03 2D Viewing (Engl)

Computergraphik 1 Textblatt engl-03 Vs.
10
Werner Purgathofer, TU Wien
2D-Viewing
2D
Viewing Pipeline
The term Viewing Pipeline describes a series of transformations, which are passed by geometry data to end
up as image data being displayed on a device. The 2D viewing pipeline describes this process for 2D data:
objectcoord.
construction
of objects
and scenes
worldcoord.
definition of
mapping region
+ orientation
viewing- projection onto

unity image
coord.
region
norm.
device- transformation
coord. to specific
device
device
coord.
The coordinates in which individual objects (models) are created are called model (or object) coordinates.
When several objects are assembled into a scene, they are described by world coordinates.
After transformation into the coordinate system of the camera (viewer) they become viewing coordinates.
Their projection onto a common plane (window) yields device-independent normalized coordinates.
Finally, after mapping those normalized coordinates to a specific device, we get device coordinates.
Window-Viewport
Transformation
A window-viewport transformation describes

the mapping of a (rectangular) window in one
coordinate system into another (rectangular)
window in another coordinate system. This
transformation is defined by the section of the
original image that is transformed (clipping
window), the location of the resulting window
(viewport), and how the window is translated,
scaled or rotated.
The following derivation shows how easy this transformation generally is (i.e. without rotation):
The transformation is linear in x and y, and (xwmin/ywmin) (xvmin/yvmin), (xwmax/ywmax) (xvmax/yvmax) .
For a given point (xw/yw) which is transformed to (xv,yv), we get:
=> xv = xvmin + (xvmax -xvmin)
xw = xwmin + (xwmax-xwmin) where 0<<1
Calculating from the first equation and substituting it into the second yields:
xv = xvmin + (xvmax xvmin)(xw xwmin)/(xwmax xwmin) = xw(xvmax xvmin)/(xwmax xwmin) + tx
where tx is constant for all points, as is the factor sx = (xvmax xvmin)/(xwmax xwmin).
Processing y analogically, we get an ordinary linear transformation: xv = sxxw + tx, yv = syyw + ty,
The chapter Transformations describes, how such transformations can be notated even simpler.
Line
Clipping
Clipping is the method of cutting away parts

of a picture that lie outside the displaying
window (see also picture above). The earlier
clipping is done within the viewing pipeline,
the more unnecessary transformations of
parts which are invisible anyway can be
avoided:
5
- in world coordinates, which means analytical calculation as early as possible,

- during raster conversion, i.e. during the algorithm, which transforms a graphic primitive to points,
- per pixel, i.e. after each calculation, just before drawing a pixel.
Since clipping is a very common operation it has to be performed simply and quickly.
Clipping lines: Cohen-Sutherland-Method
Generally, line clipping algorithms benefit from the fact, that in a rectangular window each line can have at
most one visible part. Furthermore, they should exploit the basic principles of efficiency, like early
elimination of simple and common cases and avoiding needless expensive operations (e.g. intersection
calculations). Simple line clipping could look like this:
for endpoints (x0,y0), (xend,yend)
intersect parametric representation
x = x0 + u*(xend - x0)
y = y0 + u*(yend - y0)
with window borders:
intersection 0 < u < 1
The Cohen-Sutherland algorithm first classifies the endpoints of a line
with respect to their relative location to the clipping window: above,
below, left, right, and codes this information in 4 bits. Now the following
verification can be performed quickly with the codes of the two endpoints:
1. OR of the codes = 0000 line completely visible
2. AND of the codes 0000 line completely invisible
3. otherwise, intersect the line with the relevant edge of the window and
replace the cut away point by the intersection point. GOTO 1.
Intersection calculations:
with vertical window edges:
with horizontal window edges:
y = y0 + m(xwmin x0),
x = x0 + (ywmin y0)/m,
y = y0 + m(xwmax x0)
x = x0 + (ywmax y0)/m
Endpoints which lie exactly on an edge of the window have to be interpreted as inside, of course. Then the
algorithm needs 4 iterations at most. As we can see, intersection calculations are performed only if it is
really necessary.
There are similar methods for clipping circles, however they have to consider that circles can be divided into
more than one part when being clipped.
Polygon
Clipping
When clipping a polygon we have to make sure, that the result is one
single polygon again, even if the clipping procedure cuts the original
polygon into more than one piece. The upper figure shows a polygon
clipped by a line clipping algorithm. Afterwards it is not decidable
what is inside and what is outside the polygon. The lower figure
shows the result of a correct polygon clipping procedure. The
polygon is divided into several pieces, each of which can be filled
correctly.
Clipping polygons: Sutherland-Hodgman-Method
The basic idea of this method yields from the fact that clipping a
polygon at only one edge doesnt create many complications.
Therefore, the polygon is sequentially clipped against each of the 4
window edges, and the result of each step is taken as input for the next one:
6
There are 4 different cases how an edge (V1, V2) can be located relative to an edge of the window. Thus,
each step of the sequential procedure yields one of the following results:
1. outin (output V1, V2)
2. inin (output V2)
3. inout (output V1)
4. outout (no output)
Thus the algorithm for one edge works like this:

The polygons vertices are processed sequentially. For
each polygon edge we verify which one of the 4
mentioned cases it fits, and then create an according
entry in the result list. After all points are processed, the
result list contains the vertices of the polygon, which is
already clipped on the current windows edge. Thus it is
a valid polygon again and can be used as input for the
next clipping operation against the next window edge.
These three intermediate results can be avoided by
calling the procedure for the 4 window edges
recursively, and so using each result point instantly as input point for the next clipping operation.
Alternatively, we can construct a pipeline through these 4 operations, which has a polygon as final output
the polygon which is correctly clipped against the 4 window edges.
When cutting a polygon to pieces, this procedure creates connecting edges along the clipping windows
borders. In such cases, a final verification respectively post-processing step may be necessary.
Text
Clipping
At first glance clipping of text seems to be trivial, however, one little point has to be kept in mind:
Depending on the way the letters are created it can happen, that either only text is displayed that is fully
readable (i.e. all letters lie completely inside the window), or that text is clipped letter by letter (i.e. all
letters disappear that do not lie completely inside the window), or that text is clipped correctly (i.e. also half
letters can be created).

03 2D Viewing (Engl)

Uploaded by

Copyright:

Available Formats

03 2D Viewing (Engl)

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

03 2D Viewing (Engl)

Uploaded by

Copyright:

Available Formats

Computergraphik 1 Textblatt engl-03 Vs.

viewing- projection onto

A window-viewport transformation describes

Clipping is the method of cutting away parts

- in world coordinates, which means analytical calculation as early as possible,

1. outin (output V1, V2)

2. inin (output V2)

3. inout (output V1)

4. outout (no output)

Thus the algorithm for one edge works like this:

You might also like