G3dView.mesa
Copyright Ó 1988, 1992 by Xerox Corporation. All rights reserved.
Bloomenthal, July 14, 1992 1:51 pm PDT
DIRECTORY G3dBasic, G3dMatrix, Imager, Rope, ViewerClasses;
G3dView: CEDAR DEFINITIONS
~ BEGIN
Errors
Error:    ERROR [code: ATOM, reason: ROPE];
Types and Constants
Pair:    TYPE ~ G3dBasic.Pair;
Triple:   TYPE ~ G3dBasic.Triple;
PairSequence: TYPE ~ G3dBasic.PairSequence;
TripleSequence: TYPE ~ G3dBasic.TripleSequence;
Matrix:   TYPE ~ G3dMatrix.Matrix;
Viewport:   TYPE ~ G3dMatrix.Viewport;
ROPE:    TYPE ~ Rope.ROPE;
Viewer:   TYPE ~ ViewerClasses.Viewer;
origin:   Triple ~ G3dBasic.origin;
xAxis:    Triple ~ G3dBasic.xAxis;
yAxis:   Triple ~ G3dBasic.yAxis;
zAxis:    Triple ~ G3dBasic.zAxis;
Camera:   TYPE ~ REF CameraRep;
CameraRep:  TYPE ~ RECORD [
move, rotate:    Triple ¬ [],
scale:      REAL ¬ 1.0,
fieldOfView:    REAL ¬ 60.0,
roll:      REAL ¬ 0.0,
eyePoint, lookAt, up: Triple ¬ [],
matrix:     Matrix ¬ NIL
];
PairClip:   TYPE ~ RECORD [pair: Pair, clipped: BOOL];
GetViewport: PROC [viewer: Viewer ¬ NIL, context: Imager.Context ¬ NIL]
RETURNS [Viewport];
Return viewport to fit [-1..1] (in x, y) image to context or viewer (whichever is non-NIL).
If viewer = NIL and context is for an Interpress master, viewport is set to 8-1/2 by 11 inches.
Transformation
TransformAndClipInZ: PROC [point: Triple, view: Matrix, viewport: Viewport ¬ []]
RETURNS [PairClip];
Transform the point and test if it's clipped in z only (let Imager clip in x and y).
PairsFromTriples: PROC [
triples: TripleSequence,
view: Matrix,
pairs: PairSequence ¬ NIL]
RETURNS [PairSequence];
Return sequence of vertices transformed into screen space by view; store in pairs if non-nil.
Transformation Creation and Translation
TransformByViewport: PROC [viewer: Viewer, in: Matrix, out: Matrix ¬ NIL]
RETURNS [Matrix];
Transform in by the viewer's viewport; use out if non-NIL.
Use of this procedure is discouraged, as it will cause clipping errors.
MakeCameraMatrix: PUBLIC PROC [
worIncr:   Matrix ¬ NIL,
scale:    REAL ¬ 1.0,
move:    Triple ¬ [0.0, 0.0, 0.0],
rotate:    Triple ¬ [0.0, 0.0, 0.0],
eyeIncr:   Matrix ¬ NIL,
fieldOfView:  REAL ¬ 0.0,
swapYZ:   BOOL ¬ TRUE,
out:    Matrix ¬ NIL]
RETURNS [Matrix];
Return the world to view transformation matrix, using out if non-NIL. This transforms world
space coordinates into view space. The canonical transformation sequence used here is:
(world) WiSTRXEi (eye) P (view),
where S=scale, T=translate, R=RxRyRz=rotate about x, y, and z axes, X=swap y and z, P=perspective; the parenthesized coordinate systems are only labels.
World space is right-handed, generally having x=right, y=away ("into" the screen), z=up,
Eye and view spaces are left-handed with x=right, y=up, z=away.
swapYZ true if changing from right-handed world space to left-handed view space.
If fieldOfView # 0, view space is a perspective space; the other spaces are orthogonal.
WorldToViewFromVectors: PROC [
eyePoint:  Triple ¬ origin,
lookAt:  Triple ¬ zAxis,
upDirection: Triple ¬ yAxis,
fieldOfView: REAL ¬ 40.0,
out:   Matrix ¬ NIL]
RETURNS [Matrix];
Compute a camera matrix, using out if non-nil.
FromScaleMovesRots: PROC [scale: REAL, moves, rotates: Triple]
RETURNS [eyePoint, lookAt, upDirection: Triple];
Return the camera position and orientation given the standard moves and rotates in x, y, z.
FromEyeLookUp: PROC [eyePoint, lookAt, upDirection: Triple, fieldOfView: REAL]
RETURNS [moves, rotates: Triple];
Return the standard moves and rotates in x, y, z given the camera position and orientation.
END.