Modeling the Details - Blender Master Class: A Hands-on Guide to Modeling, Sculpting, Materials, and Rendering - page 50

closing the eyes or opening the mouth won't stretch
edges awkwardly or cause parts of the mesh to inter-
sect unpleasantly. This kind of loop-based topology
also helps when creating further variations on the
shape, and it makes it easier to place UV seams and
to UV unwrap the mesh without too much stretch-
ing (see Chapter 8 for more on unwrapping).

Avoid poles with lots of edges. A pole is a vertex
where three, five, or more edges meet—that
is, a point in a mesh that deviates from a grid-
like structure. Like triangles, poles can create
artifacts when subdividing a mesh. Poles with
three or five edges aren't so bad—indeed, it's just
about impossible to create anything but toroids
and grids without creating a few poles—but poles
with six or more edges subdivide poorly.

Create loops around important forms. This allows
you to easily select, deform, and animate your
meshes, and it also ensures they will subdivide
cleanly. For example, in Figure 5-1, the use of
edge loops that flow around the eyes makes it
easier to adjust their shape.

Align edges with the form. If your object is roughly
cylindrical, the edges of the mesh should flow
around its circumference and along its length.
If your object is roughly cuboidal, create it by
starting from a cube and adding loop cuts. In
general, try to create a mesh structure that goes
with the “grain” of the shape you are trying to
create, as shown in Figure 5-2.

Figure 5-1: The same head shape with three very different
meshes

Another reason that the middle mesh in
Figure 5-1 is the better choice is that its topology
is the best suited for use with the Subdivision Surface
(Subsurf) modifier, . The Subdivision Surface modifier,
which we covered in Chapter 4, is used to subdivide
and smooth a mesh. The algorithm used by the
Subdivision Surface modifier, Catmull-Clark sub-
division, works best when given a mesh constructed
like this one. When the Subdivision Surface modi-
fier is used with a mesh containing a lot of triangles
or long, oddly shaped faces, it can give poor results,
but when given well-constructed, flowing topology, it
produces very predictable, smooth forms.

Dealing with Difficult Topology
The rules listed above are simple, but you may run
into trouble following them from time to time, espe-
cially when trying to eliminate triangles and poles
from your models. Here are some tips for dealing
with difficult topology:

Plan ahead. Most topology woes can be sidestepped
simply by planning ahead. That's why, for
example, we made sure there were eight vertices
in the loops around the arms and legs when
creating the base mesh for the Bat Creature: It
made joining the hands easy, as there were no
surplus edges to join together when it came to
bridging the gap. Powers of 2 (8, 16, or 32) are
often a good way to think about this, but regard-
less, try to keep to even numbers when creating
edge loops. If you are new to 3D modeling, it
can be helpful to sketch your desired mesh over
a photo or your concept art, either in GIMP or
on paper, as shown in Figure 5-3.

What Is Good Topology?
Good topology for animation is usually good for
subdivision and vice versa. But what constitutes
good topology? While there are no absolute rules,
there are a few important principles. It's a mix of
art and science.

Avoid triangles and n -gons where possible. This
is the big one. While triangles are fine in a
static mesh that you don't intend to subdivide
or in a low-poly object for a game, if you plan
to subdivide your mesh, use as few triangles as
possible because triangles don't subdivide as
well as quads. Equally, n -gons are converted to
triangles before being subdivided, resulting in
the same kinds of problems.

Two tris make a quad. You can join two adjacent
triangles to make a quad, killing two birds with
one stone. To convert multiple triangles into
quads automatically, select your mesh and press
alt -J to turn suitable pairs of triangles into
quads.