|
Hi
>> That's bad, because I only have tricks
for unions, not for intersections. But did
>> you try to exchange the
kernel? The demo program does not use the
>>
Extended_predicates_exact_constructions_kernel
Hmmm... I am bit confused about
Extended_predicates_exact_constructions_kernel. Are you trying to say about
Extended_homogeneous or Extended_cartesian ? because i did not found
Extended_predicates_exact_constructions_kernel anywhere in header.
Let me know it is already in any CGAL header or i have define myself ?
----- Original Message -----
From:
To:
Sent: Monday, May 11, 2009 9:39 PM
Subject: Re: [cgal-discuss] Nef_3
Question
That's
bad, because I only have tricks for unions, not for intersections. But
did
you try to exchange the kernel? The demo program does not use
the
Extended_predicates_exact_constructions_kernel.
Peter
-----"naresh" <> wrote:
-----
To:
<>
From:
"naresh" <>
Date:
05/11/2009 02:42PM
Subject: Re: [cgal-discuss] Nef_3 Question
Hi
Basically i have 100 silhouette from 2d images
and want to build 3D Polyhydron using Visual Hull. I am trying demo which
given nef_3 package.
Let me know if you need anything more ....
Alex
----- Original Message -----
From:
To:
Sent:
Monday, May 11, 2009 5:53 PM
Subject:
Re: [cgal-discuss] Nef_3 Question
You
have to be more specific. The operations themselves have an overhead,
because they
allow all degeneracies and use exact arithmetic. We would
like to offer more flexibility, here,
but it's unclear when we find the
time to redesign the data structures. In the current state, you
can win much by a) using the perfect kernel (which often is the
Extended_predicates_exact_constructions_kernel), and by b) performing the
operations in
the optimal order. So, if you give me an idea of what
operations you perform, I might give you
a hint on how to save
time.
Peter
-----"naresh" <> wrote:
-----
To:
<>
From: "naresh"
<>
Date: 05/11/2009 01:10PM
Subject:
Re: [cgal-discuss] Nef_3 Question
Hi Peter
>> We use a kd-tree for point
location and ray shooting. But I guess
>> that was not what you
were asking for, was it?
You are right. Basically I read in
Geometeric Tools by Eberly about BSP tree. Acording to him BSP tree will
faster then linear intersection.
But you said it use Kd-tree so it will
do same Job as BSP so speed will be same. In fact i was trying to
search why CGAL implemetation of nef_3 is slower.
BTW. Since you are the auther let me
ask .
Is there any method i can increase
speed because my application has lot of operations. ?
Alex
----- Original Message -----
From:
To:
Sent: Monday, May 11, 2009 3:51 PM
Subject: Re: [cgal-discuss] Nef_3 Question
We
use a kd-tree for point location and ray shooting. But I guess
that
was not what you were asking for, was it? Look into
the manual to
get an idea of the data structure, or have
a look at the papers on
our implementation for a deeper
insight.
http://www.win.tue.nl/~phachenb/Nef/Nef3D-DataStructureAlgorithms.pdf
Peter
-----"naresh" < >
wrote: -----
To: <>
From: "naresh"
<>
Date: 05/11/2009
10:52AM
Subject: Re: [cgal-discuss] Nef_3 Question
Hi
>
Unfortunately, the algorithms from the Nef package can not work
with
> inexact numbers. I think one would need to use a
completely different
> approach to make booleans work with
inexact math, but that would
> probably sacrifice correctness
in certain cases.
Is the polyhydron intersection is
implemented with BSP tree ?
----- Original Message
-----
From: "Bart Janssens"
<>
To:
<>
Sent: Saturday, May 09,
2009 8:57 PM
Subject: Re: [cgal-discuss] Nef_3
Question
> On Sat, May 9, 2009 at 1:20 PM, naresh
<> wrote:
>> I am trying to
find two polyhydron intersection in nef_3 package. But i
>>
found process is very slow because of exact arithmatic
calculations.
>>
>> So i have question can i use
use inexact arithmatic that gives me speed ?
>
> Hi
Alex,
>
> Unfortunately, the algorithms from the Nef
package can not work with
> inexact numbers. I think one would
need to use a completely different
> approach to make booleans
work with inexact math, but that would
> probably sacrifice
correctness in certain cases.
>
>
Cheers,
>
> --
> Bart
> --
> You are
currently subscribed to cgal-discuss.
> To unsubscribe or
access the archives, go to
> https://lists-sop.inria.fr/wws/info/cgal-discuss
>
--
You are currently subscribed to
cgal-discuss.
To unsubscribe or access the archives, go to
https://lists-sop.inria.fr/wws/info/cgal-discuss
|
