Path: chuka.playstation.co.uk!news
From: "SCEE" <dev_support@interactive.sony.com>
Newsgroups: scee.yaroze.programming.3d_graphics
Subject: Re: Accessing screen translated vertices
Date: Wed, 18 Feb 1998 18:09:07 -0000
Organization: PlayStation Net Yaroze (SCEE)
Lines: 107
Message-ID: <6cf7ug$beo41@chuka.playstation.co.uk>
References: <345BFB81.51488BFC@ibm.net> <345C1288.E288E7A@ibm.net> <345C279D.BD5E35D4@chat.carleton.ca> <345CEAA2.446B@peace.co.nz>
NNTP-Posting-Host: 194.203.13.10
X-Newsreader: Microsoft Outlook Express 4.71.1712.3
X-MimeOLE: Produced By Microsoft MimeOLE V4.71.1712.3

Hi James,

I think the lib function ApplyMatrix/SV/LV does this for you. If you look at
the chrome demo ( somewhere on our site ) Stuart did, he used this to
perform the environment mapping for each poly's uv's.

Cheers,

Colin.

James Russell wrote in message <345CEAA2.446B@peace.co.nz>...
>Tim O'Neil wrote:
>>
>> Well, you're right, the process (in theory) is sorta complicated, but
>> really its very simple to do.
>
>I managed to get it working by doing the calculation manually (although
>I didn't realize that when you get a RotMatrix, each value is multiplied
>by "ONE" (4096). So you have to divide by 4096 when you're done.
>
>Essentially my code does this now (of course, it's optimised much more than
this):
>
>
>
>Xo = vertex.x    // Copy the (X,Y,Z) of this vertex.
>Yo = vertex.y
>Zo = vertex.z
>
>rotate.x = pitch;
>rotate.y = heading;
>rotate.z = roll;
>RotMatrix(&rotate, &RotationMatrix); // Creates a rotation matrix from my
heading, pitch and roll
>
>RotationMatrix.t[0] = translateX; // Move the object around.
>RotationMatrix.t[1] = translateY;
>RotationMatrix.t[3] = translateZ;
>
>RM = &RotationMatrix;
>
>Xw = (RM->m[0][0] * Xo  +  RM->m[0][1] * Yo   +  RM->m[0][2] * Zo) / 4096
+  RM->t[0]; // Calculate World space coords
>Yw = (RM->m[1][0] * Xo  +  RM->m[1][1] * Yo   +  RM->m[1][2] * Zo) / 4096
+  RM->t[0];
>Zw = (RM->m[2][0] * Xo  +  RM->m[2][1] * Yo   +  RM->m[2][2] * Zo) / 4096
+  RM->t[0];
>
> // Screen space is the same as world space for me, except the viewpoint is
at (0,0,-1000)
>Zw -= -1000;  // Move everything forward 1000, which is the same as the
viewpoint moving BACK 1000.
>
>
>Using this handy little ASCII art diagram of similar triangles:
>
>          Screen   Point
>        (Ys)| ___---+ (Yw,Zw)
>      ___---+-      |
> __---      |       |
>+___________+_______+_   Z axis  +ve ->
>Eye       Origin
>
><---------->
>Projection distance (h)
>
>We see that (using similar triangles):
> Ys      Yw
> -- = --------
> h   (Zw + h)
>
>Therefore: Ys = (h * Yw) / (Zw + h);
>
>(There are nicer ways of doing this, but they're just optimisations).
>
>Finally:
>
> Xscreen =  (PROJECTION_DISTANCE * Xw) / (Zw + PROJECTION_DISTANCE);
> Yscreen =  (PROJECTION_DISTANCE * Yw) / (Zw + PROJECTION_DISTANCE);
>
>Yay! I finally got the coordinates I was looking for.
>Of course, this only works if your screen space is the same as World space
>
>Something unusual though - The aspect ratio of these points is different to
>what I'd expect - so I guess the Yaroze does some scaling to counter the
>effects of different screen resolutions.
>
>
>I haven't got my Foley-Van Damme book in front of me so I can't check -
>but isn't the transpose of a Rotation Matrix the inverse of that matrix?
>
>
>
>If it is, this would help if I ever wanted to use a dynamic screen space.
>
>Cheers,
>
>J
>
>
>--
>==PEACE COMPUTERS
>==James.Russell@peace.co.nz - 64(9)3730400 -Fax 64(9)3730401
>
>Make like a shepherd and get the flock out of here.