 |
 |
|
 |
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Has anyone come up with fast rendering solid poligonal shapes? What I'm
looking for is the tetrahedron, octahedron, etc. The ones in shapes2.inc
are SLOW, more than 15 times slower than a sphere, for instance. I've
looked through the links on the POV server, but have yet to find anything I
can use.
The shapes need to be solid so I can CSG some numbers and/or letters into
the faces. Even if the shapes are just a little faster than the ones in
shapes2.inc, I'd appreciate seeing them. The picture I posted the other
day called GlassSpheres.jpg was rendered in a little over an hour;
replacing the 64 spheres with dodecahedrons and re-rendering took 17.5
hours!!
--
Rich Allen
(Remove SPAM from my address to reply by e-mail)
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Rich wrote:
>
> Has anyone come up with fast rendering solid poligonal shapes? What I'm
> looking for is the tetrahedron, octahedron, etc. The ones in shapes2.inc
> are SLOW, more than 15 times slower than a sphere, for instance. I've
> looked through the links on the POV server, but have yet to find anything I
> can use.
> The shapes need to be solid so I can CSG some numbers and/or letters into
> the faces. Even if the shapes are just a little faster than the ones in
> shapes2.inc, I'd appreciate seeing them. The picture I posted the other
> day called GlassSpheres.jpg was rendered in a little over an hour;
> replacing the 64 spheres with dodecahedrons and re-rendering took 17.5
> hours!!
>
Since these shapes are intersections of planes, you sould try manual
bounding which can speed up rendering a lot, esp. if many planes are
invloved.
Another possibility would be using a mesh, but you can't CSG it, unless
you use megapov and it's solid mesh feature.
Christoph
--
Christoph Hormann <chr### [at] gmx de>
IsoWood include, radiosity tutorial, TransSkin and other
things on: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Christoph Hormann <chr### [at] gmx de> wrote:
: Another possibility would be using a mesh, but you can't CSG it, unless
: you use megapov and it's solid mesh feature.
This is definitely the fastest solution.
--
#local D=array[6]{11117333955,7382340,3358,3900569407,970,4254934330}
#local I=0;#macro M()<mod(D[I],13)-6,mod(div(D[I],13),8)-3,10>#end
blob{#while(I<6)cylinder{M()#local D[I]=div(D[I],104);M().5,2
pigment{rgb M()}}#local I=(D[I]>99?I:I+1);#end} /*- Warp -*/
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Wasn't it Rich who wrote:
> Has anyone come up with fast rendering solid poligonal shapes? What I'm
>looking for is the tetrahedron, octahedron, etc. The ones in shapes2.inc
>are SLOW, more than 15 times slower than a sphere, for instance. I've
>looked through the links on the POV server, but have yet to find anything I
>can use.
> The shapes need to be solid so I can CSG some numbers and/or letters into
>the faces. Even if the shapes are just a little faster than the ones in
>shapes2.inc, I'd appreciate seeing them. The picture I posted the other
>day called GlassSpheres.jpg was rendered in a little over an hour;
>replacing the 64 spheres with dodecahedrons and re-rendering took 17.5
>hours!!
>
Try this:-
#include "shapes.inc"
#include "shapes2.inc"
#declare Bounded_Dodecahedron = object {Dodecahedron
bounded_by {sphere {0,1.26}}
}
If you've got lots of small dodecahedrons in your scene, then these
Bounded_Dodecahedrons will render *much* faster.
The reason is that POV isn't smart enough to work out that the
intersection of the 12 planes that define the ordinary dodecahedron is
bounded, so automatic bounding slabs don't get generated. So, for every
ray, POV has to check the ray's intersection with 12 planes for each
dodecahedron.
When you manually apply bounds to the object, POV performs one sphere
intersection test for each bounding sphere, and then only goes on to
perform the 12 plane tests in the few cases where the bounds check
succeeds.
The speed improvement depends on how likely each ray is to miss each
dodecahedron. I'd guess that, for your image, Bounded_Dodecahedrons
would render in less than twice the time taken for spheres. In an image
with a few large dodecahedrons there wouldn't be much speed improvement.
--
Mike Williams
Gentleman of Leisure
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Warp <war### [at] tag povray org> wrote in <3aeff675@news.povray.org>:
> Christoph Hormann <chr### [at] gmx de> wrote:
>: Another possibility would be using a mesh, but you can't CSG it, unless
>: you use megapov and it's solid mesh feature.
>
> This is definitely the fastest solution.
Thanks guys, I'll see what I can do with meshes. Any pointers to
already-calculated mesh files? <smile>
--
Rich Allen
(Remove SPAM from my address to reply by e-mail)
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Mike Williams <mik### [at] nospam please> wrote in
<Cjf### [at] econym demon co uk>:
> Try this:-
>
> #include "shapes.inc"
> #include "shapes2.inc"
>
> #declare Bounded_Dodecahedron = object {Dodecahedron
> bounded_by {sphere {0,1.26}}
> }
Thanks for that explanation! I had discarded manual bounding as an
option without even trying it, since I didn't know POV wasn't auto-bounding
these objects. I'll try your suggestion out while I'm looking for mesh
versions of the solids. Thanks again!
--
Rich Allen
(Remove SPAM from my address to reply by e-mail)
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
SrP### [at] ricosweb com (Rich) wrote in
<Xns9096AD6DC7321spammindspringcom@204.213.191.228>:
> Mike Williams <mik### [at] nospam please> wrote in
><Cjf### [at] econym demon co uk>:
>
>> Try this:-
>>
>> #include "shapes.inc"
>> #include "shapes2.inc"
>>
>> #declare Bounded_Dodecahedron = object {Dodecahedron
>> bounded_by {sphere {0,1.26}} }
>
> Thanks for that explanation! I had discarded manual bounding as an
> option without even trying it, since I didn't know POV wasn't
> auto-bounding these objects. I'll try your suggestion out while I'm
> looking for mesh versions of the solids. Thanks again!
>
Woohoo! This option works great! Replacing the spheres in my scene with
the bounded dodecahedrons resulted in a 1h13m render, versus 1h2m render
for spheres, and 17h30m render for unbounded dodecahedrons. I can't
believe the difference bounding makes in this case.
Thanks again!
--
Rich Allen
(Remove SPAM from my address to reply by e-mail)
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
Rich <SrP### [at] ricosweb com> wrote:
: Woohoo! This option works great! Replacing the spheres in my scene with
: the bounded dodecahedrons resulted in a 1h13m render, versus 1h2m render
: for spheres, and 17h30m render for unbounded dodecahedrons. I can't
: believe the difference bounding makes in this case.
Don't yet discard triangle meshes completely. With them that same scene
will probably render in 10 minutes or something like that...
(I have rendered a scene with more than 300 millions of triangles at
800x600 with antialiasing and it took a bit more than 1 minute to render
in this Ultra5, which is about the same speed as an P-II 400MHz.)
--
#local D=array[6]{11117333955,7382340,3358,3900569407,970,4254934330}
#local I=0;#macro M()<mod(D[I],13)-6,mod(div(D[I],13),8)-3,10>#end
blob{#while(I<6)cylinder{M()#local D[I]=div(D[I],104);M().5,2
pigment{rgb M()}}#local I=(D[I]>99?I:I+1);#end} /*- Warp -*/
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
In article <3af29a80@news.povray.org>, war### [at] tag povray org says...
> Rich <SrP### [at] ricosweb com> wrote:
> : Woohoo! This option works great! Replacing the spheres in my scene with
> : the bounded dodecahedrons resulted in a 1h13m render, versus 1h2m render
> : for spheres, and 17h30m render for unbounded dodecahedrons. I can't
> : believe the difference bounding makes in this case.
>
> Don't yet discard triangle meshes completely. With them that same scene
> will probably render in 10 minutes or something like that...
>
> (I have rendered a scene with more than 300 millions of triangles at
> 800x600 with antialiasing and it took a bit more than 1 minute to render
> in this Ultra5, which is about the same speed as an P-II 400MHz.)
How on earth was it possible to render such a monster model in 1 minute ?
I have a mesh of approximately 1 million smooth triangles which takes a
long time to render, mainly because of the long parse time (the file
containing the triangles is approximately 150 MB). Is the parse time not
included in the 1 minute time ? And doesn't this also require a *lot* of
memory ? I'd love to be able to render bigger stuff than my 1 million
triangles; they are a 'subsampled' version of the original model in order
to get acceptable speed.
Bjorn Jonsson / bjj### [at] zzzmmedia is
Address changed to avoid junk email. Remove yyy and
zzz to reply.
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|  |
|
 |
> > (I have rendered a scene with more than 300 millions of triangles at
> > 800x600 with antialiasing and it took a bit more than 1 minute to render
> > in this Ultra5, which is about the same speed as an P-II 400MHz.)
>
> How on earth was it possible to render such a monster model in 1 minute ?
> I have a mesh of approximately 1 million smooth triangles which takes a
> long time to render, mainly because of the long parse time (the file
> containing the triangles is approximately 150 MB). Is the parse time not
> included in the 1 minute time ? And doesn't this also require a *lot* of
> memory ? I'd love to be able to render bigger stuff than my 1 million
> triangles; they are a 'subsampled' version of the original model in order
> to get acceptable speed.
I think he means that he had a scene with multiple copies of one smaller
mesh object.
A single object with that many triangles will always, as far as I know,
take a long time to parse.
Bye for now,
Jamie.
Post a reply to this message
|
 |
|  |
|  |
|
 |
|
 |
|  |
|
 |