Multilayered Complexity Evaluation within Configurators for Design Responsible collaborative systems for architectural and product design

Abstract

This paper describes the concept of integrating several complexity evaluation methods, previously developed and tested by the authors, into one product configurator through a technical prototype. In this case variations of an online configurator for design products based on a choice of these digital complexity evaluation methods developed between 2015 and 2020 are presented. This research shows that an integration of complexity evaluation for several Gestalt qualities in one product configurator is feasible, though the amount of aspects of each of these qualities and the necessary effort to be invested to achieve an integration that is suitable for customer use may vary. The concept is illustrated using a simple test case, i.e. an online shelf configurator.

Full text

Multilayered Complexity Evaluation within Configurators
for Design
Responsible collaborative systems for architectural and product design
Matthias Kulcke1, Wolfgang E. Lorenz2
1HafenCity University and Hamburg University of Technology 2TU Wien
1matthias@kulcke.de 2lorenz@iemar.tuwien.ac.at
This paper describes the concept of integrating several complexity evaluation methods,
previously developed and tested by the authors, into one product configurator through a
technical prototype. In this case variations of an online configurator for design products
based on a choice of these digital complexity evaluation methods developed between 2015
and 2020 are presented.
This research shows that an integration of complexity evaluation for several Gestalt
qualities in one product configurator is feasible, though the amount of aspects of each of
these qualities and the necessary effort to be invested to achieve an integration that is
suitable for customer use may vary. The concept is illustrated using a simple test case, i.e.
an online shelf configurator.
Keywords: Configuration, Mass Customization, Complexity, Gestalt.
INTRODUCTION
Complexity is a common denominator as a starting
point to identify possible areas of objectification
within several aesthetic Gestalt qualities like
proportion, color, sculptural consequence and
others (see e.g. Bense 1971, von Cube 1965, Kulcke
2019 and Lorenz and Wurzer 2020) to make them
tangible for use in mass customization. While some
areas of producing, interpreting and judging objects
in regards to each of these Gestalt qualities will very
likely remain forever subjective, others are bound to
be continually and further objectified through
design research and are thus more and more
suitable for integration into (semi-) automated
collaborative design systems such as online product
configurators handled by customers.
The authors are aware of differences in the
experience of Gestalt aesthetics due to cultural and
regional differences; therefore the quantitative and
qualitative measurement methods presented here
allow for a separation of data and their interpretation
(in terms of absolute values and ranges). Thus
regional adjustments can be integrated utilizing IP
address evaluations. Regarding untrained co-
designers, the need for feedback is indispensable in
order to provide them with an aid on the way to their
own design, taking into account production-related
and other practical rules and constraints (see e.g.
Felfernig 2014) as well as aesthetic expertise to
ensure long-term satisfaction with the product.
To achieve responsible professional practice
when designing collaborative systems for
asynchronous customer dialog such as online
product configurators, especially for design
products like architectural building parts, the
integration of complexity evaluation on several
layers of aesthetic Gestalt quality in one system is a
Volume 2 – Co-creating the Future – eCAADe 40 | 9

valid and necessary strategic element to be
considered by professional practitioners.
It may very well be a most crucial question of
work ethic (and not least employment) for architects
and designers, if mass customization either leads to
a mere transfer of design responsibility and work
from professional designer to customer or to an
increase of support (see Zuboff and Maxmin 2002),
aiding the customers to make an informed and
professionally guided choice of the optimum within
a reasonable timeframe truly meeting his or her
momentary and long-term needs. It still remains to
be seen if a democratization of architecture through
mass customization (see e.g. Steinbusch and
Walcher 2013 and recently Kolarevic and Duarte
2019) will be in reach in the near future.
MASS CUSTOMIZATION AND
CONFIGURATORS
Mass customization is the attempt to manufacture
individualized products according to customer
requirements under conditions of mass production
(see Pine 1993); within the scope of available
production techniques. The various strategies such
as modular construction, Computer-Aided Design
and Computer-Aided Manufacturing, paired e.g.
with Computerized Numerical Control milling, are
not the subject of this paper, but rather the
interaction with the customer acting as co-designer.
Regardless of whether the possible adjustments
concern cosmetic, material, surface-related,
dimensional or geometric customization, the
authors examine the (intermediate) results of the
design process; primarily as 2-dimensional outputs
to provide automated responses aiding informed
customer decisions during configuration.
Since most consumers are not trained designers,
guidance is needed to avoid frustration due to lack
of experience. This support is realized by means of
knowledge-based configuration. Online
configurators are well established but do not yet
provide responsiveness aiding customers to
evaluate their product instantiations in accordance
to professional aesthetic criteria (see also Kulcke
2019). This is not least owed to the fact that some
facets of gestalt aesthetics are highly subjective.
Nevertheless, there are design criteria that can be
measured and are therefore objectifiable.
This paper deals with some of these objectifiable
criteria and proposes exemplary workflows for their
sound implementation during the configurational
design process (Figure 1).
OBJECTIFYING AESTHETIC GESTALT
QUALITIES THROUGH MEASURES OF
COMPLEXITY
The matter of complexity as a key to objectify
different aesthetic Gestalt categories, among others
with the aim to make them tangible for
informational strategies within configurator
systems, has already been discussed by the authors
(see e.g. Kulcke and Lorenz 2015 and Kulcke 2019),
but research on this subject is still ongoing.
It is this objectification of Gestalt categories
while keeping in mind possible adaptation to
cultural differences (see in the field of color e.g. John
Gage 1993) that is decisive for a successful merging
of aesthetic analysis and designer/customer dialog
within configuration systems. This constitutes the
ground work for this research and several Gestalt
categories have been explored in detail.
The aesthetic areas that the authors have already
evaluated for their possible implementation through
multiple (online) applications include proportion,
color and sculpture. Complexity as their common
denominator for objectification and analysis in these
areas is increased by the number of colors and
proportions respectively, and in the case of the
fractal dimension marked by the tendency toward a
higher value of the box-counting dimension
(although, the focus here is more on the
development of characteristics across several scale
levels).
10 | eCAADe 40 – Volume 2 – Co-creating the Future

Proportion
In reference to Vitruvius as translated e.g. by
Rowland and Howe (1999) the authors have
developed two strategies of proportion analysis, one
referring to the Vitruvian take on the term
Symmetria called grid analysis (Kulcke, 2019), the
other in reference to the term Eurythmia called
gradient analysis (Kulcke and Lorenz, 2015 and
Lorenz and Kulcke, 2021). Both of these methods are
characterized by a decisive common denominator;
they don't evaluate the quality of a certain
proportion, like the golden section for example, but
rather they indicate and quantify the presence of
proportional relationships.
The grid analysis detects the presence of
previously defined proportions like the golden
section in objects by trying to fit grids in different
resolutions based on these proportions within a
given rectangle (constituting the object itself or a
part of the object) and measuring grid resolution as
well as deviation from a perfect fit (see Kulcke 2019).
Both methods examine the repetition of
proportions within a two-dimensional design, such
as those displayed on facades, using corner points of
the overall shape, openings such as windows and
doors, or decoration elements (see Figure 2 for an
example of gradient analysis), thereby identifying
dominant proportions.
Figure 1
Schematic Image of
responsive design
process
Figure 2
A. Loos, House
Steiner, North
elevation: 20 most
repetitions of
gradients
Volume 2 – Co-creating the Future – eCAADe 40 | 11

It becomes obvious that these are complexity
measurements based on redundancy. However, the
interpretation of the output data is still a subject of
research and depends not least on the cultural
background. For further details see Kulcke and
Lorenz (2015) and Lorenz and Kulcke (2021).
Color
The choice of the color of a product is to many a
highly subjective matter, in a majority of cases
connected to very personal preferences and/or
individual experiences and associations. A handhold
to objectify the aspect of color is again complexity,
which is increased with the number of color
contrast-related layers in a given or chosen set.
Contrasts to be taken into account have been named
and defined by several artists and scholars; a
predominant name of modern times in this area is
still Bauhaus teacher Johannes Itten (see e.g. Itten
2009). But there are quite a number of color theories
and models to organize hue, saturation, brightness
and additional criteria (see e.g. Küppers, 2000).
Some contrasts are precisely calculable within a
certain color theory by referring to a matching
geometrical model for color location, e.g. the
complementary color, which may be located exactly
opposite to a chosen field of a specific color wheel.
Others, like the contrasts of cool and warm or light
and dark may be at least approximated by a digital
algorithm in reference to specified color theories.
Quantitative color analysis as in identifying
overall color distribution within given images is
already available online as a tool created
accompanying research to combine color analysis
and fractal analysis (Lorenz and Wurzer, 2020). Both
methods have been implemented as web
application, freely available online [1]. Some of those
color analysis methods are based on (or
implementations of) algorithms (and javascript
programs) that are freely available on the internet,
such as 'getAverageColourAsRGB.js' [2] (which
calculates the average color of an image as RGB),
'vibrant.js' [3] (which computes the prominent colors
named 'vibrant', 'muted', 'dark vibrant', 'dark muted'
and 'light vibrant') and 'color-thief.js' [4] (which
computes the dominant color and the dominant
color palette, each with aggregation); other
methods include calculating the ten most noticeable
(main) colors with tolerance, the number of
predefined color ranges, saturation and lightness.
As an example serves the method of counting colors
in an image, divided into twelve predefined
intervals: Each of these intervals covers a color range,
reaching from reddish over orange, yellowish,
greenish, blueish to violet, each with different
lightness gradients. Figure 3 shows an evaluation for
Alexander Calder's 1974 sculpture 'Flamingo'
situated at the Chicago's Federal Center Plaza.
Sculpture
Within the field of architecture fractal analysis plays
a dominant role in measuring sculptural complexity
or roughness by utilizing box-counting (see Ostwald
and Vaughan 2016; Lorenz and Wurzer, 2020; Lorenz
and Kulcke 2021). Again, this method quantifies
complexity as a basis for collaborative aesthetic
judgment from this measurement.
The visual complexity results from the presence
of elements on different visual levels. If the size of
Figure 3
FRACAM: Color
distribution of
Alexander Calder's
1974 sculpture
'Flamingo' situated
at the Chicago's
Federal Center
Plaza.
12 | eCAADe 40 – Volume 2 – Co-creating the Future

architectural elements corresponds to the viewer's
field of vision and his or her distance from the
building, then this is a continuous characteristic
across several levels of scale. It is this property that
can be measured using the box-count method. With
this method, a grid is placed over the two-
dimensional black and white representation of e.g.
the facade and the boxes that cover lines of this
representation are counted. After several reductions
of the grid size (size of the boxes), the results are set
in relation to the grid scale in a double logarithmic
graph with the slope of the regression line is
corresponding to the box-count dimension.
It is not the height of the resulting box-counting
dimension that is decisive, but rather the continuity
of the roughness over several viewing levels.
The web application FRACAM integrates several
variants of the box-counting method, including
those that can be used not only on 2-dimensional
black and white graphics, but also on grayscale
images and color images, whereby brightness
shading or colors are included in the measurement
(see Figure 4).
INTEGRATION OF AESTHETIC GESTALT
QUALITIES IN PRODUCT
CONFIGURATORS
For an integration of aesthetic Gestalt qualities into
product configurators, preference should lie on
qualities that may be objectified to be suitable for
implementation in digital algorithms. Also,
potentially underlying subjective judgments on part
of the designers and authors of such systems should
be made transparent to the users. Thus, such
configurational system containing functions to
enable analysis and response in the realm of Gestalt
qualities are also bound to serve pedagogical
purposes – they teach their users i.e. customers in
this case about aesthetic Gestalt qualities.
It is crucial to note, that there is no one ideal
online product configurator including analysis and
response in the realm of Gestalt qualities, as well as
there is no definite most suitable order of customer
definition of parameters to instantiate his or her
order. What types of analyses are to be included and
in what order they appear has to be considered and
decided on different levels of system design.
With an online configurator, which is basically
structured step by step, i.e. sequentially, the
question arises to which extent it should be possible
to jump back and forth between the individual
construction steps. To define mutual influence or co-
dependency of components (e.g. size of the shelve,
color, or selection of material) becomes part of
designing the solution space and its accessibility to
the customers. Such considerations of
interdependence is also crucial to allow subsequent
reorganization of the design steps or their
adjustment on the fly. The definition of which step is
based on which must be taken into account when
designing the configurator.
Figure 4
FRACAM: Result of
Box-Counting of
Frank Lloyd
Wright’s Robie
House.
Volume 2 – Co-creating the Future – eCAADe 40 | 13

The Product
Specific products and product design approaches
will have different prioritization concerning
aesthetic parameters. This is due to the product itself
and associated customer expectations regarding a
product or a product group but it is also dependent
on the specific design and the designers' intentions.
Mass customization is the interaction of what is
technically possible, which enables cost-effective
and practicable production, on the one hand, and
the customer's desire to individualize the product in
certain aspects on the other hand. The former means
evaluating the process of production for a balance
between flexibility and productivity; while the latter
means using market research to find out which areas
of a product the customer would like to customize at
all. Too many options that the customer may not be
interested in or want at all can frustrate him/her
within the configuration path and lead to
abandonment. The number of choices for each
parameter (e.g. number of colors) must also be
limited so that the customer is not overwhelmed
with a multitude of choices. This is called the
‘paradox of choice’ (see Schwartz 2004).
The Target Group
Those targeted by online configuration services may
cover a wide scope of design competence, ranging
from total laymen to design professionals who are
able to judge their manipulations of design
parameters in an expert fashion.
This calls for specific product designs as well as
specific product configurator software architecture ;
tailored to the target group.
Functional, Constructional, Logistic and
Other Product Related Constraints
The flexibility, respectively the adaptability of a
specific product according to the customer's wishes
not only requires consideration from a Gestalt
aesthetic point of view, but also include restrictions
due to technical feasibility and transport. E.g.
regarding mass customized architecture, using
modules, restriction derive from the maximum
allowed height during transportation and the
minimum mandated height according to
regulations. Further restrictions relate to static
criteria or machine sizes for production. Although
CNC milling machines allow flexibility in production
(for the machine, it doesn't matter whether the same
piece is produced or different ones, since this is
about changes in the input data, not machine
settings), the maximum possible size and processing
of the workpiece is directly dependent on the
machine used.
Since mass customization is about making (at
least in part) individualized products under mass-
produced conditions production and logistics will
always be crucial factors. To realize its full potentials,
among others avoiding waste (through just-in-time
production), not only functional but also aesthetic
long-term customer satisfaction is to be integrated
in configurational systems, balanced with an overall
sensible solution space design.
Technical Feasibility
Although computational capability and customer
hardware power is constantly on the rise, not every
concept of analysis is technically feasible. Therefore,
layers of aesthetic analysis need to be prioritized, if
necessary simplified and well adjusted to match
performance of an online user interface.
By taking this aspect into consideration certain
technologies will still have to be ruled out for
implementation in online applications.
In recent research the authors have discussed
the question of computer aided multilayered
complexity analysis, combining tools for detecting
and quantifying proportion with those aimed at
sculptural complexity (Lorenz und Kulcke, 2021).
AN EXEMPLARY PROTOTYPE
CONTAINING MULTILAYERED
COMPLEXITY EVALUATION
To apply these strategies in combination a
parametric shelf system, a comparatively simple
product was chosen. This modular shelf system may
be manipulated in regards to the number as well as
14 | eCAADe 40 – Volume 2 – Co-creating the Future

height, width and depth of elements to form a
rectangular shelf. In addition the color of its modules
may be manipulated as a group or individually.
These decisions on part of the consumer may be
used as an introduction into a semi-automated
dialog concerning complexity as a step on the path
toward the Gestalt of the concluding instance of the
configuration process. To enable this semi-
automated dialog regarding aesthetic Gestalt
qualities, the latter have to be integrated as sets of
rules and constraints guiding the way toward
balanced design complexity (Figure 1).
Responsive Configuration of Proportion
The gradient analysis has been put to work in VBA for
AutoCAD, PHP and C# (see Kulcke, Lorenz 2015, 2016
and 2021).
The grid analysis has been tested as PHP code
and is thus suitable for online applications.
These instruments of proportion analysis, if
integrated in the configurator system, aid customers
in finalizing the measurements of their product
instantiations.
From a combinatorial point of view the number
of dialog scenarios which may result from adding
Gestalt analyses on multiple levels are manyfold,
therefore they are presented here in an exemplary
manner.
A conceivable scenario implementing grid
analysis in a shelf configurator for defining vertical
and horizontal shelf spacing is represented as a
flowchart in Figure 5.
Responsive Configuration of Color
Quantitative color analysis has been realized in
JavaScript (Lorenz 2017) and is therefore ready for
integration in online product configurators.
Integrated in a semi-automated customer dialog
these tools for color analysis will aid to chose
successive colors after a first choice or embedded in
a context with already existing colors. The latter
would be consistent with the assumption that
reducing the number of choices reduces possible
customer confusion (or frustration) during their
selection process.
A scenario is represented as a flowchart in Fig. 6.
Figure 6
FRACAM: Flow
Chart of color
configuration
Figure 5
FRACAM: Flow
Chart of proportion
configuration
Volume 2 – Co-creating the Future – eCAADe 40 | 15

Responsive Configuration of Sculpture
Fractal analysis using different box-counting
methods have been realized as online apps e.g. as
the tool FRACAM (Lorenz and Wurzer, 2020).
In the simplest form, measurements apply to
two-dimensional black and white images of facades
(i.e. elevations), with the grid size being halved in
each iteration (for details of the process, see the
"sculpture" section). To improve accuracy, the
reduction factor can be adjusted or the empty space
around the object to be measured (for more detail
on influences see Lorenz and Kulcke 2021). In order
not to lose performance in the case of an online web
application (and to be able to gibe quick feedback to
the customer), some simplifications were made in
FRACAM (see Lorenz and Wurzer 2020). This includes
downscaling the image size or limitation of
reduction factors. Accuracy and speed have to be
weighed up here (tests with FRACAM so far it has
been shown that the results are sufficient for
estimating a trend). Integration of a database is
planned for a better assessment of the results. This
allows comparison of the measurements with
examples of iconic architecture. A general
estimation of the design direction in the sense of
typology thus seems possible.
An exemplary scenario is represented as a
flowchart in Figure 7.
The Multilayered Configurational System
In 2016 a simple shelf configurator has been
described and prototypically realized in HTML5, PHP,
JavaScript and WebGL which now forms the basis of
more complex configuration to integrate
multilayered responsive Gestalt quality analysis (see
Kulcke 2016).
Semi-automated desynchronized designer/
customer dialog can now be refined in the area of
proportion by implementing gradient and grid
analysis, in the field the color with quantitative color
analysis and concerning sculpture regularity
through implementation of box-counting
algorithms that all have been programmed suitable
for online applications in PHP and Java Script, as
needed for implementation in online product
configurators.
Especially the use of PHP in the specific online
configuration environments discussed here is crucial
for the implementation of differentiated Gestalt
analysis tools in multiple layers. Since PHP is
characterized by the ability to use processing power
on the web server as opposed to client side
processing it is ideal for resource intense analytical
calculations as part of the co-design man machine
dialogue.
The analytic modules allow for testing of
different product types, e.g. furniture like shelf
systems or building parts like facade elements.
DISCUSSION AND OUTLOOK
We have to acknowledge that possible
democratization of design through collaborative
systems like online product configurators is more
than being able to chose your favorite color, but
about determining what truly will be your favorite
color for a particular product, purpose and context.
This is where the profession concerned with design
and architecture and its protagonists come in; to be
highly informed and thus enabled to inform those in
need of high quality design. This information and its
asynchronous communication may at least partially
be integrated into digital configuration systems
Figure 7
Flow Chart of
sculpture
configuration
16 | eCAADe 40 – Volume 2 – Co-creating the Future

using rules, constraints and responsiveness utilizing
the Gestalt analysis tools presented in this paper.
It does not lie in the scope of this research to
prove that a balance of complexity across several
Gestalt categories will enjoy unanimous approval in
the design community or with layman. This research
shows, however, that it is possible to integrate
Gestalt discussion into online configuration in a
manner of balancing complexity across several
aesthetic layers.
The task remains to push the boundaries of
aesthetic analysis based semi-automated
responsiveness within product configurators and to
refine the underlying digital algorithms while
staying aware of the limits of automated design
systems as well as customer design competence.
Specific areas of future improvements are: online
performance of analysis algorithms, pairing of
specific design products with specific analytic
algorithms and semi-automatic consumer
communication. The evaluation and improvement
of the described approach embedded in lectures
and design studios with architectural students held
by both authors is ongoing.
References
Bense, Max (1971): Zeichen und Design –
Semiotische Ästhetik. Agis-Verlag. Baden
Baden.
Birkhoff, George D (1933): Aesthetic Measure.
Harvard University Press. Cambridge, MA.
von Cube, Felix (1965): Kybernetische Grundlagen
des Lernens und Lehrens. Ernst Klett Verlag.
Stuttgart.
Felfernig, Alexander; Hotz, Lothar; Bagley, Claire;
Tiihonen, Juha (2014): Knowledge-Based
Configuration. Morgan Kaufmann. Waltham,
MA.
Gage, John (1993): Colour and Culture: Practice and
Meaning from Antiquity to Abstraction. Thames
and Hudson. London.
Itten, Johannes (2009): Kunst der Farbe.
Christopherus Verlag GmbH & Co.KG. Freiburg
Kolarevic, Branko; Duarte, José Pinto (2019): Mass
Customization and Design Democratization,
Routledge, London.
Kulcke, Matthias; Lorenz, Wolfgang E. (2015):
Gradient-Analysis: Method and Software to
Compare Different Degrees of Complexity in
the Design of Architecture and Designobjects.
Proceedings of eCAADe 2015, Volume 1,
Vienna, pp. 415-424.
Kulcke, Matthias; Lorenz, Wolfgang E. (2016):
Utilizing Gradient Analysis within Interactive
Genetic Algorithms. Proceedings of eCAADe
2016, Volume 2, Oulu, pp. 359-364.
Kulcke, Matthias (2016): Connecting Online-
Configurators (Including 3D Representations)
with CAD-Systems - Small Scale Solutions for
SMEs in the Design-Product and Building
Sector. Proceedings CAADence in Architecture.
Budapest. 61-66.
Kulcke, Matthias (2019): Gestaltkonfiguration und
Verantwortung. HafenCity University Hamburg
Küppers, Harald (2000): Harmonielehre der Farben.
Theoretische Grundlagen der Farbgestaltung.
DuMont. Köln.
Lorenz, Wolfgang E.; Wurzer, Gabriel (2020):
FRACAM: A 2.5D Fractal Analysis Method for
Volume 2 – Co-creating the Future – eCAADe 40 | 17

Facades Test Environment for a Cell Phone
Application to Measure Box-counting
Dimension; Proceedings of eCAADe 2020,
Berlin, Germany, pp. 495-504.
Lorenz, Wolfgang E.; Kulcke, Matthias (2021):
Multilayered Complexity Analysis in
Architectural Design: Two Measurement
Methods Evaluating Self-Similarity and
Complexity; In Fractal Fract. 2021, 5(4), 244;
https://doi.org/10.3390/fractalfract5040244
Naboni, R; Paoletti, I (2015): Advanced
Customization in Architectural Design. Springer.
Cham/Heidelberg/New
York/Dordrecht/London.
Ostwald, Michael J.; Vaughan, Josephine (2016): The
Fractal Dimensionof Architecture, Birkhäuser,
Basel.
Piller, Frank; Tseng, Mitchell M (2010): Handbook of
Research in Mass Customization and
Personalisation. World Scientific Publishing
Co.Pte.Ltd. Singapore.
Piller, Frank; Piroozfar, Poorang A E (2013): Mass
Customisation and Personalization in
Architecture and Construction. Routledge.
London and New York.
Piller, Frank; Walcher, Dominik (2016): Mass
Customization und Social Commerce –
Grundlagen, Kundennutzen und
Konsumentenverhalten. Center for Co-Creation,
Salzburg University of Applied Sciences.
Salzburg.
Pine, Joseph B. (1993): Mass Customization - The
New Frontier in Business Competition. Harvard
Business Review Press. Boston.
Rowland, I.D.; Howe, T.N. Vitruvius (1999) Ten Books
on Architecture. Cambridge University Press.
Cambridge.
Schwartz, Barry (2004): The Paradox of Choice: Why
More is Less. Harper Collins. New York.
Steinbusch, Michael; Walcher, Dominik (2013): Open
Architecture - Wie Partizipationsgesellschaft
und technologischer Wandel die Architektur
der Zukunft verändern werden. Lulu Press Inc.
Raleigh NC.
Zuboff, Shoshana; Maxmin, James (2002): The
Support Economy - Why Corporations Are
Failing Individuals and the Next Episode of
Capitalism. Viking Penguin. New York.
[01] http://www.iemar.tuwien.ac.at/processviz/
fracam/index.html (last accessed March, 26th,
2022)
[02] gist.github.com/nucliweb/a3a23febee3ccb9a1
b5dc93dadf94fe7 (last accessed March, 26th,
2022)
[03] https://github.com/jariz/vibrant.js/ (Zwarts, J.,
2015) (last accessed March, 26th, 2022)
[04] https://lokeshdhakar.com/projects/color-thief/
(Dhakar, L., 2015) (last accessed March, 26th,
2022)
18 | eCAADe 40 – Volume 2 – Co-creating the Future