Path: chuka.playstation.co.uk!news
From: Jake Turner <jaket@revolution.co.uk>
Newsgroups: scee.yaroze.freetalk.english
Subject: Re: Triangles, Quads???
Date: Wed, 14 Oct 1998 22:21:11 +0100
Organization: PlayStation Net Yaroze (SCEE)
Lines: 89
Message-ID: <362515C7.B9E07223@revolution.co.uk>
References: <362238AE.DF3414FF@usa.net> <36228105.D38787F2@datasys.net> <3622FF14.B0A6AA9D@hinge.mistral.co.uk> <3624E1B7.DB009494@usa.net> <01bdf79b$33250e60$016f6f6f@fffaaasssttt>
NNTP-Posting-Host: 62.172.82.78
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.05 [en] (Win95; I)

Peter Dollochan wrote:

> Doesn't the Sony tool DXF2RSD.exe do just that with one of its
> parameters -QUAD sounds about right. I think this alters the vertex
> data instead of adding points.......
>
> Of course I could be talking rubbish....
>
> Pete.
>

No, it does support the feature supposedly.  But converting tris to quads is a
trivial task except for the obvious planar case, so I have never had much joy
out of the dxf2rsd conversion.  There are plugins for 3DS MAX e.g. gamut PSm
but that is costly.

I would look for a tri-strip rather than quads, that has a lot of shared
verticies and so can reap the same benefit as a quad on the playstation.


--

Jake Turner : Court Astrologer
=========================================
Practive VI it may save your life one day


> K.Wutzke <kwutzke@usa.net> wrote in article <3624E1B7.DB009494@usa.net>...
> >
> >
> > Craig Graham wrote:
> >
> > > Darco wrote:
> > >
> > > > The playstation intreprets a quad as 2 triangles. (I know this is
> true
> > > > because I've got the numbering wrong on a quad many times, and the
> GPU
> > > > renders 2 distinct triangles) Look at this rough sketch. (If it
> doesn't
> > > > look right, look at it with a fixed size font)
> > > >
> > > >         0------1
> > > >         |     /|
> > > >         |    / |
> > > >         |   /  |
> > > >         |  /   |
> > > >         | /    |
> > > >         |/     |
> > > >         2------3
> > > >
> > > > Because it does this, it really doesn't matter which you choose. If
> > > > there is a speed difference it will be minimal. However, the data for
> 2
> > > > triangles is larger than 1 quad. So if you are dealing with large
> > > > models, it may be more efficient to use quads rather than triangles.
> > >
> > > A quad should be faster than 2 triangles because the gte only has to do
> > > 4vertex transformations instead of 6 (2 of the vertices are common
> between
> > > the two triangles). It's the same reason that mesh lists are faster on
> > > Direct3D
> > > (less calculations per polygon).
> > >
> > > NOTE: The GPU is much faster than the gte. You can render more
> triangles
> > > than you
> > > can calculate, so any reduction in calculation is a benefit.
> > >
> > > > 'Darco
> > >
> > > Craig.
> >
> >   re there any modelers/tools that can optimize an output model for
> quads?
> > I don't want it to "expand" (quadrangulate) the tris to quads (add a
> vertex),
> > but instead find almost coplanar, adjacent triangles and make these a
> quad,
> > accepting minimal vertex position corrections.
> > Any suggestions? Or will I have to write my one tool? Hnnngh... :-(
> > What can 3D Studio MAX R2 do for that?
> >
> > Karsten
> >
> >