> I had one plane with a Y normal. I wanted to change in into an
> infinitely long "runway." I CSG'ed it with two planes with normals
> parallel to the X line. Now I know I had to try several iterations
> of +x, -x, and / or +/- distance along normal before I got the strip
> to look that way I wanted it to look.
>
> Is there a "wrong" way to get the right output? If I had made an
> error (and I did several times), I would imagine I'd get something
> totally wrong looking, as opposed to an extra-slow-rendering version
> of exactly what I wanted. Again, the render time was 2X for a CSG'ed
> plane (runway) than for a simple plane.
>
Compare the following with the script you are using.
camera{location z*-3 look_at 0}
light_source{<0,0,-30>rgb 1}
intersection{
plane{y,-5}
plane{x, 5}
plane{x,-5 inverse}
pigment{rgb 1}}
vs.
box{-1,1 scale<5,2,10000> pigment{rgb 1}translate y*-5}
vs
your CSG
I had no time difference between these two objects when rendered
seperately
and they appeared identical in the scene.
They each took exactly 6 sec. to render at 320x240 AA off.
Ken
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Greg M. Johnson wrote:
> I am working on an animation. I had the characters running on a plane.
> The render time for 50 frames was maybe 12 hours. I wanted to make them
> running instead on one long narrow strip. I set up some CSG with the
> plane, by looking at the intersection of three planes. The render time
> increased to over 24 hours. I don't have the parsing vs. true render
> time, etc., in front of me. Does CSG with infinite objects greatly
> increase overall render time?
>
> Would a 2000 unit long box take less render time than a CSG'ed plane?
> ________________
> Greg M. Johnson
> "Volunteer firefighters is fine, ma'am, but don't tell me
> there ain't nobody burnin' in that buildin'," said the Fire Chief to the
> Libertarian Lady.
I found a good on line tutorial detailing the use of csg
operations in POV-Ray. It offers straight forward examples
and text based explanations that are well thought out and
easy to understand.
Take a look at:
http://tqd.advanced.org/3285/language/csg.html#intersection
Ken Tyler
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Greg M. Johnson <"gregj56590:-)"@aol.com> wrote:
: I am working on an animation. I had the characters running on a plane.
: The render time for 50 frames was maybe 12 hours. I wanted to make them
: running instead on one long narrow strip. I set up some CSG with the
: plane, by looking at the intersection of three planes. The render time
: increased to over 24 hours. I don't have the parsing vs. true render
: time, etc., in front of me. Does CSG with infinite objects greatly
: increase overall render time?
Yes.
Each infinite object is tested for each pixel. If you have 1000 infinite
objects (like planes), povray will make 1000 tests for each pixel although
the intersections of the planes was a little polyhedron in the middle of
the screen.
This can be speeded up with bounding boxes. If you bound this little
polyhedron with a properly scaled box (which encloses the polyhedron completely
inside it), then the 1000 tests will be made only when the ray hits the
box and nowhere else.
: Would a 2000 unit long box take less render time than a CSG'ed plane?
Absolutely. A box object is highly optimized and it will surely render
much faster than a box made by intersecting planes.
Only the size of the box on screen affects the render time, not the size
of the box on povray units.
--
- Warp. -
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
