0 ratings0% found this document useful (0 votes)
12 viewsDocuments
There are some computer graphics documents attached here.
Uploaded by
Marshia MohanaCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0 ratings0% found this document useful (0 votes)
12 viewsDocuments
There are some computer graphics documents attached here.
Uploaded by
Marshia MohanaCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 34
Chapter Ten
Hidden Surfaces
Opaque objects that are closer to the eye and in the line of sight of other objects will block those
cies oF portions of those objects from view. In fact, some surfaces of these opaque objects
themselves are not visible because they are eclipsed by the objects’ visible parts. The surfaces that
ae blocked ot hidden from view must be “removed” in order to construct a realistic view of the
3D scene (see Fig, 1.3 where only three of the six faces of the cube are shown). The identification
and removal of these surfaces is called the hidden-surface problem. The solution involves the
determination of the closest visible surface along each projection line (Section 10.1).
There are many different hidden-surface algorithms. Each can be characterized as either an
image-space method, in which the pixel grid is used to guide the computational activities that
determine visibility at the pixel level (Sections 10.2, 10.5 and 10.6), or an object-space method, in
which surface visibility is determined using continuous models in the object space (or its
transformation) without involving pixel-based operations (Sections 10.3 and 10.4),
Notice that the hidden-surface problem has tis to the calculation of shadows. If we place a light
source, such as a bulb, at the viewpoint, all surfaces that are vi Je from the viewpoint are lit
directly by the light source and all surfaces that are hidden from the viewpoint are in the shadow of
some opaque objects blocking the light,
10.1 DEPTH COMPARISONS
We assume that all coordinates (x, Ys z) are described in the normalized viewing coordinate system
(Chapter 8). . :
Any hidden-surface algorithm must determine which edges and surfaces are visible either from
the center of projection for perspective projections or along the direction of projection for parallel
projections. Computer Graphics
256
The question of visibility reduces to this: given two points Py (1, Yi» 2) and Po (xp,
92, 2s dogs
cither point obscure the other? This is answered in wo steps:
jection line?
aunt ites srrat obscures the other. IF so, depth comparison tll Us which pons
front of the other
For an orthographic parallel projection onto the xy plane, P; and Py are on the same Projector ig
sy = sya; = Yo In this ease, depth comparison reduces to comparing 2, and zy
the normalized perspective to parallel transform preserves lines, planes, and 42>!
e)
y
ts! Hidden Surfaces a]
Parallel projection of
this configuration is
[ABCDEFG)
Perspective projection of
this configuration is.
> [ABCDEFG)
%
y ree
1
Fig. 10.2
IF our display device has display coordinates H x V, application of the scaling matrix
( H00O0
fs 0vo0o0
wvi=lo 9-10
ooo01
< 1 onto the region 0 $x H,
transforms the normalized display space 0 Sx <1,0SyS1,0Sz
lay transform DT,:
O 0 (the angle mea Nand
the z axis is less than 90°). The polygon is also classified as a back-face when C = 0, since in this
case it is parallel to the line of sight and its projection is either hidden or overlapped by the edge(s)
of some visible polygon(s).
Although this method identifies and removes back-faces quickly it does not handle polygons
that face the viewer but are hidden (partially or completely) behind other surfaces. It can be used
asa preprocessing step for other algorithms.
10.4 THE PAINTER’S ALGORITHM
Also called the depth sort or priority algorithm, the painter's algorithm processes polygons as if
they were being painted onto the view plane in the order of their distance from the viewer. More
distance polygons are painted first. Nearer polygons are painted on or over more distance polygons.
panially or totally obscuring them from view. The key to implementing this concept is to find a
priority ordering of the polygons in order to determine which polygons are to be painted (i.e. scan-
Converted) first.
Any attempt at a priority ordering based on depth sorting alone results in ambiguities that must
be resolved in order to correctly assign priorities. For example, when two polygons overlap, how
do we decide which one obscures the other? (See Fig. 10.4.)
y 4
@,
P
°, ie
u // Projection of Projection of polygon
ie / polygon PandQ
D \A— 4, B,C, and D @
— >
* ©
(a)
Fig. 10.4 Projection of Opaque Polygons 260 Computer Graphics
Assigning Priorities
We assign priorities to polygons by determining if a given
polygon P obscures other polygons. If the answer is no,
then P should be painted first. Hence the key test is to
determine whether polygon P does not obscure polygon Q.
The z extent of a polygon is the region between the planes
2 = Emin ANd Z = Zpay (Fig. 10.5). Here, Zin is the smallest of
the = coordinates of all the polygon’s vertices, and Zax i8
the largest.
Similar definitions hold for the x and y extents of a
polygon. The intersection of the x, y, and z extents is called r
the extent, or bounding box, of the polygon. Fig. 10.5
Testing Whether P Obscures Q
Polygon P does not obscure polygon Q if any one of the followin,
order, is true.
Test 0: The z extents of P and Q do not overlap and
Fig. 10.6.
ig tests, applied in sequential
Zo... of Q is smaller than zp of P. Refer to
e
oe x
Proletionota = Projection of P
Fig. 10.6 Fig. 10.7
Test I:
oe 1: The y extents of P and Q do not overlap. Refer to Fig. 10.7.
* x extents of P and Q do not overlap.
"All the vertices of P tie-on that side of the plane containing Q which is far
Viewpoint. Refer to Fig, 10.8,
Test
hest from Hidden Surfaces
261
Pplane
Fig. 10.8 Fig. 10.9
ret dz All he vertices of Q lie on that side of the plane containing \
viewpoint. Refer to Fig. 10.9. a ining P which is closest to the
rest 5: The projections of the polygons P and
checked by comparing each edge of one pol
search for intersections.
Q onto the view plane do not overlap. This is
lygon against each edge of the other polygon to
The Algorithm
1, Sort all polygons into a polygon list according tO Zax
polygon’s vertices). Starting from the end of the list, assign pri
order, as described in steps 2 and 3 (below).
2 Find all polygons Q (preceding P) in the polygon
z coordinate of each
olygon P, in
(the largest
orities for each pi
list whose z extents overlap that of P (test
3. For each Q, perform tests 1 through 5 until true.
(a) If every Q passes, scan-convert polygon P. .
(b) If false for some Q, swap P and Q on the IS: Ta Oe
tagged, use the plane containing polygon P to divide poly
and Q, [see Fig. 10.4(b)]. The polygon" ipping teciaues <6
‘ division. Remove Q from the list and pI
has already been
wapped. If Q
‘on Q into two polygons
lescribed in Chapter 5 can
race Qy and Qs on the Uist
be used to perform the
in sorted order. cing th
5 a assing, thus reducing the
Sometimes the polygons are subdivided int triangles before processing;
; n in step 3
COmputati,
Putational effort for polygon subdivis
10
5 SCAN-LINE ALGORITHM soe
A s0, 5. an x-scan loop
ant ; 7 sted loops. 4"
ean ieee algorithm consists essentially of two neste 262 Computer Graphics
y Scan
ect the polygons to be rendered with the scan plane y = g This
For cach y value, say, y = @ inters : er ne se Ys
and the resulting intersections are I -gments in this Plane
scan plane is parallel to the xz plane,
(see Fig. 10.10).
Display
screen
Peat, B, 22)
aplane
line segments
Fig. 10.10
x Scan
the x-scan line
1, For each x value, say, x = , intersect the line segments found above with te
hat lies 0”
x= B lying on the y-scan plane. This intersection results in a set of points tl
x-scan line.
2. Sort these points with respect to their z coordinates. The point (x,
z value is visible, and the color of the polygon containing this point is the color s°
pixel corresponding to this point.
st
yy 2) with the smal
pe at the
advantage of
In order to reduce the amount of calculation in each scan-line loop, we try to take pris?
1
relationships and dependencies, called coherences, between different elements that ¢
scene. Hidden Surfaces fa
3
types of Coherence
1. Scan-line coherence. If a pixel on
likely lie within the polygon.
. Edge coherence. If an edge of a polygon i i AN a
; aoe scan lines near the given ca sMerseets a given scan line, it will most likely
3, Area coherence. A small area of an image will most likel
4, Spatial coherence. Certain properties of an object can be determined by evansining th
extent of the object, that is, a geometric figure which circumscribes the given object. Usually
the extent is a rectangle or rectangular solid (also called a bounding box) sien
# sean Tine lies within a polygon, pixels near it will most
y lie within a single polygon.
Scan-line coherence and edge coherence are both used to advantage in scan-converting polygons
(Chapter 3).
Spatial coherence is often used as a pre-processing step. For example, when determining whether
polygons intersect, we can eliminate those polygons that don’t intersect by finding the rectangular
extent of each polygon and checking whether the extents intersect—a much simpler problem (see
Fig. 10.11). [Note: In Fig. 10.11 objects A and B do not intersect; however, objects A and C, and B
and C, do intersect. In pre-processing, corner points would be compared to determine whether
there is an intersection. For example, the edge of object A is at coordinate P; = (6, 4) and the edge
of object B is at coordinate P, = (7, 3).] Of course, even if the extents intersect, this does not
guarantee that the polygons intersect. See Fig. 10.12 and note that the extents of A’ and B’ overlap
even though the polygons do not.
oanes
iv
1234567891011
Fig. 10.11 Graphics
264 Computer Grap
calculations by making them
‘oherences can simplify :
r : is is illustrated
incremental, as opposed to absolute. Thi
in Solved Problem 10.13.
A Scan-line Algorithm
In the following algorithm, scan-line and edge coherence
are used to enhance the processing done in the y-sean
loop. Since the y-scan loop constructs alist of potent lly
visible line segments, instead of reconstructing this list
each time the y-scan line changes (absolute calculation),
we keep the list and update it according to how it has
hanged (incremental calculation). This processing is
facilitated by the use of what is called the edge list,
and its efficient construction and maintenance is at the
heart of the Algorithm (see Chapter 3, Section 3.
under “Scan-line Algorithm”).
The following data structures are created:
Fig. 10.12
1. The edge list contains all non-horizontal edges (horizontal edges are automatically displayed)
of the projections of the polygons in the scene. The edges are sorted by the edge’s smaller y
coordinate ()'pj,)- Each edge entry in the edge list also contains:
(a) The x coordinate of the end of the edge with the smaller y coordinate.
(b) The y coordinate of the edge’s other end (Ymax)+
(c) The increment Ax = I/m.
(@) A pointer indicating the polygon to which the edge belongs.
2. The polygon list for each polygon, contains
(a) The equation of the plane within which the polygon lies—used for depth determination,
ice., to find the z value at pixel (x, y).
(b) An IN/OUT flag, initialized to OUT (this flag is set depending on whether a given scan
line is in or out of the polygon).
(©) Color information for the polygon.
The algorithm proceeds as follows.
I. Initialization,
) Initialize each screen pixel to the background color,
(b) Set y to the smallest y,,in value in the edge list.
Repeat s .
I. yan lene oe (below) until no further processing can be performed.
x - Activate edges whose yin, is equal to y. Sort active edges in order of increasing
I. x
IH. x-scan loop. Process, from left to right,
(a) inves the IN/OUT flag of the pol
Sunt the number of active polygor
cach active edge as follows:
lygon in the polygon list which contains the edge
ns whose IN/OUT flag is set to IN. If this number iS (b)
10.6
The subs
Hidden Surfaces 265
1, only one polygon is v
are set to the color of the
polygon by the smallest z
z values are found from the equation of the pt
from this edge and up to the next edge are an
polygon becomes obscured by another before th
set the remaining pixels to the color of the obscuring polygon. If this number is 0, pixels
from this edge and up to the next
: h One are left unchanged.
When the last active edge is Processed, we then proceed as follows:
1, Remove those edges for which the value Of Ynax Equals the present scan line value y.
if no edges remain, the algorithm has finisher” .
2, For each remaining active edge, in order,
intersection with the next scan line y
3. Increment y to y + 1, the next scan li
SUBDIVISION ALGORITHM
‘sible. All pixel values
Polygon. If this number
value of each polygo
from this edge and up to the next edge
is greater than 1, determine the visible
n at the pixel under consideration. These
lane containing the polygon. The pixels
to the color of this polygon, unless the
le next edge is reached, in which case we
replace x by x + I/m. This is the edge
+ 1 (see Solved Problem 10.13).
ine, and repeat step II.
division algorithm is a recursive procedure based on a two-step strategy that first decides
which projected polygons overlap a given area A on the screen and are therefore potentially visible
in that area. Second, in each area these polygons are further tested to determine which ones will be
visible within this area and should therefore be displayed. If a visibility decision cannot be made,
this screen area, usually a rectangular window, is further subdiv
can be made, or until the screen area is a single pixel.
Starti
four sm:
ided either until a visibility decision
ing with the full screen as the initial area, the algorithm divides an area at each stage into
aller areas, thereby generating a quad tree (see Fig. 10.13).
Tale] [2]a[2
‘|? | '| 2) [etetstatstaare
1 2 ajetifalifatie
S\4 8/4) Blatarstarste
telat feti2
1)? ] "1? | (laptapatetata
3 4 tat fati feta
Efi ca Bal 3[4i3[4[slafala
Initial area First subdivision Second sutavon asa
(a) (o)
Fig, 10.13
7 veons P with res| a given screen
The processing exploits area coherence by classifying polygons P with respect oa given ser
tea A ji
the area
8) cont,
disjoint polygon—polygon that is disjoin
i iygon—polygon that completely contains
ing categories: (1) surrounding polygon—polygon te i
mo the een eae eiink polygon—polygon that intersects 7 area ed anes
Fae ners Completely contained within the are (Fig. 10.14(e)),
Fined Polygon— polyenes { from the area (Fig. 10.14(b)]. Computer Graphics
266 om)
Pintersects A
surrounds A
Ps ae te)
Pls disjoint from A
Pis contained in A
(d)
©
Fig. 10.14
The classification of the polygons within a picture is the main
computational expense of the algorithm and is analogous to the clipping
algorithms discussed in Chapter 5. With the use of one of these clipping
algorithms, a polygon in category 2 (intersecting polygon) can be clipped
into a contained polygon and a disjoint polygon (see Fig. 10.15).
Therefore, we could proceed as if category 2 were eliminated. A Pi
For a given screen area, we keep a potentially visible polygons list
(PYPL), those in categories 1, 2, and 3. (Disjoint polygons are clearly
Fig. 10.15
not visible.) Also, note that on subdivision of a screen area, surrounding
and disjoint polygons remain surrounding and disjoint polygons of the ;
newly formed areas. Therefore, only contained and intersecting polygons need to be reclassified.
Removing Polygons Hidden by a Surrounding Polygon
The key to efficient visibility computation lies in the fact that a polygon is not visible if iis bet!
fh runing Polygon. Therefore, it can be removed from the PVPL. To facilitate processing. in
list is sorted by zniq. the smallest z coordinate of the polygon within this area, In addition, for &#*
Surrounding polygon S, we also record its largest z coordinate, zy " 7
{1 for a polygon P on the list, zp. > zs. (for a surrounding polygon 5), then P is hidée® |
and thus is not visible. In addition, all other polygons after P on the list will also be hidden by
Wwe can remove these polygons from the PVPL. | Hidde
len Surfaces Th
gptivision Algorithm
1 ste the area 0 be the whole sercen
a PYPL with respect
Create Place at BA: Sorted on Zaig (the smallest z coordinate of the
* olygon within the area). Place the Polygons in thei
Pron DY 2 surrounding polygon and ie Aa categories. Remove polygons
4. Perform the visibility decision tes Heint polygons.
list is empty, set all pi
{a) If the Ply, set all pixels to the backs:
s Pixel ackground color.
(b) I hed (eneuoy eee the list and it is a classified as intersecting (category 2)
2 aoa) Scan-convert) the polygon, and color the remaining area
(o) If there is exactly one polygon on the list a is z
ihe colorof the surreunine polyeod. and it is a surrounding one, color the area with
@ i the ages enGeand (x, y), and neither a, b, nor ¢ applies, compute the z coordinate
x(a, y) at pixel (x, y) of all polygons on the PVPL. The pixel is then set to the color of
the polygon with the smallest z coordinate.
4, If none of the abov
area, go to step 2.
© cases has occurred, subdivide the screen area into fourths. For each
10.7 HIDDEN-LINE ELIMINATION
Although there are special-purpose hidden-line algorithms, each of the above algorithms can be
modified to eliminate hidden lines or edges. This is especially useful for wireframe polygonal
models where the polygons are unfilled. The idea is to use a color rule which fills all the polygons
with the background color—say, black—and the edges and lines a different color—say, white. The
use of a hidden-surface algorithm now becomes a hidden-line algorithm.
10.8 RENDERING OF MATHEMATICAL SURFACES
In plotting a mathematical surface described by an equation z= F(x, y), where pin SX $Xmax and
Vain SYS Yar WE could use any of the hidden-surface algorithms so far described. However,
these general algorithms are inefficient when compared
to specialized algorithms that take advantage of the
structure of this type of surface. |
The mathematical surface is rendered as a wire frame
model by drawing both the x-constant curves ’
F(const, y) and the y-constant curves 2 = pe ae
(see Fig. 10.16). Each such curve Is ren red 2 F
polyline, where the illusion of smoothness 1s fe ite
by using a fine resolution (i.e., short line segments
drawing the polyline (Chapter 3)- Computer Graphics
268 |
Choose an Che N pening ron ec ant Sg envevee
ding. F(x, yj). An x-constant polyline, say
PAR, Yeo
Similarly, the y = yy polyline has vertices
Qi Os vee Oyu Yd
Choosing a view plane and a center of projection or viewpoint C(a, b, c), we create a
Iw S Yenax
are zy = Xj, has vey
The corresponding z values are Z)j = ‘ Ttices
Pray, Yy)
i na Perspective
view of the surface onto this view plane by using the transformations developed in Chapter 7. So
point [x,y FQ) projects to a point (p, q) in view plane coordinates, By apply
an appropriate 2-dimensional viewing transformation (Chapter 5), we can suppose that p an Ae
within the horizontal and vertical plotting dimensions of the plotting device, say, H x V pixels ad
Perimeter Method for Rendering the Surface
Each plotted x and y constant polyline outlines a polygonal region on the Plotting screen (Fj
The algorithm is based on the following observations: (1) ordering—the x-
curves (i.e. polylines) are drawn in order starting with the one closest to the
(2) visibility—we draw only that part of the polyline that is outside the perimeter of all Previously
drawn regions (Fig. 10.18). One implementation of the visibility condition uses a min-max aray
of length H (that of the plotting device), which contains, at each horizontal Pixel position i, the
maximum (and/or minimum) vertical pixel value drawn thus far at i; that is AG) = (venieal
pixel values drawn so far at 4) (Fig. 10.19). Selection of the max results in a drawing of the top of
the surface. The min is used to render the bottom of the surface, and the max and min yields both
top and bottom.
i. 10.17,
and conan
Viewpoint and
10} 10)
u th 4
1 10
Fig. 10.17 Fig. 10.18
The Visibility Test
Suppose that (p', 9’) is the de
cither nt @'+ 4°) is the pixel that corresponds to the point (p, q). Then this pixel is visible”
where Ais the min-max array, 1 7 4® or 4" < A(p’) 269
Hidden Surfaces
@
1
1
1
1 10 t Li
0000000000 ASsazacaae® A 1565466544
Fig. 10.19 Updating min-max Array A (Using Maximum Values Only)
The visibility criteria for a line segment are: (1) the line segment is visible if both its endpoints
les (2) the line segment is invisible if both its endpoints are not visible; and (3) if only one
visible, the min-max array is tested to find the visible part of the line.
the x or y-constant polylines thus consists of testing for the visibility of each
the min-max array as necessary. Since a line segment will, in general. span
10.19), the computation of A(i) for
ment or by using Bresenham’s
are visibl
endpoint is
Drawing
segment and updat
several horizontal pixel positions (see segment FP; in Fig.
these intermediate pixels is found by using the slope of the line seg
method described in Chapter 3.
Wright Algorithm for Rendering Mathematical Surfaces
line
F(x, y) proceeds as follows.
The drawing of the surface <
1. To perform initialization, determine whether the viewpoint is closer to the x or the y axis.
Suppose that it is closer to the x axis. We next locate the x-constant curve that is closest to
the viewpoint at x = 1.
(a) Initialize the min-max array to some base value, say, zero.
(b) Start with the x-constant curve found above.
2. Repeat the following steps using the visibility test for drawing line segments and updating
the min-max array each time a line segment is drawn:
(a) Draw the x-constant polyline.
(b) Draw those parts of each y-constant polyline that lie between the previously drawn x-
constant polyline and the next one to be drawn.
(¢) Proceed, in the direction of increasing x, to the next x-constant polyline.
10.9 WARNOCK’S ALGORITHM
Wamoct" ;
eons S algorithm divides the screen into sample areas. If there are no faces of the object within
‘oe then it is left blank. If the nearest polygon completely covers it, then it can be filled in
the color of that polygon. If neither of these conditions holds, then the algorithm subdivides Computer Graphics
270
reas and considers each of them in turn. This process i,
smaller sample a i : h t
a ins only a single pixel or it falls into one of the ye
area size remal
Fig. 10.20).
the sample are:
repeated till the sample
conditions said above (see
Fig. 10.20 Subdivision of a Scene
10.10 WEILER-ATHERTON ALGORITHM
Warnock’s algorithm was forced to use image precision operation for termination because it could
clip a polygon only against a rectangular area. Weiler-Atherton algorithm subdivides the screen
area along polygon boundaries rather than along rectangular boundaries. This requires a powerful
clipping algorithm that can clip even a concave polygon against another. This involves the following
steps.
1. Sort the polygons according to the depth z.
2. Clip all the polygons based on the polygon closest to the viewer.
3. Remove the polygons behind that polygon nearest to the eye point.
4. If required, use a final sorting method with respect to the depth z and recursive subdivision
to remove any ambiguities,
10. :
0.1 Given points Pi(1, 2, 0), P3(3, 6, 20), and P,(2, 4, 6) and a viewpoint C(0, 0 =
letermine which points obscure the others when viewed from C. Hidden Surfaces
271
soining the viewpoint C(0, 0, -10) and point P,(1, 2, 0) is (Appendix 2
The ine Jone yet ye2t ~10 + 101
i is line, we see that x = 3 when 1 =
4 P,(3, 6, 20) lies on this that
‘0 dere gent z= 20. So P, lies on the projection line throu:
ait = 3.45 imine which point is in front
Yext we determine which poi
rarer att= 1, and P, occurs at f=
ath respect to C; that is, P; obscures P),
We now determine whether Ps (2, 4, 6) is on the line. Now x= 2 when = and then
4,and ¢ = 10. Thus P3 (2, 4, 6) is not on this projection line and so it nether obscures
y=4and ‘
nor is obscured by P, and P>.
3, and then
gh C and P,,
ith respect to C. Now C occurs on the line at
3. Thus comparing 1 values, P\ is in front of P,
12. Construct the perspective to parallel transform T,, which produces an object whose parallel
projection onto the xy plane yields the same image as the Perspective projection of the
original object onto the normalized view plane z = ¢//(c! +b) (Chapter 8, Solved Problem
8.6) with respect to the origin as the center of projection.
The perspective projection onto the plane z=
(Chapter 7, Solved Problem 7.4):
cic] + b) with respect to the origin is
x 0 0 0
5 \0 x 0
Realy 5 z 0
0201 dos0
where 2, = c{(e! + b). The perspective projection onto the view plane of a point P(x,
is the point
Define the perspective to parallel transform
T, to be
2 0
0 0 1
0 0 0
(where z
- a 5 Ived
Zr is the location of the normalized front clipping plane; see Chapter 8, Solv
Problem 8.6). Computer Graphics
272
rm T;, to the point PC, ¥, 2), we get the
Now, applying the perspective to parallel transfor
point
7)
The parallel projection of Q’ onto the xy plane produces the point
of Ato)
So Q’ and P’ produce the same projective image. Furthermore, 7, transforms the Normalizeg
Y= UY=-Z,2= Zand
cf Z= 1 to th
-Zy. 2 = 0, and z= 1, °
perspective view volume bounded by x = z=
rectangular volume bounded by x = z,
3. Show that the normalized perspective to parallel transform NT, preserves the relationships
of the original perspective transformation while transforming the normalized perspective
S
view volume into the unit cube.
From Solved Problem 10.2, the perspective to parallel transform 7, transforms a point
P(x, ¥. 5) to a point
z a4)
The image under parallel projection of this point onto the xy plane is
of
The factor z, can be set equal to 1 without changing the relation between points Q and Q’.
x YSU Tis then
a)
The matrix that transforms P(x, y, z) to the point of
i 0
-%
jo the
{n addition, this matrix transforms the normalized perspective view votnme
\220,and2=1
fectanguar view volume bounded by x = 1, x=-1, y= 1 = A the
© next translate this view volume so that the corner point (-1, -1, 0) transla
origi "
gin. The translation matrix that does this is Hidden Surfaces
273
1000
0100
oo10
1101
‘The new region is a volume bounded by x = 0, x=2, y=0,y=2,2=0,and 2<1,
Finally, we scale in the x and y direction by a factor 1 so that the final view volume is
= 0, and z= 1° The scaling matrix is
the unit cube: x= 0, x= 1, y=0,
°
00
Sirana =
10
o1
eo oN
cont
The final normalized perspective to parallel transform is
1
2 0 0 0
1
Oo 0 0
2
NT, =F, Taro Sinina=|L 1 1
2.2 Inzy
00 —~o
my
10.4 Why are hidden-surface algorithms needed?
Hidden-surface algorithms are needed to determine which objects and surface will obscure
those objects and surfaces that in back of them, thus rendering a more realistic image.
5) obscures
10.5 What two steps are required to determine whether any given point P(x.
another point P3(x>, y3, 23)? (See Fig. 10.1.)
It must be determined (1) whether the two points lie on the same p'
(2) if they do, which point is in front of the other.
rojection line and
10.6 Why is it easier to locate hidden surfaces when parallel projection is used?
There are no vanishing points in parallel projection, As a result, any point P (a, 6, 2) will
other point having the same x and y coordinates (a, b).
line on the same projector as any £00
be compared to determine which point is closest to the
Thus only the z component must
viewer. Computer Graphics
274
10.7 How does the Z-butfer algorithm determine which surfaces are hidden?
The Z-buffer algorithm sets dimensional array which is like the frame bute,
x the Z-buffer stores the depth value at each pixel rather than the color, which
stored in the frame buffer. By setting the initial values of the Z-buffer to some large
Humber, usually the distance of back clipping plane, the problem of determining whe,
surfaces are closer is reduced to simply comparing the present depth values stored in the
Zbuffer at pixel (x, y) with the newly calculated depth value at pixel (x, y). If this ney
value is less than the present Z-buffer value (i.c. closer along the line of sight), this vaty
replaces the present value and the pixel color is changed to the color of the new surface. °
p a two-
howev
10.8 Using a 2 x 2 pixel display, show how the Z-buffer algorithm would determine the colo
of each pixel for the given objects A and B in Fig. 10.21.
Display space
Display
screen
u 1 2 x
Pixe! coordinates
Fig. 10.21
‘The display space
Play space for the 2 x 2 pixel display is the region 0 $x $2, 0S
1 9) and Ax?
line A is o .
is On the displa wean tn’ f, ce
isplay sereen in front of square B. B is the square with display
coordinates B (0 0! nat?
er (2.0.5), ,(2.2,1). ana a,(0,2, 1) Te asve
2
Of A (after projecti
T projection and scan conv ne
$1. In Fig. 10.21, 4 ii ine wi
+ Ais the line with display space coordinates A, (3
2.
ersion) would appear on a2 x 2 pixel displ Hidden Surfaces
1[y]a
ofaly
oT
where @ is the color of A and y is the background color. We have used
Section. 3.1) that a point (x, y) sean-converts (o the pixel [Floor (x). Flo
sistency that 2 = 1.99...) The displayed image of B is
the fact (Chapter 3,
r(y)]. (We assume
1 b |b}
0 bib
0 1
where b is the color of B. We apply the Z-buffer algorithm to the piture composed of
objects A and B as follows: :
1, Perform initialization. The Z-butfer is set equal to the depth of the back clipping plane
1, and the frame buffer is initialized to a background color y. -
lfyTy ifiti
Frame buffer = 2 Zee
oly Ty so, Laoha
oT 0 1
2. Apply the algorithm to object A.
(a) The present Z-buffer value at pixel (0, 0) is that of the back clipping plane. ie. Zp
(0, 0) = 1. The depth value of A at pixel (0, 0) is z = 0. Then Z,,,(0. 0) is changed
to 0 and pixel (0, 0) has the color of A.
1 py 1 fafa
Fra 3 aa Zoot =
vame butfer= | FEL 2S Aue FG
01 00
(b) Object A is not seen from pixel (1, 0), so the Z-butfer value is unchanged.
1 fy Ty 1 fifi
rame=§ [211 aay ts
01 ot
(©) Object A is not seen from pixel (0, 1), so the Z-buffer value is unchanged:
yyy Zant tj
Frame=y Foo) Zu = 9 Tol
Ot one cent Z-butter
n the pres
(€) The depth value of A at pixel (1, 1) is 0. Since this is less
Value of 1, pixel (1, 1) takes the color of A.
I
Frame =
rame = 5 76 Computer Graphics
3. Apply the algorithm to B.
i = 0. So the color of pi
(a) The depth value of B at pixel (0, 0) is 4, and Ziyi(0, 0) = 0. So the color of pixey
(0, 0) is unchanged
Tai 1 1] 0
1 [ya =
Frame = 9 fa fy Sout ololi
041 oot
a py F
(b) The depth value of B at pixel (1, 0) is. The present Z-buffer value is 1. So the
Z-buffer value is set to j and pixel (1, 0) takes the color of B.
l fy ]a 1etly
Frame = Zou
Olalb 0 10
0 1 0
Hlri-lo
(©) The depth value of B at pixel (0, 1) is 1. The present Z-but
color at pixel (0, 1) is set t
fer value is 1, so the
fo that of B, and the Z-buffer is updated.
fo pa 1Jt to
Frame = ae 2 ;
0 b 0 lo
“ 2
01 01
(4) The depth value of B at pixel (1, 1) is 7 - The present Z-buffer value is 0. So the
color at pixel (1, 1) remains unchanged.
t}ola 1 |i to
Frame = Zu
olelo o|t
rn] oT
The final form of the Z-buffer indicates th
‘at line A lines in front of B.
10.9 What is the max
i r 5 z-bulfer
imum number of objects that ea eeenty sing the Z-bu!
algsrithm? OF objects that can be presented by using
; . itrary
, The tota number of objects that can be handled by the Z-buffer algorithm is abit
Ssause each object is processed one at time, a
to How does the basic sean-line method determine which surfaces are hidden?
10. ?
‘The basic scan-line method looks one at a time at each of the horizoy
the display area. For example, at the horizontal pixel line y =
(consisting of all scan-converted polygons) is searched to
horizontal (y) pixel values equal toa
Next, the algorithm looks at each individual pixel in the a row. At pixel (a B), the
depth values (< values) of each polygon found above are compared to find the polygon
having the smallest z value at this pixel. The color of pixel (c, B) is then set tothe vole
the corresponding polygon at this pixel.
nal lines of pixels in
@ the graphics data structure
find all polygons with any
he four pixel display and the graphics objects A and B from Solved Problem 10.8,
show how the basic scan-line method would display these objects,
10.11
First we initialize the display to the color y of the back clipping plane (located at z = 1)
lfyyy]
Frame = 5
01
1, Set y = 0. The scan-converted representations of A and B contain pixels on the y = 0
scan line.
(a) Set x = 0. Comparing the z values of A and B at pixel (0, 0), we find that the
smaller z value is 0, which belongs to A. Thus A is seen from pixel (0, 0): that is,
Pixel (0, 0) is set to the color of A.
T}yly
olely
0.1
() Set x= 1. Since A is not seen from pixel (1, 0) while B is seen, the color of pixel
(1, 0) is set to that of B.
Ifyfy
Olaly
01 _
> Set y= 1. The scan-converted representations of A and B contain pixels on the»
Scan line,
(®) Set.x=0, Since A is not seen at pixel (0, 1) while B is seen, pixel (0, 1) is set the
color of B,
I}yjy
ofalb
xel (1, 1)
(b) 6 01
y Stx= 1, Both A and B are “seen” from pixe
'80, that of Bis 1
2
1 (1, 1). The depth of A at pi
Thus A is visible at pixel (1, 1). 278
10.12
10.13
Computer Graphics
ify
ofa|s
01
This trame buffer represents the final image displayed.
iona 9
How does edge coherence help to reduce computational effort?
Tt is based on the assumption that, if an edge or line intersects a given scan line, it wil}
most likely intersect the scan lines next to it. Thus if the pixels that intersect the edge are to
be found, instead of intersecting each scan line with the edge, it is sufficient to locate only
one intersection pixel and then find the others using the slope m of the edge (see Solved
Problem 10.13).
Show how the calculation of the intersection of an edge with a scan line can be made
incremental as opposed to absolute.
The absolute calculation requires that we find the x intersection value of the edge (e.g.
with equation y = mx + b) with the scan line y = o for each a.
The incremental solution to the problem is based on the following observations. Suppose
that x, is the x intersection of the edge with the scan line @. Then the intersection Xqa OF
the edge with the next scan line y= @ + 1 can be found as illustrated in Fig. 10.22. From
Fig. 10.22, where m is the slope of the edge,
Ay (a+l)-a@
Ax
=m
Nast ~ Xa
Edge y=mx+b
Scan tine y= a+ 1
1 Xawe+d
Scan line y= ——__f/ i
(av a) rime
m
‘Fig. 10.22
Solving for x,.,), we obta
Nas) = Xqt —
; m
Thus the calculation of the next intersecti
i i fro
pen ion point is inci ; is, it is found
the previous intersection point by E remental; that is, it
adding 1/m to it, meee ae
How does area coherence reduce computational effort?
low S anes
10.14
Area coherence is based on the assumption that a small enough region of en
inst likely tie within a single polygon. This reduces, as in the subdivaiy cen
computational effort in dealing with all those polygons that are potentially vise
given screen area (region of pixels).
10.18 How is spatial coherence determined?
Spatial coherence is determined by examining the extent of an object. Th rectangular
extent (bounding box) of an object is determined by finding the minivan and maximum
and : | Surrounding Hidden Surfaces
285,
4, visibility decision. Applying tests (a) through (c), we now apply test (d) since the
region is pixel size. The < coordinate of A at pixel (0, 0) is 0, and that of Bis +. Thug
‘Ais visible at pixel (0, 0):
1
Frame buffer = ole
ool
Region 2: pixel (0, 1).
2, Forming the PVPL. Note that A is disjoint from this region:
Polygon | min Category
PVPL =
e x Surrounding
»
Visibility decision. From test (c), there is only one
polygon and it is surrounding, so we
color the pixel (0, 1) that of B:
1
o fats
01
Region 3: pixel (0, 1).
2. Forming the PVPL. Since A is disjoint from this region, we have
Polygon | Znin | Category
PVPL =
B 3 Surrounding
3.
Visibility decision. From test (c), there is only one polygon B and it is surrounding. So
region 3 is colored b:
1[o
ofalb
01
Region 4: pixel (1, 1),
2. Forming the PPL,
Polygon Zain Category
PVPL = A 0 Intersecting
1 :
+ Surrounding
B 2
e
Visibility decision, Having applied tests (a) through (c), we now apply test (d). The z
Coordinate of A at pixel (1, 1) is less than that of B. Thus pixel (1, 1) is set to the color
of A: 286 Computer Graphics
l| bia
Olalob
ool
This is the final image in the frame buffer.
10.24 How can we use the special structure of a convex polyhedron to identify its hidden faces
for a general parallel or perspective projection?
Suppose that on each face of the polyhedron
there is an outward-pointing normal vector N,
attached ata point P of the face (Fig. 10.24),
For h face of the polyhedron, let the line-
of-sight vector L be the vector pointing from
the face to the viewer. For a parallel projection,
this is the direction of projection from the
object to the projection plane. For a perspective
projection, it is the vector PC from the normal
vector attached at point P to the viewpoint at
point C (Fig. 10.24).
is visible if the angle @ made by the
line-of-sight vector L and the normal vector N
is less than 90°. It is hidden if this angle is
larger than or equal to 90°. If 0° < @ < 90°,
then 0 < cos @< 1; if 90° < @< 180°, then -1
1), the first polygon written will determine
the color of the pixel (see Problem 10.7).
10.3 A priority flag could be assigned to break
the tie resulting from the use of Z-buffer
algorithm.
10.4 A system that distinguishes 274 depth
values would require three bytes of
memory to represent each z value. Thus
3 x 1024 x 768 = 2304 KB of memory
would be needed.
10.:
in
The scan-line method can take advantage
of (a) scan-line coherence, (b) edge cohe-
rence, (c) area coherence, and (d) spatial
coherence.
10.6 Scan-line coherence is based on the
assumption that if a pixel belongs to the
scan-converted image of an object, the
pixels next to it will (most likely) also
belong to this object.
10.7 Since this figure is a polygon, we need
only find the maximum and minimum
coordinate values of the vertices A, 3,
and C, Then
Xmin Fact 2
Ymin Ymax = 2
Zin Zmax = 2
The bounding box is shown in Fig. 10.27.
10.8 Horizontal line segments (Yin )
lic on only one scan line: they are auto-
matically displayed when non-horizontal
edges are used in the scan-conversion
proces
We search the z coordinates of the vertices
of the polygon for the largest value, zp,,.
The depth of the polygon is then set to
10.
&
10.10. Area coherence is exploited by classify-
ing polygons with respect to a given screen
area as either a surrounding polygon, an
intersecting polygon, a contained poly-
gon, or a disjoint polygon. The key fact
is that a polygon is not visible if it is in
back of a surrounding polygon.
10.11 When using a hidden-surface algorithm
to eliminate hidden lines, we set the fill
color of the polygons determined by the
lines to the background color.
(ETP
You might also like
- Hidden Surface Removal - Computer GraphicsNo ratings yetHidden Surface Removal - Computer Graphics10 pages
- Visible Surface Determination: CMSC 161: Interactive Computer GraphicsNo ratings yetVisible Surface Determination: CMSC 161: Interactive Computer Graphics34 pages
- Hidden Surface, Visible Surface & Hidden-Surface Removal ProblemNo ratings yetHidden Surface, Visible Surface & Hidden-Surface Removal Problem2 pages
- CG Module 6 - Visible Surface Detection Algorithm & AnimationNo ratings yetCG Module 6 - Visible Surface Detection Algorithm & Animation25 pages
- Computer Graphics: BY N.Sathish Kumar AP CSENo ratings yetComputer Graphics: BY N.Sathish Kumar AP CSE24 pages
- Visible-Surface Detection Methods: CS3162 Introduction To Computer GraphicsNo ratings yetVisible-Surface Detection Methods: CS3162 Introduction To Computer Graphics10 pages
- Computer Graphics Computer Graphics: Visible Surface Detection Methods Visible Surface Detection MethodsNo ratings yetComputer Graphics Computer Graphics: Visible Surface Detection Methods Visible Surface Detection Methods38 pages
- UNIT - 3 3D Clipping & Visible Surface Detection MethodNo ratings yetUNIT - 3 3D Clipping & Visible Surface Detection Method7 pages
- VisibVisibility and Hidden Surface RemovalilityNo ratings yetVisibVisibility and Hidden Surface Removalility75 pages
- Rendering Pipeline: Viewing: Geometry Processing Rendering Pixel ProcessingNo ratings yetRendering Pipeline: Viewing: Geometry Processing Rendering Pixel Processing16 pages
- Unit-4 Visible-Surface-Detection - sp-1No ratings yetUnit-4 Visible-Surface-Detection - sp-112 pages
- +Lgghq#6Xuidfh# (Olplqdwlrq: (Visible Surface Determination)No ratings yet+Lgghq#6Xuidfh# (Olplqdwlrq: (Visible Surface Determination)9 pages
- Visible Surface Detection Algorithms-Cg Unit 7100% (1)Visible Surface Detection Algorithms-Cg Unit 712 pages
- Visible Surface Detection (V.S.D) : (Chapt. 15 in FVD, Chapt. 13 in Hearn & Baker)No ratings yetVisible Surface Detection (V.S.D) : (Chapt. 15 in FVD, Chapt. 13 in Hearn & Baker)29 pages
