Page 20 - Anthropometry, Apparel Sizing and Design

New directions in the field of anthropometry, sizing and clothing fit 15

this pool. During 2000–10, when most large-scale 3-D anthropometric surveys were
conducted, data sources were proprietary and/or costly due to the sheer cost involved
in conducting a survey. Participants of the survey got the first right to use the data,
and subsequently, they became available to other users. While companies continue
to use the linear data available with them, raw and processed data from 3-D scanners
are now mostly available in 3-D form. Anthropometric data can be represented in
the form of 3-D body scan, multiview images, one image, body contour, etc.
(Molyboga and Makeev, 2018). As a result the analysis, management, mining, and dis-
bursing of such diverse data have become a humongous task. Demand for algorithms
that can extract precise information about the human body from digital data is growing.
Accordingly, agencies that specialize in data management and analysis are now a part of
the data collection team.
Anthropometric data can be purchased online from various sources such as CAE-
SAR store (Store.sae.org, 2019) for data of CAESAR survey and DINED (Data.4tu.nl,
2019) for data of the Netherlands. World Engineering Anthropometry Resource
(WEAR) (https://www.bodysizeshape.com/page-1855750) is a consortium that shares
nearly 200 anthropometric data sets comprising 2-D and 3-D data from 20 countries on
a single platform. Fee-based bundled services are offered by companies such as
Anthrotech (Anthrotech, 2019) that scan and analyze specific population data pro-
vided by clients or provide solutions from their own data sets. Human Solutions
GmbH through their portal iSize provides 3-D processed data of French and German
populations (Anon, 2019). Global databases that cover all populations of the world are
not yet available.
Avatars created from statistically derived body scan data of a large number of sub-
jects of a particular population group are available publicly. These templates represent
typical body shapes and sizes extracted from a large database of scans and bring 3-D
and 4-D data directly to the designer (Ballester et al., 2016). Grouped and parame-
trized avatars of adult males and females obtained by principal component analysis
and partial least squares regression method are available (www.inkreate.eu)(www.
i-size.net). These are statistically derived sets of high-resolution 3-D templates. In
some cases, body movement data are incorporated into the forms through static poses
(Inkreate.com).
Prof Michael Reed of the University of Michigan has made several 2-D data sets
available on his website (Mreed.umtri.umich.edu, 2019) in public interest. Download-
able, 3-D parametric forms of adults and children in different poses are also available
on their site http://humanshape.org/. It is also possible to individualize the process of
avatar making at the level of the designer. Tools are available for converting any raw
body scan from any scanning device into a simulation-ready individual avatar (https://
www.inkreate.eu/) for use of the designer.
In all examples discussed earlier, the avatars are available in fixed static poses, and
data about movement are not available. The latest and most exciting advancement in
this field aims to fix this gap. Max Planck Institute, Germany (http://smpl.is.tue.mpg.
de/), has developed a fully animated parametric skinned multiperson linear model
(SMPL) (Fig. 1.5). SMPL is a realistic learned model of human body shape and pose
that is compatible with existing rendering engines, allows animator control, and is