Game Development Reference
In-Depth Information
are needed. For example, an avatar must have certain prerequisites in order to
meet these requirements. The three essential components are:
1.
A 3D mesh representing the shape of the body,
2.
Texture mapping on this mesh to provide the avatar with a realistic
appearance,
3.
An H-Anim model taken as a basis for the definition of motions. This also
comprises the definition of joints with certain degrees of freedom, as well
as adjustable length relations.
The term eGarments denotes digital, 3D models of real pieces of clothing. Most
online product catalogs only consist of 2D pictures. The easiest way to produce
eGarments is to generate data from a CAD program, which is used for the design
and the cutting construction of the clothing. State-of-the-art of cutting construc-
tion is, however, in most cases only two-dimensional. In order to generate a 3D
volume model of the garment, these faces must be sewed together virtually and
then transferred into a 3D grid model. eGarments are produced in FashionMe in
a multistage process. A real dummy is equipped with the specific garment and
is then scanned in 3D. The basis for this method is that the naked dummy was
scanned in the first instance so that the system knows the gauges of the dummy.
In a second step, the dummy is scanned wearing the garment. By subtracting the
known rough model from the dressed model, the necessary geometric data is
computed. The eGarment consists of the offset of the dummy, the garment's
surface, and the graphical information, which maps the surface and is used as a
texture.
The focus of the scanning technology used in FashionMe (provided by the
AvatarMe company) basically lies on the simple and fast generation of Internet-
enabled, personalized avatars that have a realistic appearance, and not so much
on an exact rendering of the actual gauges of a person. The scanner is
accommodated in a booth that can also be set up in public places (Figure 6). The
scanning process comprises digitally photographing the model from four differ-
ent perspectives. Based on these digital views, the avatar is computed by means
of existing rough models. In order to be able to move the model realistically, the
avatar is assigned a skeleton. The assignment of all necessary points in the avatar
for the individual joints is realized using predefined avatars. After size, weight,
age and sex of the scanned person have been recorded, the most appropriate
avatar is automatically selected from about 60 different pre-defined body
models. Then, the avatar is personalized using texture mapping. This procedure
provides a first skeleton, which can be further refined manually, in order to equip
fingers with knuckles, for example.
Search WWH ::
Custom Search