 |
|
|
|
|
|
| |
| |
|
|
From: Dejan D M Milosavljevic
Subject: ANNOUNCE: Acceleration of Plane primitive
Date: 13 Apr 2000 08:35:32
Message: <38f5bf14@news.povray.org>
|
|
|
| |
| |
|
|
Entire source and test scenes at -- Plane patch -- page via
http://members.xoom.com/dmilos/prg/pov/mypovpage.htm
Test no 0:
Official: Time For Trace: 0 hours 1 minutes 33.0 seconds
(93seconds)
Patched : Time For Trace: 0 hours 1 minutes 30.0 seconds (90 seconds)
Official: Peak memory used: 87668 bytes
Patched : Peak memory used: 87597 bytes
Test no 1:
Official: Time For Trace: 0 hours 1 minutes 53.0 seconds
(113seconds)
Patched: Time For Trace: 0 hours 1 minutes 21.0 seconds (81 seconds)
Official: Peak memory used: 132958 bytes
Patched: Peak memory used: 127847 bytes
Any comment?!
--
***************************
dmi### [at] xoommail com
http://members.xoom.com/dmilos/
***************************
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>Any comment?!
This is a nice patch. For those who haven't looked at it, Dejan has used the
fact that linear transformations always transform a plane to a plane
to remove the 'current transform' variable from the plane object, and
has updated the transformation functions to directly modify the normal
and distance of the plane as appropriate. This means that the intersection,
inside, and normal functions can be marginally faster for untransformed
planes, and significantly faster for transformed planes.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 13 Apr 2000 14:28:15 +0200, "Dejan D. M. Milosavljevic"
<dmi### [at] xoommailcom> wrote:
>Entire source and test scenes at -- Plane patch -- page via
>http://members.xoom.com/dmilos/prg/pov/mypovpage.htm
<snip>
>Any comment?!
Yes. I am not a programmer, but I must tell you I really respect your
work with these patches. Correct me if I'm wrong in thinking that the
real art of it seems to be in finding the right spot to correct and
then leaving everyone else slapping their foreheads asking themselves
how they haven't thought of that first :)
This plane patch is a really nifty one (now that Ron has explained
what you've done). Can a similar approach be taken for triangles? IIRC
the intersection point of the test ray and the plane of the triangle
is tested for insidedness using a lemma of one of Jordan's theorems.
This requires several steps of manipulating 3D vectors. Maybe if one
of the coordinates is stripped down (the triangle and intersection
point are projected on a coordinate plane), these calculations would
only require 2D vectors. Of course the right projecting plane should
be chosen.
What do you think? Is it worth the trouble?
Peter Popov ICQ : 15002700
Personal e-mail : pet### [at] usanet
TAG e-mail : pet### [at] tagpovrayorg
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Great! So I guess this one will become part of the official?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>Any comment?!
Just keep 'em comin', Dejan! :)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 13 Apr 2000 19:24:57 -0400, TonyB wrote:
>Great! So I guess this one will become part of the official?
Hard telling. There's the matter of getting permission from the
author, for one thing. There's also the matter of my sig, which
I seem to have misplaced. I don't speak for the POV-Team.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>Hard telling. There's the matter of getting permission from the
>author, for one thing. There's also the matter of my sig, which
>I seem to have misplaced. I don't speak for the POV-Team.
I never said anything about your sig. I just know that if something better
is out there, then it should become official. (With the author's permission,
of course, although I don't know why an author wouldn't want his work to
enter the official version.)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On the other hand... How many times have you used a transformed plane?-)
--
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Peter Popov <pet### [at] usanet> wrote:
: Can a similar approach be taken for triangles?
Why not? Transforming a triangle results in another triangle. The
transformation could be done by changing the vertices of the triangles at
parse time.
--
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Warp wrote:
>
> On the other hand... How many times have you used a transformed plane?-)
>
Is a rotated plane a transformed plane ???
If so, the answer is:
Many, in CSG difference or intersection...
Bouf.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
