Professor Celestino Soddu, Architect
Director of the Generative Design Lab. DST, Milan Polytechnic University, Italy
Chairman of GA, Generative Art International Conference (GA'98, GA'99)
With Generative Design it is possible to design and produce, with the same project and production line, a sequence of different and unique objects, Where the homologation and the leveling is not more acceptable and where subjective and multi-cultural identity is required as precious good, the Generative intelligent approach finds its irreplaceable space in the process design/manifacturing and intelligent production/communication/e-commerce.
Generative Design is a possible answer to aesthetic needs of contemporary man. As product of artificial environment using AI, natural complexity of industrial objects can breed a new fantastic world.
Argenia is the Generative Industrial Design software that I have designed. It is able to generate an endless sequence of different 3D models of objects, all different and unpredictable, but all recognizable as belonging to the same Idea, to the same design concept. They all toghether define a species and they can be produced by the same industrial equipment.
The system is based on two original software engines that works together:
1st, Artificial Intelligence device. The Idea is represented as Generative Code, performed as Generative DNA. This software engine is the same in all the design occasions and represents my design concept. I upgrade this code in front of the evolution of my Idea of Industrial Object.
2nd, Artificial Life device. The emulation of the project time evolution as Artificial Life. This engine manages the increasing complexity of each project, and it is able to generate a sequence of unpredictable and different 3D models.
The variuos scenarios of the same generative code can
be evaluated as adequate to predefined goals, production technical requests
and customer's functional requests. Being the system open, the code is
suitable to be adjusted in progress.
The Industrial Design activity is based on two fundamental phases, the creative and the evolutionary ones.
The 1st one is the phase of conceiving and expliciting the idea as operable system. This approach is built activating a logical connection between the existing world and a future possible world representing our wishes and thoughts. A design idea is the identification of a set of incoming possibilities that transcend specific "solutions" but identify the sense of the reachable quality. This designing moment involves the research on "how" the world may be transformed, as opposed to what may be the possible incoming scenario.
The 2nd one is the evolutionary moment, the moment of the development of the idea. This approach runs inside paths of refinement and increase of complexity of the projects. It involves the management of the project to reach the customer's requests and the wanted quality.
I have designed Argenìa, a form generation system which follows this approach and operates directly in the field of managing the evolution of a design concept: it is like a well organized design office, with hundreds of designers that know very well my design concept. In response to the incoming requests of the customers, they work representing and explicating my idea in a sequence of different design scenarios.
My virtual office Argenìa is founded on the possibility to explicitly separate the design activity in the creative moment and in the evolutionary moment. The creative one is reserved to my direct work on the code following the increasing quality of my idea of industrial object. This job builds and increases at all times the artificial DNA that represents the recognizability of my design imprinting. This creative process is activated from subjective interpretations with an adductive procedure. It is not possible to emulate these procedures using machines. The second moment, the managing of the evolutionary path of design activity, may be carried out using artificial devices able to emulate logical procedures, like Artificial Life procedures. The emulation of the evolutionary logic is useful for a very simple reason: to obtain the best operative design control on the complexity.
Designers know very well that the quality of a project depends, as conclusive factor, from the time spent to design. Following a creative and not an analytical approach, we must not try to accelerate the design developing path "deducing" from the requests the formalization choices. Instead we can perform the evolution of our idea using the requests and constrains as incoming opportunities of increasing complexity. This approach allows us to reach a higher quality. A more "creative" approach needs a longer time to spend in design activity.
But we can emulate it with machines. With Argenìa we can spend a lot of virtual time to manage the evolution of our design concept. With Generative Design we can fit the contemporary times and rhythms of the project management to the need of reaching a high standard of quality and complexity. The complexity is the only way to reach a quality adaptive to the changes and the uniqueness of each user. In this way the user finds the designed objects in tune with his uniqueness and unrepeteability as a human being.
These characters of complexity are the same as in the natural objects.
We can form the hypothesis that the actual passion for the natural world is not a reaction to the artificial worlds but to the artificial simplified and flattened products of the industrial era. The fascination of the craved natural world could be anything else than the desire of a high quality "artificial" environment fitting to the human needs of naturality. We look at a possible answer not only to "essential" needs, the confirmed and simplified needs, but also to the most complex needs of each person, belonging to his own identity and uniqueness.
Learning from Renaissance how to manage the complexity.
Leon Battista Alberti defines the Beauty of an artificial object "a coming together of all the parts, performed with proportion and logic in something in which it is possible to find again in each event, in a way that will not allow the inserting, extracting or changing of anything without decreasing its Beauty".
With generative design, and particularly with Argenìa, we can approach, directly, this complex paradigm of proportions and logic, and we can directly design the Beauty, or better our idea of beauty, before the realization of each single possible artificial event. This is the heart of the generative approach. Generative Art works for beauty, in the sense of the humanistic approach to the Renaissance, because the generative code, which is the core of each generative design, is the real structure of the idea. It defines how to bring together all the parts and the dynamic relationship amongst these parts in the evolution of complexity. The generative project defines what is the law of proportion and which logic the dynamic evolution will follow. All the events that this code can generate will be, in a humanistic sense, beautiful, or, if we prefer will belong and represent our concept of the world.
Furthermore the generative design produces events that are unique and complex. This uniqueness and complexity are strongly related to each other. As in Nature, each event is generated through an artificial life, which, as in the natural life, produces uniqueness, identity and complexity during an identifiable time. In the evolution of a project (performed as artificial life), the unpredictability of events is the opportunity to explicit choices, and then to build identity and recognizability of the idea.
This complexity is a natural-like complexity. We can recognize, in the artificial objects we produce through this generative approach, the harmony and the beauty of natural-like complexity that refers to the Humanistic approach of Renaissance: Man, Geometry, and Nature as references for "the harmony which is not thought as an individual caprice but as conscious reasoning." (L.B.Alberti, De re aedificatoria).
And, moreover, each designed object is unique and unpredictable. It is, like natural objects, an individual belonging to a species. The species is the design concept.
The hypothesis of my first generative projects (1988) was that the identity and recognizability of each designed object can be strongly connected with its evolutionary laws (its DNA). The random events that occur in its life cannot do anything but increasing its identity. The clouds are strongly modified by the wind, and they enlarge and change their image. The dunes shift at all times, modifying the desertís shape. And this endless regeneration performs impressive and unpredictable beauties.
But the clouds are always recognizable as clouds, the dunes hold, the shape of a dune. The natural DNA is recognizable not withstanding each event is different, unique and unrepeatable.
The generative project is the DNA of artificial objects.
So I performed my generative software to use the random events to increase the identity and the complexity of artificial systems (identity and complexity come together).
The first study case in the field of industrial objects was the design of a species of chairs. The random aspect of this system was realized with a different and unpredictable "speed" of a set of different generative devices belonging to the multiple possible faces of the same formalization: geometry, material, technology, etc.
This approach allowed me not to use a data base of pre-designed shapes, but only a set of generative machines that worked together.
A generative design, from my point of view, cannot use an array of data, but a set of different generative devices, like a set of different dynamic chaotic systems, that work together and use the unpredictable contamination each other to access to different "points of view", to different formalization of the same request. The recognizability of the project grows, increases its identity in front of the ever-larger quantities of alternatives
Constructing an Artificial DNA
First step, the construction of a generative code.
The devices: the code of a concept is built through a parallel series of interpretative codes/matrices that operate transformations of each input in an output event with an increasing complexity. How these transformations happen belongs to my concept of industrial design.
The structure: a primary paradigm, or better a set of superimposed and autonomous paradigms, that represent and manage the use of the procedures of transformation and their omothetic symmetry from the totality to the detail.
As each designer does, this conceptual moment is the construction of a proper world of interpreted references. This is the fruit of the passage from the exegesis to the ermeneutica that starts from the comprehension of our existing environment to the interpretation of the same environment.
This code is like the notebook of each designer. Where we cannot (only) find the observed, analyzed and systematized events, but one possible interpretation of them. Each sketch traces and discovers the subtended logic that the designer found in the event and, I would say, some evolutionary logic identified as procedure of transformation and increase of complexity of an idea.
In Argenìa this notebook is realized through a series of sketches/algorithms. In other terms the sketch, reporting to the interpretation of its own references in terms of transformation, of code of evolution of complexity of an event, can be annotated with an algorithm that, as it is known, defines how the transformation of the input in an output happens. Every possible result, obviously, "remembers" the initial event but the process of transformation is able to characterize the final event on the base of identified objectives (objectives of sense, identified on the concept and abstract field using specific meta-project configurations).
The storage of these matrices with algorithms is the historical memory of Argenìa. These procedures are the motor of each design activity. They act directly on the characters and on the recognizability of the final projects.
This approach brings, in the generative design that operates through repertoires of procedures and not of database of "solutions", to the possibility to reach, simultaneously, a whole series of objectives also if they are one in contrast with the other. For example, it is possible to reach goals belonging to different fields of reference as the world of the designer and his cultural referents, the customer's world, the world of the engineering of the product, the world of the communication, of the marketing, and so on.
The final project will have the ability to surprise each one of the design partners. It will be a pertinent representation of the idea of each one. Each partner will recognize the project as belonging to his proper idea for the reason that he will identify, following his specific point of view, the project as carrying the characters that, in terms of "procedures of increase of complexity", he had given
Figure 1: C. Soddu (3D generated models) and H. Dehlinger. (2D generated textures) 1999, design of a species of lamps with Argenìa.
In this sense the generative design, and especially Argenìa, performs an approach to the project able to put together different people and different competencies that may also be apparently conflicting. Because of this reason the generative approach to design is developing in sectors, as mechatronics, where the complexity of products is high. The complexity requires to improve the design activity with the simultaneous holding of different scientific and technological approaches. In the generative approach, the agreement among partners must be the usage of a common language: the language of the procedures, of "how" to transform and not "what" are the solutions. Because of the procedures are usable together, one after the other, the solutions are not.
The evolutionary paradigm.
We need to construct a peculiar paradigm in relationship to the specific data of a particular design occasion. Also this is multiple because it is reported to the various fields involved in the designing action. It manages the transformations in progress.
The customer's requests and the relative connected constrains must be transformed in fields of the possible evolution, in new fields of design possibilities. We must redefine these requests of the customer into procedures of increasing complexity and of increasing performances. It is clear that this "translation" is not only understanding the requests, but also operating a design interpretation of the same requests. It is already a design choice, belonging to our concept
It is possible, and it is necessary for a good generative design, to realize these translation-interpretations in multiple ways. It is possible to build a set of devices of transformation that presumably operate toward performances in line with the requests of the customer.
Following this approach we can represent the concept design as an evolutionary paradigm whose characteristics are:
In Argenìa this job is:
- the indication of "how" these events are being progressively transformed activating specific relationships with the others,
- the degree and the plausibility of the mutual contaminations,
- the structure and the quantity of the acceptable exceptions and on "how" these are managed to reach the goals and/or to capsize the following evolutions.
The charm of the generative design is that it is possible to use complex structures of proliferation and omothetia as the fractals, not acting on the forms, but acting on the logics. After all, the experience on the fractals has taught us that, in the recognizability of the final image, the form from which has departed doesn't count, but the idea, the recognizability belongs to the procedures adopted and repeated.
Artificial Life: managing the evolution of the project
The following moment is the manegement of the evolution. We use a design hypothesis, an evolutionary paradigm that checks and verifies the simultaneity of the possible evolutions and increases of complexity of the project. In the professional offices this moment concerns the management of the teamís project, the identification of the operational hypotheses to develop, the verification of congruity of the results to the hypothesized quality and the idea.
In Argenìa this moment means the realization of a structure of "Artificial Life" able to let evolve the project, testing and increasing its complexity, and surveying the multiplicity of the possible results as endless representation of the same idea.
So we can consider the double face of the generative design:
Artificial DNA and Artificial Life are the two systems that must be designed for activating a generative design. They are two separate projects, sometimes contrasted, but they represent the two faces of a same idea of project. With a deep difference. In the first one random doesn't exist, in the second random is one of the factors of control and amplification of the idea. In every case random must not be used "to produce random shapes" but to upset the code (that represents the idea) letting it react in a way to increase (and to render explicit in the project) its identity and recognizability, putting up unexpressed potentiality of the idea (or holes to fill).
Therefore the code will be always the same, at least in a single generation of events. The artificial life that will serve as environment to its evolution has to be always different, even if it has to be able to maintain a predefined degree of "difficulty" in way that the evolution of the project is completed with a sufficient degree of complexity.
Which is the result? Certainly the ability of the project, so trained by an "hard" artificial life, to answer to the requests, also unpredictable, of the consumer. A degree of complexity that can be similar to some natural objects or, I like to think, to the complexity of a historical city that owes its beauty and charm, and therefore its ability to answer to the specific needs of each man, to the lived long and difficult history, to a stratification of cultures, phases of expansion and contraction, and from the acquired (in this lived artificial life) ability to grow, rather to increase, its own uniqueness and recognizability.
But we need to explicit the main character of this "artificial design life": it produces not forms, but ever-new incoming fields of design opportunities.
Interactive processes between customers and Argenìa.
The hypothesis to pursue is that every following requests of the customer not "changes" the project, but increases its complexity. If the designer translates the objectives of the customer in terms of procedures of transformation and not of change of forms, the customer will find more and more in the project the mirror to his needs. But the designer will also have, from these following increases of complexity, the occasion to give to the project a more strongly characterization and recognizability of the design idea. The project will acquire a more evolved order and will be able to answer to a multiplicity, in progress, of requests. This means that the project will be more adaptive and flexible, and will be able of pertinent answers to the further unpredictable needs of the customer and, also, to the unpredictable needs of the final consumer.
In the generative design the traditional relationship with the customer has substantially changed in the form but not in the substance. The sequence: need > transformation procedure > generation of possible scenarios get the relationship between designer and customer in the correct respect of the mutual competencies. To the customer it is not asked of pointing out "how" to realize the project but only "what" performances he desires. To the designer it is asked of "interpreting" the customer's need using a procedure, a "how to transform" the project. And these interpretations of the designer have not communicated, as often happens, through a form-idea, but through evolutionary procedures that represent, in the plurality of the possible generated scenarios, the idea of the industrial designer. Eliminating the misunderstanding of possible false associations between form and idea but proposing the idea in its potentiality of transcendence.
The result is the fully possibility of carry out the feedback between customer and designer in times that are unthinkable with the traditional design management and, in more, respecting the role of both.
Recognizability, Identity and Designer's imprinting.
Generative Art is a possible answer to aesthetic needs or contemporary man. Natural complexity as a product of artificial environment can breed a new fantastic world.
1. The idea as generative project exists only if the idea is a recognizable code.
We build up our recognizing codes since infancy. This process is peculiar to Art. We learn, since first infancy, to recognize everything in the world, to recognize a chair among other similar objects. We were able to build our peculiar structure of chair idea. When we are in front of a chair that we never have seen before, notwithstanding that this shape is new and different from any other known chair, we succeed in recognize it as a chair. The recognizability is, therefore, our abstract synthesis of already lived for the valuation of the unpredictable incoming events. But the recognizability is not an objective category. We cannot learn to recognize objects if through our own experience. This synthesis belong to the field of subjectivity, it is a distinguishing mark of our uniqueness and identity of human persons. It is the subjective building of categories, of species for the identification of individual ware. This synthesis is also the first step to approach design processes.
2. The recognizability arises from the subjectivity.
The structure of the recognizability and of its identification from the user is therefore, fundamentally, a structure of connection between two subjectivities, between two syntheses of the world, between two ways to see the incoming possible environment: one subjectivity is belonging to the designer, the 2nd to the user.
And here the only possible field of investigation emerges. To manage, in the designing acts, these possible connections between different subjectivities: the complexity. We mean the complexity as an answer in terms of flexibility, of adaptivity to the inter-subjectivity; as a pertinent answer to the unpredictable subjective needs but, in the other side, being able to be recognized as it really is, therefore to exist as subjective idea of species.
We could say that the designer builds his own idea-product on the field of sense, building the complexity and multiplicity of this sense over his own references, while the subjectivity of the user is rendered explicit in the identification of a meaning (but also identifying a possible use) between those possible ones.
3. Subjectivity means not simplification (objective) but complexity.
Wanting to use the representations as metaphor of the relationship between subjectivity, objectivity and complexity, we could say that with a representation as the axonometry, that pretends to be objective, we would never succeed in representing the endlessly, while with the perspective, fundamentally subjective, we can represent and "check" the endlessly. The complexity of an idea-product, performed as generative project, cannot rise if from the subjectivity of the designer. And, more the subjectivity of the designer is recognizable, more the idea-product acquires quality and market.
4. Argenìa is a direct approach to the complexity, and in this it finds its strength irreplaceable power in the contemporary age.
Argenia don't use the random generation of forms. It works with the recognizability of the designing idea. It is an approach of human creativity improved by artificial tools. It is an easy designing approach to the complexity.
The complexity, in fact, is closed to an argenic project; it is its necessary fuel. As I often say to my students, if a designer complicates his designing life, he accelerates the project evolution, because he identifies more quickly the fields of possible design choices. In the structures of Argenia, this means that more the requests of the customer are complicated, more these requests are also contradictory, more constrains of different nature will be set to the project, more the process of construction of the recognizability of the idea will evolve.
Not only. Given that the customer's requests must be "translated" by the designer into transformation procedures, these requests are adding up one over the other, with a sequentially proceeding. So the management of the complexity becomes easy. One of the greatest potentialities emerged in the Generative Design is that it is a suitable approach for projects of great complexity where the designers must control a whole series of disciplines not directly connectable one each other.
In the contemporary age, where artificial objects become more and more complex, where the homologation and the leveling is not more acceptable and where subjective and multi-cultural identity is a more and more required as precious good, the Generative approach finds its irreplaceable space in the design and production.
 Celestino Soddu. L'artificiale progettato (designing
the artificial space). Casa del Libro Publisher, 1979
 Celestino Soddu. L'immagine non euclidea (not-euclidean image). Gangemi Publisher, 1987
 Celestino Soddu. Citta' Aleatorie. Masson Publisher, 1989
 Celestino Soddu. Simulation tools for the dynamic evolution of town shape. Oxford Polytechnic 1991
 Celestino Soddu, Enrica Colabella. Il progetto ambientale di morfogenesi (environmental morphogenetical design). Esculapio Publisher, 1992
 Celestino Soddu, Enrica Colabella. Spatial and image dynamical models for controlling and planning the development of differences in an increasingly international and interdipendent world, in the proceedings of the conference "Planning in a time of Change", Stockholm 1992
 Celestino Soddu. Human machine interaction in design processes. A new search software to generate the simulation of the increasing complexity of an idea about architecture, environment and industrial design, in the proceeding of the congress "Computer Science, Communications and Society: A Technical and Cultural Challenge " Neuchatel, 22/24 September. 1993.
 Celestino Soddu, Enrica Colabella. The project of morphogenesis. A way to design the evolution code of the environment, AESOP congress, Istanbul 23-26 August 1994.
 Celestino Soddu, Enrica Colabella, Artificial Intelligence in the teaching of architectural design, in the proceedings of the conference Hypermedia in Sheffield, Sheffield 1995
 Celestino Soddu, Enrica Colabella, Recreating the city's identity with a urban morphogenetical design", International Conference on Making the Cities Livable, Freiburg 1995.
 Celestino Soddu, Enrica Colabella, "Artificial Intelligence and Architectural design", CAAD Futures '95, Singapore 1995.
 Celestino Soddu, Enrica Colabella, A Natural Approach to Industrial Design: Argenic Design, New Simplicity? international furniture design conference, Helsinki Aug 1997
 Celestino Soddu. Generative Art. GA'98 proceedings, Editrice Dedalo Publisher, 1998
 Celestino Soddu. Argenia, the natural approach to Architecture and Industrial Design, in the proceedings of Mecatronic workshop at the Universtity of Lancaster, 27/28 May 1998. (ISBN 1-86220-057-2).
 Celestino Soddu, Recognizability of the Idea: The evolution process of Argenìa, in the proceedings of AISB99 Edinburgh 6-9 April 1999
 Celestino Soddu, Argenia, the generative process, Workshop Morphogenetic Design, Institute for higher education in the sciences and in the Art. Institute for Architecture Sint-Lucas, Brussels, 10 Aprile1999.
 Celestino Soddu. Recognizability of designer imprinting in generative artwork. GA'99 proceedings pre-print, 1999
 Celestino Soddu, Argenia, artís idea as generative code, proceedings of GECCO2000, 2000 Genetic Evolutionary Computation Conference, Las Vegas July 8 2000
 Celestino Soddu, New Naturality: a generativa approach to art and design, proceeding of Artificial Life 7 Conference, Portland August 2000
 Celestino Soddu, Basilica, an Architectural Generative Virtual Office based on Italian Renaissance Codeís interpretation, proceedings of Mosaic 2000 Conference, Seattle August 21, 2000.