TEAnO, or the computer assisted generation of manufactured aesthetic goods seen as a constrained stream of technological unconsciousness
Paolo Ferrara
TEAnO – Etnoteam spa
e-mail: pferrara@etnoteam.it
Gabriele Foglia
TEAnO – Infostrada spa
e-mail: gabriele.foglia@infostrada.it
Marco Maiocchi
TEAnO – I.NET spa –
Politecnico di Milano
e-mail: m.maiocchi@inet.it
Abstract
TEAnO (Telematica, Elettronica,
Analisi nell'Opificio) was born
in Florence, in 1991, at the age of 8, being the direct consequence of years of
attempts by a group of computer science professionals to use the digital
computers technology to find a sustainable match among creation, generation (or
re-creation) and recreation, the three basic keywords underlying the concept of
“Littérature potentielle” deployed by Oulipo in France and Oplepo in Italy (see
“La Littérature potentielle (Créations Re-créations Récréations) published in
France by Gallimard in 1973).
During
the last decade, TEAnO has been involving in the generation of “artistic goods”
in aesthetic domains such as literature, music, theatre and painting. In all
those artefacts in the computer plays a twofold role: it is often a tool to
generate the good (e.g. an editor to compose palindrome sonnets of to generate
antonymic music) and, sometimes it is the medium that makes the fruition of the
good possible (e.g. the generator of passages of definition literature). In
that sense such artefacts can actually be considered as “manufactured” goods.
A
great part of such creation and re-creation work has been based upon a rather
small number of generation constraints borrowed from Oulipo, deeply stressed by
the use of the digital computer massive combinatory power: S+n, edge
extraction, phonetic manipulation, re-writing of well known masterpieces,
random generation of plots, etc. Regardless this apparently simple underlying
generation mechanisms, the systematic use of computer based tools, as well as
the analysis of the produced results, has been the way to highlight two
findings which can significantly affect the practice of computer based
generation of aesthetic goods:
·
the deep structure of an aesthetic work persists even
through the more “destructive” manipulations, (such as the antonymic
transformation of the melody and lyrics of a music work) and become evident as
a sort of profound, earliest and distinctive constraint;
·
the intensive stream of computer generated “raw” material
seems to confirm and to bring to our attention the existence of what Walter
Benjamin indicated as the different way in which the nature talk to a camera
and to our eye, and Franco Vaccari called “technological unconsciousness”.
Computers are useless. They can only give you
answers.
Pablo Picasso
(1881-1973)
1. The nature of
the beast: the potential literature
In
1960 François Le Lionnais, as an offspring of the Collège de Pataphysique, founded
the Oulipo. The name of the group is an abbreviation of Ouvroir de Littérature
Potentielle (Workshop of Potential Literature). During numerous sessions
following the foundation, the members discussed what they later referred to as
the mechanics of the art of literary creation. In 1961 François Le Lionnais
conceived the term “potential literature” to set up a literary framework for
the work Cent Mille Milliards de Poèmes (One Hundred Thousand Billion Poems) by
Raymond Queneau. Actually he didn’t provide a real definition for such new-born
notion which actual meaning could only be derived by a number of samples: the
potential literature does not fix one and for all the sequence of the textual
elements which a literacy work consist of, it fixes instead the formal methods
to be used to re-structure such elements in order to generate an instance
of the class of all possible texts..
A
potential literary work can thus be “played” rather than read: the user is
provided with the game rules and a set of and the basic elements to be combined
and recombined according to the given rules.
One of the most famous formal methods proposed by Oulipo is
the Jean Lescure's ‘Méthode M ± n’. As Lescure says:
«La méthode M ± n, que l'on propose d'abord sous la forme encore limitée
dite S + 7 (forme qui a donné à la méthode son nom), consiste à remplacer dans
un texte existant (de qualité littéraire ou non) les mots (M) par d'autres mots
de même genre qui les suivent ou les précèdent dans le dictionnaire, à une
distance variable mesurée par le nombre des mots. Aussi S + 7 veut dire
simplement que l'on remplace tous les substantifs d'un texte par le septième
qui le suit dans un lexique donné» [1]
According
to Lescure’s statement, Keats’:
A thing of
beauty is a joy for ever:
Its loveliness increases; it will never
Pass into nothingness; but still will keep
A bower quiet for us, and a sleep
Full of sweet dreams, and health, and quiet breathing
Could
be turned into N/Adj/Adv (nouns/adjectives/adverbs) + 3 Keats’[1]
A thirst of
beaver is a judge for evermore:
Its love-kindness increases; it will nevertheless
Pass into notification; but stoutly will keep
A bowler quixotic for us, and a sleep-car
Full of swell dregs, and health insurance, and quixotic breechblocks.
However,
the “combinational itch” is not confined to the literacy domain. Similar
researches have been carried out by Oupeinpo, founded for the first time in
1964 (and re-founded may times in later years); an interesting example of
formal painting method proposed by Oupeinpo is the so called “proportioned
colour painting”, consisting in the pre-definition of the exact percentage of
each colour which must be present in the final picture[2]. More formal
painting methods applied by Oupeinpo are described in [2]
A
further example of combinational painting is represented by a mobile electro-painting with on-off lights
entitled " Jazz " by Frank J. Malina, which made available 211 (i.e.
2048) different combinations to the observers [3].
Actually.
it must be noted that Oulipo and Oupeinpo are just the best known one of
several OuxPo’s imagined by Le Lionnais, who also defined the OuReliPo (Ouvroir
de Religions Potentielles), the OuMuPo (Ouvroir de Musique Potentielles), the
OuCuiPo (Ouvroir de Cuisine Potentielles), etc.
2.
Many years later, someone yelled: «Remember the Alamo!»
Since
the very first steps of the Ouplipian mice[3],
the generation of instances of texts of potential literature was performed by
hand, even when such rules were extremely arduous to follow. Or better, the
challenge was to defined more and more sophisticated rules (and consequently
sophisticated text generations or manipulations)[4] revealing in
this the underlying enigmistic spirit of the Ouplepians.
In
1982 Paul Braffort and Jacques Roubaud founded the ALAMO or the Atelier de
Littérature Assistée par la Mathématique et les Ordinateurs during a workshop
held in Villeneuve-lès-Avignon by the Oulipo; the aim of ALAMO was to explore
the potential of digital computer technology applied to the generation of
potential literature. As first results ALAMO made available a number of
software programs aimed to exploit the potential of the combinatorial
literature like Litanie by Jean Meschinot”, Baci by Quirinus Kuhlmann, Sonetti
by Queneau, Dizains by Bénabou” (strophes of ten verses), Triolets by Braffort”
(ballads of eight verses)[5]
Following
the example of ALAMO, TEAnO - , Elettronica, Analisi nell'Opificio - was born
during the conferefence “Attenzione al potenziale – il gioco della letteratura”
(Beware of the potential- the game of literature) held in Florence in may 1991.
He was born at the age of 8, being the direct consequence of years of attempts
by a group of computer science professionals to use the digital computers
technology to find a sustainable match among creation, generation (or
re-creation) and recreation, the three basic keywords underlying the concept of
“Littérature potentielle”.
The
name is the direct consequence of the trial of finding some equivalent of
fighting at Fort Alamo; Barletta was felt too much against our French guests,
so that the only possible choice was another historical place: Teano. Suddenly
TEAnO as acronym.
The
works of TEAnO spread on three main directions:
·
Automatic generation of some “artistic” works;
·
Support to the creative generation of artistic works;
·
Studies and formalisations of creative and artistic
operations.
During
its life TEAnO produced many works in the above directions, and organised many
public events to issue results and to promote activities.
In
the following paragraph some examples of the production are reported, in order
to present the working activities of the group.
2.1 Automatic Generations
We list some examples of automatic generations, ordering them according to the growing complexity of the used models.
2.1.1. Pataphysics
In 1983 “Jarry e la Patafisica” was a big event organised in Milan, with a big exhibition and various happenings related; within the exhibition a computer room was set up by TEAnO, allowing the visitors to generate automatically various literature creations; the generations followed typical oulipians rules, such as S+n, antonymic substitutions, combinatorial generations, etc.[4] Among the generation of “pataphysic” proverbs, horoscopes or manipulation of well known poems, we report below the result of the generation of plots; each character was modelled as a finite state automaton, able to perform some actions, each of them able of influencing the other automata; the concurrent execution of them provided examples such as the following:
CAST:
The Cardinal
The Religion
God
The Sin
The Religion meets God.
In a dark night, the Cardinal
meets God.
Following dirty intrigues, The
Religion meets the Cardinal.
God meets the Sin.
The Sin meets the Religion.
Suddenly, the Cardinal gets
married to the Religion.
God kills the Cardinal. and
the Sin get married to the Religion.
The Sin kills itself.
The Religion get married to
God, and both were happy forever.
The interest
in the example is not, of course, in the result, but in the possibility to
model narrative plots, we will find also in further more sophisticated works.
2.1.2. Artusi’s recipes
Following
the master of ‘800 Italian cooking, Pellegrino Artusi, a classification of the
ingredients of recipes has been organised, and a kind of substitution of them
(following an S+n algorithm) inside of the text has been performed, obtaining
new creative recipes [5]. Among them, the followings, the former absolutely
foolish, the latter more than edible:
Snail
ice cream (Original: Lemon ice cream)
Courgette: 300 grams; brandy:
half a litre; snails: no.3
Use snails from gardens,
having them a better flavour in respect to the foreign ones. Boil the
courgettes in the brandy, add some piece of loin of the snails; boil for 10 minutes
with open pan. Let the syrup cool, and squash the nails inside. Put in the ice
cream machine. This quantity will be sufficient for six persons.
Apples
with Calvados (Original: cream cheese with coffee)
Mix slowly apples and
calvados. Add dowder of a courgette. Serve in a cup.
2.1.3. Tales
According
to the Propp’s work on Russian tales scripts, a generative algorithm has been
set up, with the proper contextual constraints, in order to produce original
magic folk tales; changes in the constrains allowed to change the culture of
the generated tales.
Many
works regarding the generation of folktales using the Propp's approach have
been performed within TEAnO since 1983. Such works range from simple generation
programs, strictly adopting the syntax proposed by Propp to generate Russian
folktales, to more sophisticated ones trying to tune this mechanism for
generating folktales belonging to the Italian folklore and the latter was used
to create scripts for some performances of traditional wooden puppet theatre.
The most interesting results are described in [6].
Nevertheless,
some activities have been carried out in order to try to find a more
sophisticated and general purpose "novel generation engine", which
led to the work of Claude Bremond.
As
known, the analysis developed by Propp is based on four axioms [7]:
a) The constant, stable elements of a folktale are the actions (Propp call them "functions") performed by the characters, regardless who is actually performing the action and how. They are the basic building blocks of the folktale.
b) The number
of such functions in a folktale is limited to 24.
c) The
sequence in which the functions follow each other is always the same (that is
each function is independent from the other)
d) All the
folktales have the same structure.
In
[8], Claude Bremond tries to make a step further in the work done by Propp by
contradicting the basic axioms and generalising the morphology proposed by
Propp to any kind of novels and not only folktales. Following Bremond, the new
four axioms for the morphology of a novel are:
a) The basic
building block is still the function as in Propp's work, but such functions are
not limited and are grouped into sequences to generate the whole novel.
b) A novel is
a sequence of "elementary sequences". Each elementary function
(processes) consists of three functions corresponding to the three mandatory
components of the elementary sequence:
1) a first
function (an event to be foreseen or a behaviour to be taken) that
"opens" the process related to the elementary sequence and provides
its rationale;
2) a second
function that implements the process in terms of actual events and/or
behaviours
3) a third
function that "closes" the process related to the elementary sequence
under the form of a result that has been (or has not been achieved)
c) With an
elementary sequence, a function does not necessarily force the subsequent one
(contradicting Propp) and each elementary action is described by a
"virtuality graph":
Fig 1: the virtual graph for the development of elementary sequences
d) Elementary
sequences can be combined in different ways to form "complex
sequences". Two examples of links between elementary sequences are the
following:
"bout à bout": the third
function (the closure) of an elementary sequence implements the first function
(the opening) of the subsequent elementary sequence;
"enclave": the functions
which an elementary sequence consists of are exploded in nested elementary
sequences which functions can be in turn exploded in other elementary sequences
and so on.
Generated
examples are the following, the former in a schematic structure, se second in a
more complete narrative form:
First example:
The dragon kidnapped the king father of
the beautiful princess.
The beautiful princess proclaimed a ban.
The valiant archer left secretly taking
the brave horse.
The valiant archer flow to the bounds of
the world with the brave horse.
The valiant archer set free the king from
the dragon.
The dragon prosecuted flying the valiant
archer.
The valiant archer hid himself.
The dragon failed the prove and killed
himself.
The
valiant archer became beautiful.
The valiant archer married the beautiful
princess.
Second example (a fragment)
After having tied and gagged the
kingdom’s wizard, the cyclops Cecyl stole the magic potion. Immediately after
the misfortune, heralds were sent to inform the near kingdoms about the fact,
calling for valiant knights.
The groom Huliev, after having heard the
complaints of the courts, decided autonomously to leave for removing the
misfortune.
The wizard Basil asked Huliev to share
all the wheat reserves with the poors of the kingdom for fighting the hunger,
promising his helps.
Basil and Huliev, together, decide to
share the objects according to their own needs, and the wizard prepared a magic
potion for Huliev, cooking barks of trees hundreds years old.
Near to the river Huliev found a small
boat, went on it and, very quickly, arrived to his destination.
Huliev, confident on his capabilities and
possibilities, invited the cyclops into the arena, where they should have been
running among lions, rushing off without being eaten. The longer in surviving
will have been the winner.
The run was very difficult, and logs on
the way was adding difficulties, but Huliev arrived to the goal as first.
While the cyclops went hiding himself in
the dungeons of the castle, Huliev followed him and imprisoned him till the end
of the his days.
…………….
Following
the approach of C. Bremond seems to be possible to synthesise novels having a
structure that is much more sophisticated if compared with the folktales
originating from Propp's axioms.
A work group is evaluating the generation of really interactive novels (or interacting scripts) using the logic proposed by Bremond to be performed on a multimedial computer workstations or on the (future) interactive TV. The interaction between the user and the generation program would be possible by means of an "emotional joystick", that is a sort of "crystal sphere" changing colour depending on the evolution of the interactive novel. A first rough prototype of such emotional joystick is currently being developed.
2.1.4. Antonymic music
Antonymic transformation refers to the substitution of some (or all the) words in a text with their “opposite” (white with black, earth with see or with sky, and so on). Which is the antonym od a tune? TEAnO assumed an algorithm to define it, providing both antonymic texts and tunes.
The "antonymic poetry" is another canonical constraint proposed by Oulipo[1]: it consists of the substitution of all the "meaningful" words in a given poetry with words having exactly the opposite meaning. Typically, since it is not possible to define an "absolute antonym" of a specific word[6], a context dependent substitution is performed and, in general, many different results can be achieved from the same text to manipulate.
For example, the following short passage from T. S. Eliot:
After the torchlight red on sweaty faces
After the frosty silence in the gardens
After the agony in stony places
The shouting and the crying
.....
can be turned in:
Before the darkness black under dry feet
Before the burning shouts in the rooms
Before the birth in grassy places
The silence and the laughter
.....
TEAnO has applied this techniques to texts (both prose and poetry) and to music as well, so to obtain the so called "antonymic music".
The generation of antonymic music is based on the following mechanism, which in some sense resemble the one adopted for antonymic poetry [9]: after having drawn a given melody on the pentagram, the "average value", a sort of "musical centre of gravity" of such melody is determine as a specific position on the pentagram and, then, all the notes of the melody are reflected above and under the average position. Finally, the antonymic melody is "normalised" (shifting of all the notes by a fixed number of position on the pentagram) in order to bring all the antonymised phrases to the same tonality.
The following is an example of an antonymic transformation [9]:
Fig 2: An antonymic transformation on a musical phrase (the centre of gravity is F)
This is essentially a mathematical/statistical approach, based on some statistical assumptions:
a) a note appear more frequently in a musical phrase is considered more important for the melody than another that is repeated less frequently and has the same intensity and duration;
b) a note having a longer duration is considered more important for the melody than another with less duration but same intensity and frequency
c) a note with a higher intensity is more important for the melody than another having the same frequency and duration but a lower intensity.
With the help of commercially available computer programs, a number of experiments have been realised on some well known songs, using the antonymic techniques for processing both music and lyrics: for example, "Le feuilles mortes" by J. Prevert and J. Kosma has been transformed into "Le racines vives" and "Blowin' in the wind" by B. Dylan into "Breathin' in the lull".
The work performed on antonymic music led to the idea of the design of a "musical workstation" for standard PC's, a sort of "virtual juke-box" able to play the same record in a hundred thousand billions of different ways [9]: for example, different and interesting results can be obtained by varying the parameters for the determination of the musical centre of gravity and/or changing completely the transformation algorithm.
This musical workstation allows to enter a melody and to apply to it one ore more pre-defined transformations; it is then possible to listen to the achieved results and to print the related scores.
One of the most promising evolution of these techniques for music manipulation could consist in thinking of such transformations as algebraic operations on a an "algebraic musical structure" (e. g. a group or a linear space) and deriving from it the basic properties of different classes of melodies.
A feasibility study is currently being carried out on this subject on the basis of the overall structure [9] described in the following figure:
Fig 3: an overall architecture for the musical workstation
2.1.5. Cocktails
A
model of cocktail has been built, and a knowledge base of the official standard
cocktails has been set up, to be used as a basis for an automatic generation of
new cocktails, through artificial intelligence “ by example” methods.
Among
the cocktails generated:
Montezuma
14 parts of
cachassa
1 part of peach
brandy
2 drops of
angostura
2 drops of lemon
juice
a cherry
mix with ice
2.1.6. Recipes
The
works on the cocktail has been refined, adding complexity of working
operations, in order to generate automatically cooking recipes. The works are
provided in this conference [10].
2.1.7. Metaphors
A
model of metaphors as partially overlapping knowledge representation frames has
been set up, able to generate automatically new metaphors, as a way to
stimulate new understandings. The model was able to generate metaphorical
phrases such as:
I cant digest your ideas.
I devoured your book.
I’m
thirsty of kwowledge.
2.2 Support to creative generation
2.2.1. Anagrammatic poetry support
Some of the Oplepo productions was related to the constructions of poems in which each verse was an anagram of a previously defined set of letters; a program has been developed to help the authors in this kind of work.
Following
in the footsteps of G. Perec (who
published in France the work "Ulcérations" composed exclusively of
anagrams of the French word ulcérations) and R. Campagnoli (who published in
Italy the work "Edulcoranti" composed exclusively of anagrams of the
Italian word edulcoranti), this editor helps the user in writing poetry or
prose consisting exclusively of various anagrams of a selected
word. It allows to:
a) define the selected word to anagram;
b) to define a fixed section within the
word that will not be changed during the anagram
generation
c) generate the complete list of all
existing anagram (taking into account the constraint due
to the fixed part of the word)
d)
browse and select the desired anagram
and edit it on the screen by adding blank spaces
This is a really a suitable subject for the exploitation of the combinatorial
power of computers as a powerful help for the humans in the creation of
original works. The English word "loutishness" can generate up to
39916800 anagrams but, keeping the part "theloss" fixed and
generating all the possible anagrams, we can obtain the response: "the
loss in us".
2.2.2 A tool for the production of definitional literature
A
piece of definitional literature can be generated from a given text by
substituting each "meaningful" word (such as nouns, adjectives and
some adverbs) with its definition contained in a given dictionary and iterating
this procedure for a number of times [1].
Using
this technique, it is possible, after three steps, to generate a text of
approximately 180 words starting from a sentence of six words. For example [1]:
STEP 0: The cat has drinks the milk.
STEP1: The carnivorous, domestic and
digitigrade mammal drinks the white sweet liquid produced by the females of the Mammalian.
STEP2: The meat eating, living in places
populated by humans, and walking on fingers mammal
...
This
interactive tool helps the user in defining one or more (specialised or general
purpose) dictionaries to be used in the step by step process, allows to define
the words to be substituted and the number of steps to apply.
2.2.3. Palindromic sonnet support
The
generation of palindromic phrases is a cumbersome work, but to build a palindromic
sonnet is an extreme trial; a program helping in it has been developed.
The
sonnet is a very complex maze based on a well known and very successful set of
constraints since a number of centuries. In the specific case of the palindrome
sonnet a new constraint has been added to the usual ones: the same text must be
readable starting from the first character of the first verse as well as
starting from the last character of the last verse.
The
palindrome constraint sets up a complex grid of correspondences and reflections
that have to be satisfied in order to accomplish the required rhymes and
metrics.
The
following is an example of the templates (generated by the editor) for a
palindrome sonnet accomplishing two different rhyme structures[7],
obtained just by enter the first verse, which can give an idea of the
complexity of the composition:
|
ABBA – ABBA - CDC - DCD |
ABAC - ABAD - ABE - ABE |
|
|
|
|
was alone in the sun |
was alone in the sun |
|
*[8] |
* |
|
was* |
nus*sun |
|
*sun |
was* |
|
|
|
|
was*sun |
*sun |
|
* |
nus* |
|
nus* |
*sun |
|
*sun |
nus* |
|
|
|
|
* |
*sun |
|
nus*saw |
nus* |
|
nus* |
*saw |
|
|
|
|
*saw |
nus*saw |
|
* |
* |
|
nusehtnienolasaw |
nusehtnienolasaw |
After that, the user can proceed by adding spaces
for separating words (e. g. in the last verse) and entering the missing part of
the sonnet: any entered character will be reflected by the program in the
appropriate positions (depending on the specific rhyme structure) to ensure the
validity of all constraints. During the sonnet composition, the "*"
characters indicates to the user where it is possible to add new character
without violating the constraints.
Furthermore,
the editor provides a number of functionalities for helping the writer to select
the desired rhyme structure from a dozen of pre-defined schemes (e.g.
ABAB-BABA-CDC-DCD or ABBA-ABAB-CDE-CDE or ABAB-ABBA-CDE-DEC), to define new
rhyme structures and to choose the number of characters taken into account for
each rhyme.
2.3 Studies on formalization
2.3.1. Scripts formalisations and algebraic models of Telenovelas
Analysis
of movies script structures has been executed through Petri Nets, allowing a
formal distinction between plot and narration; trials in changing the narration
on the same plot has been carried on. More, some algebraic model of telenovelas
have been carried on toward the generation of “successful” movies.
As well known, telenovelas present a very established and peculiar structure that can be described by means of few simple rules:
a) A fixed set of characters
b) A fixed set of closed places in which
all the actions are developed
c) A number of related (but independent)
narrative processes which evolve in parallel
This
consideration can lead to the definition of an algebraic structure and a set of
operation based on it, to be used for generating "canonical"
telenovelas:
T = {t,P,S}
where t
represents
the time, P
is
the set of all involved characters and S is the set of all
possible closed places.
This
first description provides only a static view of the set of telenovelas; to
describe the development of all actions, a structure called "script"
has been defined:
x = <T, L, F>
where the
first item, T, is a static description of telenovelas, the second, L, is a function (functions are typically the
elements describing transformations of other elements) having specific
properties and called "placement"; the third component, F, is another function, called "focus".
During the
development of the script, the various parallel stories evolve in parallel,
characters interact each other and move from one place to another. At any time,
the placement function has the goal of describing the distribution of
characters over the different places and their transitions from place to place;
on the basis of the so determined distribution, it is possible to apply the
focus function in order to find out the most interesting place (that is the
place in which the performed actions have the highest narrative tension) in
order to focus the script on the actions happening in such place.
One of the
most interesting implications of a so defined structure, describing an abstract
telenovela, is that it allows the definition of a sort of telenovela engine,
consisting of "elementary telenovelas" that can be composed in order
to obtain a whole family of "canonical telenovelas".
2.3.2. Recipes
A formal model of cooking recipes has been set up, according to a definition of ingredients and working steps, introducing a metric and the concept of “distance”, allowing an evaluation of a recipe copyright violation [11].
2.3.3. Aesthetical translation
Is
it possible to translate an artistic composition into another of different
nature (e.g. a poem into a painting) preserving the “aesthetical emotion” of the
observer? This is the goal of this research line. The main lines followed are
resumed here.
An
art A can be characterised by a set of features FAi.
A={FAi} , i=1..NFA
Each
feature can assume a set of enumerated predefined values VAij.
FAi: {VAij} ,i=1..NFA ,j=1..Mai
A
manufact O belonging to the art A is characterised by a set of values:
O(A)={vaij} vaij Î Vaij
Examples:
Ex.1.
PAINTING={form, technique, size,
composition, colours, ..., subject}
form={square, horizontally rectangular,
vertically rectangular, oval, ..., round}
technique={oil, water-colour, ...,
composite}
...
subject={abstract, pastoral, town, sea,
..., feminine portrait, masculine portrait}
La Gioconda={vertically rectangular, oil,
small, ..., feminine portrait}
Ex.2
POETRY={schema, verses, phonemes, ...,
subject}
schema={sonnet, ... , prose}
phonemes={prevalent a, prevalent e,
prevalent i, ..., liquids, ...}
...
L'infinito={sonnet, endecasyllabs, ...}
To
simplify computations (as it will be clear in the following, we can represent a
manufact as an array of Boolean, using as indexes all the possible enumerative
values:
O(A)={vai} i=1..S MAk, k = NFA
In
the example above, the result would be:
La
Gioconda={0,0,1,0,...,0;1,0,...,0;...;...;0,0,0,0,...,1,0}
Given
two artistic disciplines A and B, relationships can be provided between the
related features FA and FB, to map the artistic discipline A on B (in general
the relationships are not reversible).
We
can assume that each FAi will contribute to some extent to different features
FB's, and, more in general, each vaij will contribute to different vbij.
We
can define a mapping matrix MABij which elements are defined as:
mabij i = Sum MAk,j = Sum MBh
k = NFA, h = NFB
and
represents a weight (e.g. on a range 0..1) of how much a value from A
influences values on B.
Example:
A=painting
B=poetry
MAB:
|
|
sonnet |
prose |
... |
prev.a |
prev.e |
prev.i |
... |
liquids |
... |
... |
|
square |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
horiz.rect. |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
vert.rect. |
1.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
oval |
0.7 |
0.0 |
0.0 |
0.0 |
0.0 |
0.2 |
|
|
|
|
|
... |
|
|
|
|
|
|
|
|
|
|
|
round |
0.2 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
oil |
0.0
|
0.0 |
0.0 |
0.0 |
0.0 |
0.0
|
|
|
|
|
|
watercolor |
0.2 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
... |
|
|
|
|
|
|
|
|
|
|
|
composite |
0.0 |
0.2 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
… |
|
|
|
|
|
|
|
|
|
|
|
abstract |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
pastoral |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
town |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0
|
|
|
|
|
|
sea |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
... |
|
|
|
|
|
|
|
|
|
|
|
femin.port. |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
|
masc.port. |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
0.0 |
|
|
|
|
where
sonnet, prose, ... refer to the poetic structure; prevalence of vowel a, e, ...
to the sound of the vowels; liquids, ... to other letters; and
correspondingly, square, horizontal rectangle, .. to the shape of the painting;
oil, watercolour, ... to the technique; abstract, pastoral, town, ... to the
subject, and so on.
Given
the opera O(A) composed in the art A, it is possible to translate it in the art
B:
O'(B) = maxf(O(A) * MAB)
where
the operation * represents the usual multiplication of an array by a matrix,
rows by columns and the operation maxf indicates that, for each of the NB features
of O'(B), the maximum value of O'(B) within the single feature is set to 1, and
all the others to 0.
As an
example, the transformation of a soonnet into a painting, executed following
the suggestions obtained through the method.
