Path: chuka.playstation.co.uk!chuka.playstation.co.uk!jdspugh.eecs.utas.edu.au!user
From: littledevil@trump.net.au (Richard Malka)
Newsgroups: scee.yaroze.programming.3d_graphics
Subject: vertex access for collision detection
Date: Sat, 20 Sep 1997 18:57:33 +1000
Organization: Little Devil Software
Lines: 24
Message-ID: <littledevil-2009971857500001@jdspugh.eecs.utas.edu.au>
NNTP-Posting-Host: jdspugh.eecs.utas.edu.au
X-newsreader: MT-NewsWatcher 2.2.2

Hi,

I want to access rotated/translated (i.e. world coordinate) vertices of
objects for collision detection. As I see it, the rotation/translation
matrices for an object get applied when GsSortObject is called. This then
rotates the vertices and stores them in a packet ready to send to the GPU.
It is thus no use looking at the TMD data because that always remains
constant - it just describes the shape of the object in local coordinates.

Do I then have to 'manualy' apply the matrix transforms on my TMD data to
obtain the rotated/translated positions of the vertices I need for
collision detection? This would be annoying because it means I'll have to
write code to emulate the hierachy of transforms that are done
automatically for me at the moment. It would be nice to be able to read
the rotated/translated vertices that are (I pressume) in the packet (that
has been processed by GsSortObject) ready to go to the GPU, but I don't
think this is possible because we don't know the GPU packet data format.

Is this correct? What would be the best way to obtain the
rotated/translated (i.e. world coordinate) values of vertices for
collision detection.

Thanks,
Richard