The Importance of Resolution in CG - An Introduction to Computer Graphics for Artists - page 90

to make them all? Should they be separate, or can they be blended together to reduce
the overall polygon count? If the parts do not move in such a way that otherwise
hidden parts are exposed, or the camera will not reveal currently hidden parts, then
the hidden items may be left out of the model. Leaving parts out of a model is a
gross level of optimization and may be referred to as part resolution . This refers to
the number of parts represented of the number of parts the assembly should have.
To determine the best part resolution for a subject, one must consider the size of
the parts, their exposure to the camera, and the way they are meant to move. Parts
that must be separately articulated, such as the beams of a swing-arm lamp, usually
require more polygons so that they may move freely. If they didn't need to move, it
is sometimes possible to overlap parts in such a way that hidden polygons are
deleted. Movement makes this more diffi cult to accomplish, so it must be considered
when making decisions about resolution.
Texture maps can be applied to a single polygon or many at once. A cathedral
can be textured with one map attached to one shader, or thousands of maps attached
to thousands of shaders. The number of parts in the object will affect the number of
materials, but the parts can also be subdivided into sections that receive separate
materials. This is another level of resolution. How many materials will be used?

7.3 Polygons

The number of polygons in an object or a scene is referred to as the polycount.
Based on this is what is known as the poly budget . The poly budget is the number of
polygons allowed for a certain object or class of objects. Polygons require memory
to store and bandwidth to calculate their appearance in renderings. Because of the
number of polygons that can be involved in a scene, millions to hundreds of mil-
lions, beyond certain levels they are regarded as expensive. Because of this, consid-
erable effort has been expended to determine ways of reducing the impact of high
polycounts (Pouderoux and Marvie 2005 ; Ho et al. 2006 ; Bittner et al. 2009 ). The
resolution of a scene is informally equated with the number of polygons it con-
tains but this is not an accurate measure if anything other than polygons is present
in a scene.
Poly budgets vary considerably depending on what they are used for and in what
medium they will be presented. In entertainment, there is a noticeable difference
between poly budgets for video games and feature fi lms. Video game consoles are
designed to process polygons quickly, but this comes at the cost of texture memory
needed to process texture maps. For this reason, feature fi lms tend to use more high
resolution images than will be found in video games, but their poly budgets are
about equal or lower. In feature fi lms, resolution is governed fi rst by what is needed
to be completely realistic and secondly by means of optimizing the geometry
without sacrifi cing its realism. In video games, the combination of the poly budget,
texture limits, and restricted dynamic lights cause realism to be sacrifi ced in return
for rendering speed.