From - Tue Aug 20 17:35:28 2002
X-Mozilla-Status: 0001
X-Mozilla-Status2: 00000000
Received: from [150.237.196.2] (helo=mailhub3.hull.ac.uk)
	by mailhub.dcs.hull.ac.uk with esmtp (Exim 3.03 #2)
	id 17h9uH-0005Fq-00
	for mv@dcs.hull.ac.uk; Tue, 20 Aug 2002 15:22:53 +0100
Received: from web10108.mail.yahoo.com by puccini.ucc.hull.ac.uk; Tue, 20 Aug 2002 15:21:41 +0100
Message-ID: <20020820142138.55188.qmail@web10108.mail.yahoo.com>
Received: from [137.203.131.55] by web10108.mail.yahoo.com via HTTP; Tue, 20 Aug 2002 07:21:38 PDT
Date: Tue, 20 Aug 2002 07:21:38 -0700 (PDT)
From: Ramesh Raskar <rameshraskar@yahoo.com>
Subject: Re: Two Pass Algorithm for Silhouette Rendering
To: Mahes Visvalingam <m.visvalingam@dcs.hull.ac.uk>
In-Reply-To: <006901c234bf$0d887600$7a5ced96@dcs.hull.ac.uk>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii


  Mahes, what is the difference between silhouette and occluding contour 
  as far as the displaying them for 3D polygonal scenes is concerned ?
  
  If you want the best quality silhouettes, it probably best to do
  it in a purely software method .. find
  the candidate silhouette edges (by taking dot product with view vector
  and normals of the two adjacent polgyons of that edge) and then running
  simple z-buffer visibility on them.

  If you want to avoid pre-processing, do not want to deal with
  computing connectivity information (and want to implement in hardware)
  then our methods are more useful. The wireframe method comes closest
  to the pure-software method described above but it is slow. The polygon
  fatterning method is more efficient and flexible.

  Check the latest book Real Time Rendering
  http://www.realtimerendering.com/
  http://www.acm.org/tog/resources/RTR/rtr2_npr.pdf (first few pages)

  Let me know what you think.


--- Mahes Visvalingam <m.visvalingam@dcs.hull.ac.uk> wrote:
> Hello Ramesh and Michael
> 
> I am interested in NPAR of terrain - the following page provides a brief
> introduction.
> http://www2.dcs.hull.ac.uk/CISRG/projects/Royal-Inst/index.html
> 
> We have been evaluating a few silhouette rendering algorithms and
> included yours for several reasons, esp the effort to use polygon based
> methods.  Unfortunately, the results were somewhat disappointing.  We
> have written a report for the benefit of researchers in Digital
> Cartography and GIS - and others.  I always have an uncomfortable
> feeling when I cannot get good results from the work of others, that we
> might have goofed.  I would be grateful if you could put me right if I
> have made a mistake and I would be happy to change our report.  I know
> that Ramesh is continuing to work on this approach - I have seen your
> more recent paper.  Although polygon fattening seems to pose problems in
> scientific visualisation of terrain, I can see that it has value in
> other areas.  For example, it will certainly be useful in student
> projects related to our MSc in Computer Graphics and Game Development.
> 
> http://www2.dcs.hull.ac.uk/CISRG/projects/Silhouettes/R+C.html
> 
> I would appreciate your comments.  Thanks.
> 
> mahes
> 
> =================================
> Dr Mahes Visvalingam
> Reader in Digital Cartography
> Department of Computer Science
> The University of Hull
> HULL
> HU6 7RX
> UK
> 
> m.visvalingam@dcs.hull.ac.uk
> http://www2.dcs.hull.ac.uk/CISRG/
> 


__________________________________________________
Do You Yahoo!?
HotJobs - Search Thousands of New Jobs
http://www.hotjobs.com