i
i
i
i
i
i
i
i
522 11. Non-Photorealistic Rendering
part of the silhouette, the quadrilateral’s points are moved so that it is no
longer degenerate (i.e., is made visible). This results in a thin quadrilateral
“fin,” representing the edge, being drawn. This technique is based on the
same idea as the vertex shader for shadow volume creation, described on
page 347. Boundary edges, which have only one neighboring triangle, can
also be handled by passing in a second normal that is the negation of this
triangle’s normal. In this way, the boundary edge will always be flagged
as one to be rendered. The main drawbacks to this technique are a large
increase in the number of polygons sent to through the pipeline, and that
it does not perform well if the mesh undergoes nonlinear transforms [1382].
McGuire and Hughes [844] present work to provide higher-quality fin lines
with endcaps.
If the geometry shader is a part of the pipeline, these additional fin
polygons do not need to be generated on the CPU and stored in a mesh.
The geometry shader itself can generate the fin quadrilaterals as needed.
Other silhouette finding methods exist. For example, Gooch et al. [423]
use Gauss maps for determining silhouette edges. In the last part of Sec-
tion 14.2.1, hierarchical methods for quickly categorizing sets of polygons
as front or back facing are discussed. See Hertzman’s article [546] or either
NPR book [425, 1226] for more on this subject.
11.2.5 Hybrid Silhouetting
Northrup and Markosian [940] use a silhouette rendering approach that
has both image and geometric elements. Their method first finds a list of
silhouette edges. They then render all the object’s triangles and silhouette
edges, assigning each a different ID number (i.e., giving each a unique
color). This ID buffer is read back and the visible silhouette edges are
determined from it. These visible segments are then checked for overlaps
and linked together to form smooth stroke paths. Stylized strokes are then
rendered along these reconstructed paths. The strokes themselves can be
stylized in many different ways, including effects of taper, flare, wiggle,
and fading, as well as depth and distance cues. An example is shown in
Figure 11.12.
Kalnins et al. [621] use this method in their work, which attacks an
important area of research in NPR: temporal coherence. Obtaining a sil-
houette is, in one respect, just the beginning. As the object and viewer
move, the silhouette edge changes. With stroke extraction techniques some
coherence is available by tracking the separate silhouette loops. However,
when two loops merge, corrective measures need to be taken or a noticeable
jump from one frame to the next will be visible. A pixel search and “vote”
algorithm is used to attempt to maintain silhouette coherence from frame
to frame.
i
i
i
i
i
i
i
i
11.3. Other Styles 523
Figure 11.12. An image produced using Northrup and Markosian’s hybrid technique,
whereby silhouette edges are found, built into chains, and rendered as strokes. (Image
courtesy of Lee Markosian.)
11.3 Other Styles
While toon rendering is a popular style to attempt to simulate, there is an
infinite variety of other styles. NPR effects can range from modifying real-
istic textures [670, 720, 727] to having the algorithm procedurally generate
geometric ornamentation from frame to frame [623, 820]. In this section,
we briefly survey techniques relevant to real-time rendering.
In addition to toon rendering, Lake et al. [713] discuss using the diffuse
shading term to select which texture is used on a surface. As the diffuse
term gets darker, a texture with a darker impression is used. The tex-
ture is applied with screen-space coordinates to give a hand-drawn look.
A paper texture is also applied in screen space to all surfaces to further
enhance the sketched look. As objects are animated, they “swim” through
the texture, since the texture is applied in screen space. It could be ap-
plied in world space for a different effect. See Figure 11.13. Lander [722]
discusses doing this process with multitexturing. One problem with this
type of algorithm is the shower door effect, where the objects look like they
are viewed through patterned glass during animation. The problem arises
from the textures being accessed by pixel location instead of by surface
coordinates—objects then look somewhat detached from these textures,
since indeed that is the case.
i
i
i
i
i
i
i
i
524 11. Non-Photorealistic Rendering
Figure 11.13. An image generated by using a palette of textures, a paper texture, and
silhouette edge rendering. (Reprinted by permission of Adam Lake and Carl Marshall,
Intel Corporation, copyright Intel Corporation 2002.)
One solution is obvious: Use texture coordinates on the surface. The
challenge is that stroke-based textures need to maintain a relatively uni-
form stroke thickness and density to look convincing. If the texture is
magnified, the strokes appear too thick; if it is minified, the strokes are
either blurred away or are thin and noisy (depending on whether mipmap-
ping is used). Praun et al. [1031] present a real-time method of generating
stroke-textured mipmaps and applying these to surfaces in a smooth fash-
ion. Doing so maintains the stroke density on the screen as the object’s
distance changes. The first step is to form the textures to be used, called
tonal art maps (TAMs). This is done by drawing strokes into the mipmap
levels.
2
See Figure 11.14. Care must be taken to avoid having strokes
clump together. With these textures in place, the model is rendered by
interpolating between the tones needed at each vertex. Applying this tech-
nique to surfaces with a lapped texture parameterization [1030] results in
images with a hand-drawn feel. See Figure 11.15.
Card and Mitchell [155] give an efficient implementation of this tech-
nique by using pixel shaders. Instead of interpolating the vertex weights,
they compute the diffuse term and interpolate this per pixel. This is then
used as a texture coordinate into two one-dimensional maps, which yields
per-pixel TAM weights. This gives better results when shading large poly-
2
Klein et al. [670] use a related idea in their “art maps” to maintain stroke size for
NPR textures.
i
i
i
i
i
i
i
i
11.3. Other Styles 525
Figure 11.14. Tonal art maps (TAMs). Strokes are drawn into the mipmap levels. Each
mipmap level contains all the strokes from the textures to the left and above it. In
this way, interpolation between mip levels and adjoining textures is smooth. (Images
courtesy of Emil Praun, Princeton University.)
Figure 11.15. Two models rendered using tonal art maps (TAMs). The swatches show the
lapped texture pattern used to render each. (Images courtesy of Emil Praun, Princeton
University.)
gons. Webb et al. [1335] present two extensions to TAMs that give better
results, one using a volume texture, which allows the use of color, the other
using a thresholding scheme, which improves antialiasing. Nuebel [943]
gives a related method of performing charcoal rendering. He uses a noise
texture that also goes from dark to light along one axis. The intensity
value accesses the texture along this axis. Lee et al. [747] use TAMs and a
number of other interesting techniques to generate impressive images that
appear drawn by pencil.
With regard to strokes, many other operations are possible than those
already discussed. To give a sketched effect, edges can be jittered [215, 726,
i
i
i
i
i
i
i
i
526 11. Non-Photorealistic Rendering
Figure 11.16. Two different graftal styles render the Stanford bunny. (Images courtesy
of Bruce Gooch and Matt Kaplan, University of Utah.)
747] or extended beyond their original locations, as seen in the upper right
and lower middle images in Figure 11.1 on page 507.
Girshick et al. [403] discuss rendering strokes along the principal curve
direction lines on a surface. That is, from any given point on a surface,
there is a first principal direction tangent vector that points in the direction
of maximum curvature. The second principal direction is the tangent vector
perpendicular to this first vector and gives the direction in which the surface
is least curved. These direction lines are important in the perception of a
curved surface. They also have the advantage of needing to be generated
only once for static models, since such strokes are independent of lighting
and shading.
Mohr and Gleicher [888] intercept OpenGL calls and perform NPR ef-
fects upon the low-level primitives, creating a variety of drawing styles. By
making a system that replaces OpenGL, existing applications can instantly
be given a different look.
The idea of graftals [623, 820] is that geometry or decal textures can be
added as needed to a surface to produce a particular effect. They can be
controlled by the level of detail needed, by the surface’s orientation to the
eye, or by other factors. These can also be used to simulate pen or brush
strokes. An example is shown in Figure 11.16. Geometric graftals are a
form of procedural modeling [295].
This section has only barely touched on a few of the directions NPR
research has taken. See the “Further Reading and Resources” section at
the end of this chapter for where to go for more information. To conclude
this chapter, we will turn our attention to the basic line.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.