Graphics Reference
In-Depth Information
Figure 1.7: The standard teapot, created by Martin Newell, a model that's been used thou-
sands of times in graphics.
bulb. The surface area of a sphere with radius 10 cm is about 1200 cm
2
; our lens
therefore receives about 1/10,000 of the light emitted by the bulb, or 6.6
10
14
photons per second. If we take the picture with a 0.01
sec
exposure and we have
an approximately 1-million-pixel sensor, then each sensor pixel receives about
10
6
photons. Photographing a dark piece of carpet in our imaginary office above
might result in each sensor pixel receiving only 100 photons.
×
Computer games are some of the most demanding applications at present; to make
objects appear on the user's screen, these applications send polygons to a graphics
processor. These polygons have various attributes (like color, texture, and trans-
parency) and are displayed with various technologies (antialiasing, smooth shad-
ing, and others, all of which we'll discuss in detail later). For a polygon to be
displayed, certain pixels must be colored in certain ways. Thus,
polygon rate
(the
number of polygons displayed per second) and
fill rate
(the number of pixels
colored per second) are both used to measure performance. The numbers are con-
stantly changing, and there's a huge difference between a textured, antialiased,
transparent polygon covering 500 pixels and a flat-shaded 10-pixel triangle, so
comparisons are difficult. But complex scenes for interactive display can easily
contain 1 million polygons, of which maybe 100,000 are visible (the others being
hidden by things in front of them or outside the field of view), each occupying
perhaps 10 pixels on average. In many cases, a single polygon occupies less than
a single pixel. This happens in part because complex shapes are often modeled
with polygonal meshes (see Figure 1.7). For high-quality, noninteractive, special-
effects production, the resolution of the final image may be considerably higher,
but at the same time, scenes can contain many millions of polygons; the “polygon
is smaller than a pixel” rule of thumb continues to apply.
The functioning of a standard graphics system is typically described by an abstrac-
tion called the
graphics pipeline.
The term “pipeline” is used because the trans-
formation from mathematical model to pixels on the screen involves multiple
steps, and in a typical architecture, these are performed in sequence; the results
