Re: mesh to 3D texture?

Christian Heckl <[email protected]> Fri, 26 Aug 2011 12:28:59 +0200
Newsgroups gmane.games.devel.algorithms
Message-ID <[email protected]>
This is a multi-part message in MIME format.
--===============6853149524542636484==
Content-Type: multipart/alternative;
	boundary="------------010302020403080802000007"

This is a multi-part message in MIME format.
--------------010302020403080802000007
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi Jeff,

recently did some research myself, haven't yet come to the 
implementation side of things, so can't say much about real-world 
applicability:

These are IMHO the most promising results, when it comes to 
non-exlusively-binary voxelization:

Real-time GPU-based Voxelization and Applications (March 2011)
http://www.cs.wisc.edu/techreports/viewreport.php?report=1688

Two Simple Single-pass GPU methods for Multi-channel Surface 
Voxelization of Dynamic Scenes (Pacific Graphics 2011)
http://graphics.cs.aueb.gr/graphics/docs/papers/PG-2011optimized.pdf

They both give a nice wrap-up of previous work (especially Eisemann) and 
discuss the kind of problems you are worried about.
I will go for the second paper as basis for my implementation, not 
completely sure about performance though.

@all: if anyone has experience with those two papers, I'd be glad to know :)

Chris

> So, suppose that I wanted to voxelize a mesh into a regular 3D grid (a 
> 3D texture on the GPU, fairly low res, something like 64^3). At each 
> point I'd like to have access to color, a (rough) surface normal, and 
> some kind of indicator of whether the space is "empty" or "filled".
>
> Is there some clever way I can use GPU rasterization to fill it 
> (ideally keeping the whole process on the GPU)? I thought about 
> rendering the mesh in slices with very narrow front and back clip 
> planes, once for each slice of the 3D texture, but it seems like there 
> could be issues with polygons perpendicular to my z axis not showing 
> up, creating holes...
>
> Any advice?
>
> -- 
> Jeff Russell
> Engineer, 8monkey Labs
> www.8monkeylabs.com <http://www.8monkeylabs.com>
>
>
> ------------------------------------------------------------------------------
> EMC VNX: the world's simplest storage, starting under $10K
> The only unified storage solution that offers unified management
> Up to 160% more powerful than alternatives and 25% more efficient.
> Guaranteed. http://p.sf.net/sfu/emc-vnx-dev2dev
>
>
> _______________________________________________
> GDAlgorithms-list mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list


--------------010302020403080802000007
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Hi Jeff,<br>
    <br>
    recently did some research myself, haven't yet come to the
    implementation side of things, so can't say much about real-world
    applicability:<br>
    <br>
    These are IMHO the most promising results, when it comes to
    non-exlusively-binary voxelization:<br>
    <br>
    Real-time GPU-based Voxelization and Applications (March 2011)<br>
    <a
      href="http://www.cs.wisc.edu/techreports/viewreport.php?report=1688">http://www.cs.wisc.edu/techreports/viewreport.php?report=1688</a><br>
    <br>
    Two Simple Single-pass GPU methods for Multi-channel Surface
    Voxelization of Dynamic Scenes (Pacific Graphics 2011)<br>
    <a
href="http://graphics.cs.aueb.gr/graphics/docs/papers/PG-2011optimized.pdf">http://graphics.cs.aueb.gr/graphics/docs/papers/PG-2011optimized.pdf</a><br>
    <br>
    They both give a nice wrap-up of previous work (especially Eisemann)
    and discuss the kind of problems you are worried about.<br>
    I will go for the second paper as basis for my implementation, not
    completely sure about performance though.<br>
    <br>
    @all: if anyone has experience with those two papers, I'd be glad to
    know :)<br>
    <br>
    Chris<br>
    <br>
    <blockquote
cite="mid:CANLWWykAUyCZGmS__Y=TSMUPZ0Xk-e_d46512ZSCCv06q1SKmw@mail.gmail.com"
      type="cite">So, suppose that I wanted to voxelize a mesh into a
      regular 3D grid (a 3D texture on the GPU, fairly low res,
      something like 64^3). At each point I'd like to have access to
      color, a (rough) surface normal, and some kind of indicator of
      whether the space is "empty" or "filled".<br>
      <br>
      Is there some clever way I can use GPU rasterization to fill it
      (ideally keeping the whole process on the GPU)? I thought about
      rendering the mesh in slices with very narrow front and back clip
      planes, once for each slice of the 3D texture, but it seems like
      there could be issues with polygons perpendicular to my z axis not
      showing up, creating holes...<br>
      <br>
      Any advice?<br clear="all">
      <br>
      -- <br>
      Jeff Russell<br>
      Engineer, 8monkey Labs<br>
      <a moz-do-not-send="true" href="http://www.8monkeylabs.com"
        target="_blank">www.8monkeylabs.com</a><br>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">------------------------------------------------------------------------------
EMC VNX: the world's simplest storage, starting under $10K
The only unified storage solution that offers unified management 
Up to 160% more powerful than alternatives and 25% more efficient. 
Guaranteed. <a class="moz-txt-link-freetext" href="http://p.sf.net/sfu/emc-vnx-dev2dev">http://p.sf.net/sfu/emc-vnx-dev2dev</a></pre>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
GDAlgorithms-list mailing list
<a class="moz-txt-link-abbreviated" href="mailto:[email protected]">[email protected]</a>
<a class="moz-txt-link-freetext" href="https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list">https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list</a>
Archives:
<a class="moz-txt-link-freetext" href="http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list">http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list</a></pre>
    </blockquote>
    <br>
  </body>
</html>

--------------010302020403080802000007--


--===============6853149524542636484==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

------------------------------------------------------------------------------
EMC VNX: the world's simplest storage, starting under $10K
The only unified storage solution that offers unified management 
Up to 160% more powerful than alternatives and 25% more efficient. 
Guaranteed. http://p.sf.net/sfu/emc-vnx-dev2dev
--===============6853149524542636484==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
GDAlgorithms-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
Archives:
http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
--===============6853149524542636484==--