Irenebrination: Notes on Architecture, Art, Fashion, Fashion Law, Science & Technology

Will Artificial Intelligence Be Used in Future as an Art Authentication Tool?

We do know that Artificial Intelligence (AI) can be an asset in the health field: for example, there have been studies to see if AI, trained on databases of mammograms collected from clinics and academic institutions, may help detecting breast cancer in early stages.

A research published today on the journal Heritage Science (Ugail, H., Stork, D.G., Edwards, H. et al. Deep transfer learning for visual analysis and attribution of paintings by Raphael. Herit Sci 11, 268 (2023)) analyses instead another application for AI in the field of artwork authentication.

We all know that art attribution and authentication pose intricate challenges for experts as the processes to analyse an artwork and attribute it to a specific artist are long and involve studies on provenance, materials, iconography and conditions, not to mention analysis of artists' careers, derivative works, and connoisseurship.

The paper focuses solely on connoisseurship (i.e. visual analysis of the composition, style, brush strokes, shading, etc.) and introduces a computational tool aimed at assisting scholars in analyzing and authenticating a restricted class of paintings, focusing, as suggested in the title of the study, on artworks by Italian painter and architect Raffaello Sanzio da Urbino, better known as Raphael.

Machine learning in art analysis is seen with suspicion among the scholarly art community, mainly because of the scarcity of high-quality data necessary for training robust multi-class classification models. Yet the research team, led by Hassan Ugail, professor of visual computing at the University of Bradford, found a way to avoid this by constructing smaller, artist-specific models for more accurate analysis and authentication of individual paintings.

For their study on Raphael, the team mainly employed a pre-trained ResNet50 (Residual Network50) deep neural network for detailed feature extraction, combining it with an SVM (Support Vector Machine) binary classifier for authentication. Additionally, edge detection and analysis algorithms, crucial for capturing Raphael's artistic style, including brushwork signatures, were integrated as authentication tools.

This combined and hybrid approach enhanced the model's accuracy: to classify a painting as likely by Raphael, the artwork had indeed to meet both the edge feature threshold and the SVM model's probability threshold. This composite methodology should significantly improve classification accuracy, something that was put in doubt in an analysis carried out a few months ago. In that case, two AI research groups, one from the University of Bradford/the University of Nottingham and one from a private company named Art Recognition, aimed to authenticate the de Brécy Tondo painting. Using facial recognition and brushstroke analysis, they yielded conflicting results: while the former group claimed with absolute certainty that the work is by Raphael, the latter reported an 85% certainty that it is not. The outcomes challenged the perceived infallibility of AI, while art experts criticised the results highlighting that the painting was a Victorian copy of Raphael's the Sistine Madonna.

The essay published today highlights how the system was trained with a collection of authenticated Raphael paintings and a mixed array of paintings from artists like Rembrandt, Peter Lely, van Dyck (labelled "Not Raphael"). The first results looked promising as the machine learning approach achieved a 98% accuracy in image-based classification tasks during validation with a test set of well-known and authentic Raphael paintings.

Artificial Intelligence correctly identified "The Sistine Madonna", the "Haddo Madonna" (initially attributed to Innocenzo Francucci da Imola but suspected, and later confirmed, to be a Raphael) and "Lo Sposalizio or The Marriage of the Virgin" as painted by Raphael, while the portrait of Diana Kirke, Late Countess of Oxford as painted by another artist (Peter Lely).

The most extensive research was carried out on the Madonna of the Rose (Madonna della rosa, 1518-1520, Museo del Prado, Madrid). The team's algorithm suggested a probability of just 0.57 that the entire painting was the work of Raphael. Individual sections of the artwork (mainly the heads of the main figures in the painting) were then analysed, leading to doubts and questions about whether Raphael painted the face of Joseph in this artwork.

It is worth remembering that some art historians previously questioned the full attribution of this painting to Raphael alone, suggesting that his associate, Giulio Romano, might have had a hand in it. The machine learning analysis done with the help of AI supports this academic theory.

The methodology developed in this research can be readily modified and adapted for authenticating paintings by different artists, benefiting art historians and collectors.

As machine learning and image processing technologies advance, this method could therefore become part of a comprehensive toolkit for artwork analysis, working alongside existing methods such as scholarly analysis and advanced imaging techniques like spectroscopic imaging, and dating techniques.

Yet the results confirm the obvious, reassuring art experts that their jobs will not be replaced any time soon: just like AI can be an asset for a radiologist or a doctor, providing faster and more accurate diagnostic analyses, the pre-trained ResNet50 and the SVM classifier may lead to interesting results when used judiciously by art experts, historians and collectors familiar with the context, as these tools can enhance traditional visual analysis in art authentication challenges and accelerate research processes, operating in conjunction with other methods of analysis.  

So far authentication processes employing AI were only applied to the art field, but there is another discipline in which authentication has become crucial - fashion. Yet, authenticating garments and accessories may be trickier for Artificial Intelligence as authenticity in these cases also passes through a close-up tactile analysis. That said, maybe in future we will also have AI tools that may help fashion historians in their work. 

In the meantime, those who want to discover more about the technical aspects of the research published today can access the image datasets used to train and test the machine learning model and check out the code behind it at this link.

Threads of Transformation: New Worlds from Traditional Fragments

In the introduction to the book Traditions of the Thompson River Indians of British Columbia (1898, Download Traditions of the Thompson River Indians), Franz Boas states, "It would seem that mythological worlds have been built up only to be shattered again, and that new worlds were built from the fragments."

Boas wasn't obviously referring to fashion, but we can move from this concept of shattered worlds and new visions rising from the remaining fragments to create modern fashion mythologies evolving from traditions.

In the history of contemporary fashion, using traditional patterns and motifs as starting points for collections or for specific garments often caused terrible faux pas that led to cases of appropriation. Yet it is possible to update traditional moods and inspirations and reinterpret them in modern keys, especially in those cases in which traditions seem to have been abandoned.

Let's consider the case of traditional regional costumes in Italy: in a previous post we looked at the attire of rural women in fascist times, highlighting how wearing their local and regional costumes during parades, they weren't promoting their own individuality against the regime, but were instead being manipulated into showcasing the nation's traditions, hinting therefore at the grandeur of the regime.

As the decades passed regional costumes were gradually forgotten in Italy: nowadays you may just see them in ethnographic museums and at folk exhibitions, or maybe donned by members of folkloristic dance groups or musicians, or at times worn during food and crafts exhibitions by young women to promote the delicacies and handmade products of the different regions.

By the early '80s, regional costumes were already nearly forgotten, but it was still possible to find in some shops nostalgic postcards that celebrated them.

Each of these postcards, almost resembling fashion sketches, featured a young lady in the traditional attire of an Italian region. The dress was decorated by skilled artisans with embroideries that enriched the regional costume printed on the postcard. The embroideries added therefore a three-dimensional tactile quality to the postcards.

This sort of anthropological souvenir - popular in in Italy, but also in France, Spain and Portugal - had already started disappearing in the early '80s (nowadays you may find a few examples of these postcards on eBay with prices varying, depending on the conditions and the number of postcards).

Most of the subjects on these postcards related to regional or national costumes, but there were also occasional religious themes. These postcards, usually sold in pilgrimage sites like Fatima, featured religious icons and figures like the Virgin Mary highlighted with the usual embroideries.

My grandfather used to collect stamps and postcards and also had a few of these postcards: I wouldn't play with them, but I would admire them and cherish the details like the tiny flowers on the skirts of the young women portrayed on the postcards or the delicate stitches used to add lace-like shawls and veils. In my eyes, the embroidered elements on the postcards turned them into kitsch yet mesmerizing souvenirs.

The postcards came to mind while looking at Antonio Marras' Pre-Fall 2024 collection. Marras blended in this collection two different moods – Sardinian patterns and costumes with Native Americans references.

Marras explored the points in common between the two cultures, emphasizing their shared relationship with and respect for nature. The collection featured therefore bucolic inspirations alongside elegant pieces, such as tailored coats and jackets with pleated panels, and clashes of plaids and florals, showcasing Marras' skills through textured and patterned juxtapositions.

Blankets served as a canvas for Marras that created through quilt techniques abstract motifs and figures of animals. The blanket is supposed to turn in this case into a wearable glossary, embodying both Sardinian and Native American elements, while metaphorically hinting at a displaced existence amid wars and environmental disasters.

Talking about embroideries, the collection also included a jacket that could be considered as a wearable embroidery sampler as it featured on one side a selection of traditional embroidered elements, and, on the other, the embroidery of a woman dressed in the Sardinian regional costume that almost evokes the vintage allure of costume postcards (see also last vintage postcard in this post for another version of the Sardinian costume).

This jacket aligns with Boas' concept of shattering mythologies: in this case the designer interpreted regional embroideries, patterns and costumes, building something new from their fragments and recreating in this way a transformative process of traditional mythologies.

Fashion Fusion: Exploring Interdisciplinary Inspirations in Balmain's Pre-Fall 2024 Collection

In the previous post we have seen how a broader understanding of advertising campaigns might open the door to more expansive connections. This principle extends to all domains, but, within the realm of fashion, it enables unique connections filtered through diverse disciplines and influences.

Take, for example, Balmain's Pre-Fall 2024 collection: Creative Director Olivier Rousteing last visited Miami more than 10 years ago, but the memories lingered and this collection moves from that trip.

Characterized by couture silhouettes, the collection seems divided into two sections, one revolving around a monochromatic palette, a black and white dichotomy, and another featuring glittering flamingoes and palm-tree embellishments.

The strong shoulders in the first section may point at famous crime drama Miami Vice, but the rigid geometries of the jackets seemed borrowed from Klaus Nomi's iconic black and white costume that made him famous as he sang "Lightning Strikes". The reference wasn't too absurd considering also the black and white Harlequin patterns and the fact that Klaus Nomi looked theatrical, almost like a character out of the Commedia dell'Arte.

We blame Artificial Intelligence for stealing and collaging other artist's works, but in the giant blender of fashion, everything is constantly remixed as also proved by a striped evening gown in this collection that seemed to be a clash between Marc Jacobs' Spring/Summer 2013 black and white optical dress, Dior's Haute Couture Spring/Summer 2018 dress (View this photo) and the oblique striped attire of the three girls in the frescoes of Bolzano's Church of San Domenico (Dominican Church).

Some elements for this collection were actually taken from Rousteing's own S/S 2013 collection for Balmain with its South Beach palette, but other colors and textures may have had an arty derivation.

Flamingoes appeared as prints or as three-dimensional sequined embellishments on jackets: the gleaming, square sequins pointed at different (yet similar, if you think about the material) types of flamingoes, such as Gino Marotta's acrylic sculptures of this alluring bird.

Actually Marotta's plexiglass sculptures in neon colors seemed to be evoked also by the motifs embroidered or sequined on bomber jackets and more formal designs included in the menswear section of this collection.

The palette for these motifs was vibrant and almost replicated the solarized lighting effects and bright shades in Carmelo Bene's Salome, with sets designed by Marotta.

More subdued pastels pointed instead at Miami's famous Art Deco architecture, with details of the buildings being evoked in sculptural lapels.

So, from a music inspiration we went through Op Art and fluorescent plexiglass art and finished with architecture. Doing these flights of fancy is a fun and intriguing exercise that helps us looking at a fashion collection from a variety of points of views and perspectives, considering borrowed elements (Klaus Nomi's costume and Marc Jacobs/Dior's dresses in this case), filtered through art and architecture references (Gino Marotta and Miami Art Deco).

Now it's your turn: explore a fashion collection, uncover the designer's inspirations, and then challenge yourself to establish broader connections with other disciplines.

Consumers Perceptions in the Social Media Era & Fashion Campaign Faux Pas from Zara Back to Versace

Among the fashion-related news this December there was the story of the controversial Zara ad campaign.

Shot by Tim Walker, the images, posted on Zara's official social media channels, promoted six new jackets and depicted model Kristen McMenamy among rubble and plaster, posing next to mannequins and statues (the idea was showing craftsmanship in an artistic context) or standing next to the sort of wooden boxes traditionally used to ship artworks. In one image the model was portrayed carrying a mannequin wrapped in a white sheet.

The images led to a notable backlash on social media platforms, especially on Instagram, TikTok, and X, where people started drawing parallels to news reports from the Israel-Gaza conflict depicting bodies of Palestinians covered in white sheets.

Social media users claimed the advertisements mocked the destruction of the war and also featured a cardboard cutout that resembled an inverted map of Palestine. As a consequence, a call to boycott Zara started with the hashtag #boycottzara gainining considerable traction, amassing around 20 million views on TikTok.

Inditex, Zara's parent company, confirmed that the campaign was conceptualized in July, and the photo shoot took place in September, so months prior to the recent conflict that started in the first week of October. But, as the backlash grew, Zara expressed regret for any offense caused and retired the campaign.

"Unfortunately, some customers felt offended by these images, which have now been removed, and saw in them something far from what was intended when they were created," the company stated on social media. “Zara regrets that misunderstanding and we reaffirm our deep respect towards everyone." 

Zara has actually got a history of encountering calls for boycotts related to the Israeli-Palestinian conflict: in 2021 Zara's chief designer, Vanessa Perilman, made offensive remarks about Palestinians on social media. The company faced condemnation, and Perilman received criticism for a direct message to Palestinian model Qaher Harhash on Instagram ("Maybe if your people were educated then they wouldn’t blow up the hospitals and schools that Israel helped to pay for in Gaza," Perilman said to Harhash. "Israelis don't teach children to hate nor throw stones at soldiers as your people do") in response to a pro-Palestinian post made by Harhash a month after an 11-day conflict between Israel and Hamas, which led to over 200 deaths.

In 2022 Zara came under fire after Joey Schwebel, chair of Zara's Israel franchisee Trimera, hosted a campaign event for right-wing ultranationalist politician Itamar Ben-Gvir. This led to Arab Israelis burning Zara's clothes and calling for a boycott.

In this case, before releasing the campaign Zara may have wondered if the images may have triggered the wrong connections in people (perhaps omitting the main image of the model carrying the statue wrapped in a white sheet could have reshifted the attention to the sculpture studio moods of the campaign).

That said interpreting a campaign is subjective, so each of us may project on an image a particular feeling or idea we may have, but at times, as commentators on social media, we lack the knowledge to make connections with other campaigns or images from decades ago. 

Versace Spring/Summer 1985 campaign photographed by David Bailey featured for example female models in Versace swimsuits or evening wear and male models wrapped up in coarse linen shrouds held together by thick ropes. As you may guess, this wasn't a great idea.

The male models had a sort of mummy-meets-corpse look and at the time the campaign raised a few eyebrows as it looked rather disturbing and the images left you cold: were these men dead? Were they annoying yuppies (check out the man in the suit in the last image in this post) killed for, well, being annoying yuppies; was this an erotic game gone wrong or a dominatrix going down the necrophiliac path? We will never know. What we do know, though, is that in this case the wrapped up bodies weren't supposed to be sculptures, but, well, real men.

In conclusion, both companies and social media users should maybe apply the basic principle "think before acting (or reacting)" - as a company you may plan and shoot a winter campaign during the summer, but unexpected occurrences may alter the appropriateness of a campaign.

As for users, before joining the chorus of accusers, strive for greater knowledge. Keep in mind that the history of fashion has witnessed more controversial instances than this Zara campaign (we may actually highlight that this campaign appeared a bit bland as boxes with statues or models as statues coming out of wooden boxes isn't a groundbreaking idea in fashion and Tim Walker has often turned to this solution in previous photoshoots - think about his 2018 photoshoot entitled "Handle with Care").

Nevertheless, this incident underscores the effectiveness of calling for boycotts: since the beginning of the Israel-Gaza conflict in October, the boycott movement against a number of multinational companies for their support of and ties to Israel has indeed expanded significantly.

A Tragic Word of the Year With Some Human Rights & Architectural Implications: Domicide

"Artificial Intelligence" and "Hallucinate", words of the year for the Collins and Cambridge Dictionaries were strictly linked with modern technologies. Yet, maybe, we should ponder a bit about other words that didn't make it into the official 2023 glossaries, but that are extremely relevant at the moment. Among them there is a term linked with human rights and architecture – domicide.

Mentioned in a report (Download UN_AdequateHousing) published in July 2022 compiled by Balakrishnan Rajagopal, UN Special Rapporteur on the right to adequate housing, the term indicates the destruction of houses of civilian inhabitants during wars and conflicts.

In the report entitled "Adequate housing as a component of the right to an adequate standard of living, and the right to non-discrimination in this context," the UN Rapporteur analyzes the ongoing and unprecedented violations of the right to adequate housing during and after violent conflicts.

Domicides have happened throughout history: the report highlights for example how during World War II, German Armed Forces employed the tactic of demolishing homes and civilian infrastructure, leading to notable war crime convictions at the Nuremberg Trials.

The Chief of the Operations Staff of the High Command of the German Armed Forces, Alfred Jodl, was found guilty of war crimes and crimes against humanity for, inter alia, having ordered the evacuation of all persons in northern Norway and for the burning of their houses, while Field Marshal Erich von Manstein was convicted for ordering to deport civilian inhabitants of the territories occupied by the German Armed Forces under his command in Eastern Europe and to destroy their houses as well as all other objects of economic value.

The report also remembers the atomic bombings in Hiroshima and Nagasaki and cites systematic housing destruction in global decolonization movements.

The Commission on Human Rights, in its 1993 resolution, declared forced eviction a gross violation of human rights, emphasizing the right to adequate housing. Governments were urged to take immediate measures to eliminate forced eviction. This obligation extends to situations of internal strife, violent conflict, and occupation.

The Guiding Principles on Internal Displacement (E/CN.4/1998/53/Add.2) reinforce the right to be protected against arbitrary displacement, emphasizing adherence to international law to prevent conditions leading to displacement.

Similarly, principle 5 of the principles on housing and property restitution for refugees and displaced persons (E/CN.4/Sub.2/2005/17) prohibits the "forced eviction, demolition of houses and destruction of agricultural areas and the arbitrary confiscation or expropriation of land as a punitive measure or as a means or method of war."

Yet the deliberate destruction of homes during violent conflicts continues to the current day, despite existing human rights and humanitarian laws prohibiting arbitrary destruction of housing and displacement.

The list of conflicts that have deprived civilians of their houses is long and comprises Syria, Libya, Myanmar and Ukraine. After Russia invaded Ukraine, in Mariupol alone, a city of 400,000 inhabitants, 90 per cent of all apartment buildings were damaged or destroyed.

Rajagopal reminds in the report how vulnerable people - children, women, people with disabilities, the elderly and members of the LGBTQ+ community - are affected not only differently, but often more detrimentally than others.

Elderly people, for example, are unable to leave conflict zones owing to mobility or sensory impairments and losing their homes from them has also got a psychological impact as they see something they put all their energies in being destroyed. The report mentions the case of Jawad Mahdi, who, at 68 years of age in 2021, lost his home in Gaza City to Israeli airstrikes, and compared the loss of his home and "all [those] years of hard work" to "someone ripping your heart out and throwing it."

"I have witnessed how in just a few seconds a home – the culmination of a life-long effort and the pride of entire families – can be wiped out and turned to rubble," the UN Special Rapporteur states in a press release about the report. "Destroyed is not only a home. Destroyed are the savings of entire families. Destroyed are memories and the comfort of belonging. Along with this comes a social and psychological trauma that is difficult to describe or imagine."

As of 2022, more than 100 million people worldwide have been forcibly displaced as a result of persecution, conflict, violence, human rights violations or events seriously disturbing public order. But now numbers have increased since the Israeli-Gaza war started.

On October 7th, Hamas militants initiated an unprecedented attack on Israel from the Gaza Strip, killing 1,200 people and taking 240 hostages. In response, the Israeli military conducted air strikes on Gaza and launched a ground offensive. According to the United Nations reports, so far, close to two million people were forced to leave their homes in Gaza, with hundreds of thousands facing challenges in securing adequate food and water for survival. 

Satellite imagery indicates damage to 98,000 buildings in Gaza, including the main Palestinian court, the Palestinian Legislative Council complex, over 300 education facilities and over 150 places of worship, while 26 out of 35 hospitals are not functioning.  The current devastation raises concerns about the concept of domicide with commentators wondering whether the mass destruction of buildings results from the search for Hamas or a covert plan to expel Palestinians, making the territory of Gaza uninhabitable. 

The concept of "domicide" is increasingly accepted but not recognized as a distinct crime against humanity under international law, while Rajagopal states it should be recognised as a crime.

A Professor of Law and Development at the Department of Urban Studies and Planning at the Massachusetts Institute of Technology (MIT) in Boston, Rajagopal actually spotted a gap in international law that may be allowing the destruction in Gaza: the Rome statute protects civilian homes in conflicts between states but does not categorize it as a crime against humanity in conflicts within a state or involving non-state actors. In this case, Israel may argue that the conflict is not international because it does not recognize Palestine as a state.

Rajagopal's report emphasizes the need for urgent international action to address and prevent these violations, calling attention to the lack of reparation, limited options for voluntary return, and the absence of just and durable solutions and stresses the necessity for further measures to ensure justice and respect for human rights in post-conflict and reconstruction settings.

Hopefully, domicide will be tackled also at future architecture events such as the Venice Biennale and we will see the UN involving in the discourse also architects such as Shigeru Ban, who have developed humanitarian projects and temporary structures and housing for refugees and victims of disasters across the world.

Though never mentioned in fashion, domicide actually appeared in a fashion collection: over twenty years ago, Hussein Chalayan indirectly referenced it in his "After Words" (A/W 2000-01) collection. The latter was inspired by the news from Kosovo about people fleeing their houses in a rush during the war and carrying their possessions on their backs.

Models during the show removed the covers from chairs and turned them into dresses, then the last model stepped into the middle of a round 1950s-style coffee table, lifted it up and transformed it into a skirt. In this way the refugee theme was linked to the idea of hiding and camouflaging valuable possessions or carrying them along in flight.

The Joy of Being Alive, The Misery Created by Power: Toni Negri (1933 - 2023)

As announced by his daughter Anna with an Instagram post, Italian philosopher, political scientist, activist, and politician Toni Negri died today.

For his links with radical left-wing organizations like Potere Operaio and Autonomia Organizzata, the current Italian Minister of Culture, Gennaro Sangiuliano - who is trying to reboot fascist hegemony in culture through exhibitions that pay respect to a Tolkien-inspired fantasyland imagined by right-wing powers - commenting upon his death, called Negri "a bad teacher".

In response to this statement, I'm republishing here a condensed version of an interview I did with Negri in 2002 for an American publication. In the interview Negri spoke about the concepts of Empire, labor, multitude, war and immigration, as tackled by him and Michael Hardt in the volume Empire.

The interview was actually subtitled "A lesson by Toni Negri" as it was conceived as a mini-essay about the principles analyzed in Empire. Readers of this site may also remember how I discussed the theme of a fashion empire in a previous post that was also inspired by Negri's writings. Negri is survived by his partner, the French philosopher Judith Revel, and by his children from previous relationships Anna, Francesco and Nina.

In the Spires of the Empire: A Lesson by Toni Negri

by Anna Battista (2002)

"He's in the foyer, doing an interview," a guy whispers to his friend beside me as he hastily takes a seat among the audience at the Michetti Theatre in Pescara, Italy. Shortly after, the man we've all been anticipating, quietly enters the theatre room, avoiding to draw attention. Behind me, a lady comments softly to a friend, "He looks older... after what he went through..." While Toni Negri's life has been marked by numerous dramatic events, tonight, those years seem lost in the folds of time, finally left behind, distant eons away. Interestingly, he harbors no resentment towards anyone, as he later asserts, "Resentment is the worst quality of a free man." Tonight, Toni Negri isn't dwelling on his past; he's here to discuss his latest book, co-authored with Professor Michael Hardt, titled Empire.

A step into the past

In the '60s, Antonio Negri was teaching State Theory at the University of Padua, in Italy. His works mainly focused on themes like labour theory and work. During these years he also joined the Marxist journal "Quaderni Rossi" and started taking part in Potere Operaio (Workers' Power), a radical left-wing Italian political group. When the latter split up, Negri moved with another one, Autonomia Organizzata (Workers' Autonomy). Italy was living then the so-called "anni di piombo" ("lead years"), an expression that defined the '70s and the violence that spawned terrorist groups like the Red Brigades. On 7th April 1979, Negri, accused of being the head of various terrorist organizations, was arrested in Milan. In the four years that followed, Negri was moved from one prison to another and was also accused of being the organizer of the murder of politician Aldo Moro. In 1983, when Negri's trial took place, the accusation of having been part of terrorist organisations was dropped and Negri was condemned on the substance of his writings.

On 25th and 26th June of the same year, Negri ran as candidate for the Italian Radical Party and, after obtaining parliamentary immunity, he came out of prison. Months passed and the Chamber of Deputies requested to withdraw parliamentary immunity from Negri. In September 1983, instead of going back to prison, Negri escaped to France where he spent fourteen years, teaching political science at the Université de Paris VIII (Saint Denis), studying and writing.

In 1997, Negri went back to Italy, where he entered Rebibbia prison in Rome to serve his sentence. Thousands of articles were published on international newspapers about Negri's case. Amnesty International studied his case and the trials that followed, that were considered unfair. In the meantime, Negri continued writing for various magazines about politics and focusing on developing with Michael Hardt a trilogy that started with Empire.

A step into the future

Marx Beyond Marx (Pluto Press, 1991), The Savage Anomaly (1991), Insurgencies: Constituent Power and the Modern State (1999; both published by the University of Minnesota Press) are listed among Antonio Negri's most studied works, but there is one essay which is at present considered as being the best one - Empire (published by Harvard University Press). The book was finished in June 1997, before Negri went back to Italy to be arrested. The last page of the book - in which the authors state that we are living in the same situation St. Francis of Assisi was in, divided between the joy of being alive and the misery created by power - was written then, and yet, the volume deals with something, an evolutionary process, which is happening right now.

Negri and Hardt, who previously published together the volume Labor of Dionysus (1994, University of Minnesota Press), theorise in Empire the death of the nation state and the rising of a new power, the result of the various capitalistic processes occurred throughout history. The book, divided in four parts, tackles issues such as the new Empire, the concepts of labor and multitude, immigration, wars and the final destruction of the Empire. Hardt and Negri do not tend to identify the Empire with a single nation, but they claim that the proper Empire is a sort of place without boundaries, which is forming right now, through a thick network of new laws and regulations. Though it is still forming, the new power is already going through a deep crisis, analysed in the last part of the book, in which, very optimistically, the two authors confide into the power of the multitude to destroy the Empire.

"The book I co-wrote with Hardt is rather old, we finished compiling it around 1997-98," Negri starts remembering, explaining, "The Empire forms when the nation state and the national space are incapable of taking proper control. The workers' struggles in the capitalistic countries, the anti-colonial struggles in the ex-colonial countries and the struggles for freedom in the socialist countries, have little by little transformed the space controlled by the modern sovereignty. As a first consequence, the capital had to establish new wider structures to contain these groups."

According to Negri, the formation of an "Empire" is in progress through attempts at regulation and rule creation. He emphasizes that sovereignty is shifting away from individual nations, raising questions about who exercises it and within what limits. Negri acknowledges the challenges of the present, asserting that reverting to the nation-state as a defense against the emerging global order is not viable due to the crisis faced by nation-states. He highlights the three characteristics of sovereignty: organizing an army, coining currency, and determining local culture. "But all these things aren't possible right now in a modern nation state," Negri states. "At present, only the Empire can organise warfare, only the world market influences currency minting and, from the cultural point of view, a nation can't really have a cultural monopoly." Negri underscores the significant and consequential nature of the Empire's formation process, describing it as a complex and impactful journey.

Among the various consequences of the process which is forming the Empire, Negri underlines one in particular, the concept of labor.

Negri emphasizes a significant consequence of the ongoing process shaping the Empire, focusing particularly on the concept of labor. Labor has undergone a transformation into the immaterial, a characteristic amplified by the third revolution that has supplied the technological means to develop labor in these forms.

"Today, for those engaged in intellectual and immaterial work, the distinction between the working day and the temporal day disappears," Negri says. "Labor evolves into a vital activity, embodying the intellectual or linguistic capacity to create new values. Now, individuals can independently generate a product. In the past, during the transition from capitalism to the era of large-scale industry, workers were provided with tools and instruments by the master or owner when they went to work. However, the dynamics have shifted: work now involves expressing intellectual and linguistic abilities, and individuals own the tools and instruments for their work. This elevates the virtues of each individual, marking the end of the class concept."

Previously, the concept of class and mass had a subordinate value, it was indeed the capital that formed a class. Today, the situation has changed and the capitalist no longer provides the tools. "Individuals possess the instruments for work because they are alive, engage in activities, cooperate with others, and produce language. Language, in turn, determines production and valorization," Negri explains, continuing, "while sovereignty expands, intellectual strength grows. This transformation reshapes the anthropological structure. When labor aligns with the individual, everything undergoes modification. The attempt to exert sovereignty faces a formidable force comprised of individuals possessing the instruments for work." 

In "Empire," a new concept, that of the "multitude," is also formulated, representing all individual subjects. "The multitude is a process born out of singularity, a clear inclination to come together and find unity," explains Negri. "These singularities first and foremost constitute a multiplicity."

Negri intends the term "people" as referring to the group delegating authority to the state, while with "multitude" he defines the the reassertion of individual singularities. "When we speak of multitude, we reference a power opposing the Empire," he says, adding "today, multitudes are not obstacles but represent an inherent limit that cannot be surpassed. The multitude serves as a boundary to the world market."

But beyond the concepts of multitude and mass, another issue Negri cannot avoid discussing is war. In his view, war intervenes in the process of forming the Empire as an element that undeniably renews the control technique. However, also the definition of war has changed.

"Once, with 'war,' we indicated 'politics'; war meant making politics but with other means, a definition linked to a secular Machiavellian tradition," he states. "Is war today a continuation of politics? No, it actually stands at the foundation of politics; it is the regulating principle of the world, presenting itself as the governing element of the political reorganization of the world market. War now takes place against an enemy that isn't territorially defined. The enemy is within the Empire, a public enemy inscribed in imperial space."

"In modern history, when we spoke about an enemy within the national space, it implied that the police had to face it. The war in the Empire becomes police from a theoretical point of view. War is not the destructive conflict in which millions of men were thrown into trenches and set against each other. Before 9/11, the American army system underwent reform: it transformed so that the army became a sort of police force, divided into small units capable of surviving for long periods, transported to various parts of the world whenever and wherever needed. Intelligence and secret services are no longer considered parallel institutions to the army but are integrated into the military force. War and police do not identify with each other, but the two concepts are getting closer and closer, in a relation of continuity. Low-intensity war is identified with high-intensity police. We face such a situation in Israel, and we encountered it in Genoa. The enemy is a public enemy linked to the forms and problems of the structures. Often, the enemy is identified with immigrants, it is a mobile enemy."

Thus, Negri introduces the immigration issue, also analyzed in Empire: "The struggle against emigration, attempting to confine this workforce that moves in such an impetuous way globally, is one of the most horrible examples proving that the 'war-police' relationship is forming," he highlights.

In his opinion, the Empire's inability to control demographic and emigrating movements becomes one of the top priorities on the agenda of the 'war-police' relationship. In the emigration flux, we only see the immense sufferings of the emigrating multitude, and we feel compassion towards them. "To feel sympathetic towards these people is a good thing, a right thing to do," he comments, continuing, "but within these movements, we must also see the population's search for self-expression. Moving is a rebellious act, an extremely positive act. We must consider migrating populations as people who want to build new forms of freedom, positive elements breaking down the enclosures."

After talking about Empire, Negri moves on to talk about the journal "Posse". The name of the journal comes from Latin, it indicates the verb "to be able to", "posse" indicates what a body and a soul can do together, Negri and Hardt claim in Empire. The first issue of the journal was entitled 'Vivere nell'Impero' ('Living in the Empire') and it took into consideration a few themes developed in Empire. "We wanted to verify the things we had stated in our book and we started developing an inquiry on the job situation, observing how the multitude works in a new way, living in a condition of territorial mobility and flexibility, working in a society rather than in the factory, creating production through cooperation. We wanted to understand how the Empire functions in productive situations. What is extraordinary right now is that labor has truly changed and is returning to the center of our attention, as it was for Marx."

At the end of his presentation at the Michetti Theatre, Negri concludes his speech stating, "Once we used to say 'It is right to rebel'. Now they prefer to say 'Another world is possible'".

In Empire, Hardt and Negri draw a parallel between the new power and the Roman Empire: it can be divided into monarchy, aristocracy, and democracy - the fundamental trinity at the core of the ancient Empire. The army and organizations like NATO or the G8 nations represent the monarchy, while the aristocracy is composed of multinational corporations. Finally, democracy is embodied by various non-governmental organizations and the multitude.

Like a million-footed body, the multitude moves against the Empire, confronting the new global sovereignty, ready to subvert it and tailor it to its needs. Predicting the form of government that the future multitude may attain is challenging, we might say, paraphrasing Leon Trotsky's Literature and Revolution, but it is easy to hope that, like the human beings mentioned by Trotsky, the multitude will become "immeasurably stronger, wiser, and subtler; their bodies will become more harmonized, their movements more rhythmic, their voice more musical. The forms of life will become dynamically dramatic. The average human being will rise to the heights of an Aristotle, a Goethe, or a Marx. And above this ridge, new peaks will rise."

Future in the Past: In a Permacrisis, Take Refuge in Frutiger Aero Moods

Taipei Fine Arts Museum (TFAM) has just announced that artist Yuan Goang-Ming will represent Taiwan at the 60th Venice Art Biennale (April 20 - November 24, 2024).

Yuan's project - entitled "Everyday War" (at Palazzo delle Prigioni, Venice) - revolves around themes such as "home," "dwelling" and "an uncanny tomorrow".

The artworks project anxieties about the current political and social environment and point at the geopolitical tensions with China, but also at conflicts all over the world.

The exhibition's eponymous work "Everyday War", for example, is a single-channel video presenting before-and-after scans of a domestic space: glass shatters loudly, then warplanes fly in, destroying the objects in the room. In the end aircrafts annihilate one another, and the whole house is left a ruin in the aftermath of a furious battle.

As the camera keeps panning back and forth in a straight, steady line, lights and shadows gradually bathe the entire interior, and the collapsed home slowly returns to its original unscathed appearance.

The video becomes therefore a metaphorical exploration of the hidden fears and threats in Taiwan, but also invites to consider two unsettling thoughts - "war as part of normal life" and "war becoming the new normal".

While Yuan Goang-Ming's art mainly focuses on the state of affairs in Taiwan (a theme that will be even more relevant after the local elections in January 2024), the artist also ponders on the fact that the concept of war today has evolved from the actual firing of artillery to invisible expressions. "War as part of normal life," could indeed be interpreted as post-capitalist unequal distribution, cyber attacks, pandemics, and the discrimination and oppression of different religious and ethnic groups.

In the end, we all live a tumultuous life, in countries where there is an actual war and therefore a physical and psychological threat, or in situations where conflicts are carried out in other more subtle ways. All these battles destabilize us, stripping us of the necessary calmness to get on with our lives.

In this great era of uncertainty, it is therefore only natural to be pessimistic, but, for a much needed dose of temporary optimism, there are still uplifting trends left – like Frutiger Aero.

Making a notable return in 2021, almost marking the beginning of post-pandemic times, Frutiger Aero has found a home on TikTok, with the hashtag "#frutigeraero amassing almost 300m views.

Originally unnamed, Frutiger Aero emerged between 2004 (it succeeded the Y2K aesthetic) and 2013, shaping a distinct aesthetic across advertising, media, stock imagery, and technology. The aesthetics, provide a harmonious blend of technology and nature: motifs mainly revolve around computers and early flip phones, blue skies with clouds, aquatic themes, animals such as tropical fish and dolphins or plants and flowers.

User Interface key elements employing Frutiger Aero include linear gradients, bloom/glow effects, and a sheen for a 3D appearance.

Design-wise Frutiger Aero was characterized by Skeuomorphism, that is a style that imitates the appearance of physical objects by using elements that mimic their real-world counterparts.

So, for example, creating icons and elements that closely resembled physical objects to provide users with a familiar and intuitive experience (think a floppy disk for the save function or a bin for the trash function; as design trends evolved, modern interfaces moved onto flat design and minimalism).

The main palette for this aesthetics revolved around grass green, aquamarine, aqua green and yellow (think about Super Mario Galaxy 2 and you get an idea) and the images were more refined than most Y2K imagery and characterized also by glossy textures, but, at the same time, they were not that polished.

It was Sofi Lee, a member of the Consumer Aesthetics Research Institute (CARI), an online community committed to documenting visual aesthetics within consumer culture from the 1970s to now, who identified Frutiger Aero in 2017 and defined it with the Frutiger font family (by Adrian Frutiger) and Windows Aero, the design language of Windows Vista (the User Interface theme of Windows Vista/7).

Costume-wise the Frutiger Aero palette can be found in Bo Welch's film "The Cat in the Hat" (2003), while, for what regards fashion, some of the designs in Marni's S/S 24 collection evoke this mood, especially oversized T-shirts and long azure leather sleeveless dresses, floral decoupage evening gowns, mini-dresses or coats (some of the collage-like floral embellishments in these pieces were made from repurposed tin cans molded and painted to resemble metallic flowers), and more humble designs such as basic crop tops and hipster jeans or designs with plaids and stripes made strip by strip.

The combination of craft aesthetic/DIY with more sophisticated, almost technological designs in this collection points at Frutiger Aero. Besides, also Marni's Resort 2024 seems to reference this mood in its lookbook that features images characterized by distortion and by a bokeh effect on designs with a floral patterns. 

More recently the optimism of Frutiger Aero was evoked in Coca-Cola's new Y3000 drink: co-created with the assistance of AI, its packaging aesthetic is glossy and the light-toned color palette featuring violet, magenta and cyan against a silver base, gives a futuristic feel and also incorporates the distinctive Frutiger Aero bokeh effect to hint at a morphing, evolving state, communicated through form and color changes that emphasize a positive future (View this photo). 

Frutiger Aero also spawned sub-trends and subcategories such as Frutiger Eco (futurism and environment) and Helvetica Aqua Aero (also known as Frutiger Aqua, Aquacore, Low Frutiger, or simply Helvetica Aqua, obviously with images related to the sea).

Frutiger Jolly (sometimes also called Frutiger Santa, Frutiger Christmas or Frutiger Noel) explores instead holiday themes and hints at Christmas and Winter season holidays.

Frutiger Aero's aesthetic waned in 2012 in favor of the Flat Design look (Windows 8/10), but we are drawn to it through a blend of nostalgia for happier and less complex times and because this aesthetic offers a virtual escape from the permacrisis.

Its resurgence reflects indeed a desire for simpler times when connectivity didn't dominate our lives, offering a touch of retro-futuristic happiness amidst the complexities of modern existence, amidst talks of AI armageddon, wars and climate change (think about the devastating omission of a call for the phase-out of fossil fuels by the recent Cop28).

So, get optimistic filling your eyes with retro futurist Frutiger Aero images, reinvent its aesthetic using AI text-to-image tools, wait for Demna Gvasalia at Balenciaga to do an overpriced T-shirt about it or, if you can't stand its return, simply lie low until the trend subsides.

Transparency Matters: AI-Generated Images & The Erosion of Reader Trust in Media

In previous posts we have investigated the impact of Artificial Intelligence (AI) on the job market. While there are actually different professions that have been impacted by AI, there has been one service in particular that went through some radical changes – stock images.

Indeed, it isn't that complicated to generate credible pictures of generic themes, from Christmas and New Year's Eve to business meetings using AI text-to-image tools. 

In these cases, employing AI for image creation can be a useful solution, particularly for independent publications, websites and social media (I have done it in previous posts and I found this exercise great to hone my skills at writing prompts, but also useful to come up with large images and illustrations that may turn a long post into a more visually appealing article for readers).

Yet it is clear that major publications, facing financial challenges exacerbated by the declining ad market, have turned to AI not only for efficiency, but also as a cost-cutting measure.

During this year some major publications have been using AI just to cut staff: as you may remember from a previous post, earlier on this year German leading tabloid Bild announced a €100m cost-cutting plan that resulted in hundreds of job redundancies. The layoffs were partly due to tasks that can now be performed by Artificial Intelligence (AI) and automated processes in the digital world. Other publications followed its example.

Alarming statistics from Challenger, Gray & Christmas Inc. (Download The-Challenger-Report-October-2023-1) revealed a somber reality: without considering the cuts from November onwards, nearly 20,000 jobs have been eliminated throughout the media industry for a variety of reasons.  

In the last part of this year nearly 100 staffers were let go at Vice Media Group and G/O Media shut Jezebel, letting 23 editorial staffers go.

More recently, both Vox and Condé Nast announced workforce reductions. Vox, responsible for publications like New York magazine, Grub Street, and NowThis, implemented a 4% staff reduction, equating to a minimum of 20 employees, as reported by the New York Times. This marks the second round of cuts in the year, following a prior 7% workforce reduction in various departments during January.

In the case of Condé Nast, the renowned publisher of Vogue, Vanity Fair, Architectural Digest, Bon Appétit, and the New Yorker, a 5% reduction in headcount has been announced, impacting over 300 employees. 

In the recent Vox and Condé Nast workforce reductions, AI wasn't mentioned, but the declining ad market, which makes up a large portion of revenue for these companies, has been blamed by commentators. Magazines seem to have shrinking budgets that push them to cut their staff and to invest less money on writers, but also on their photographic archives or in stock image subscriptions, with some publications turning to AI-generated images that do not even look too refined and sophisticated.

Surprisingly, even prestigious magazines like Women's Wear Daily have started using them for generic business pieces. These images, bought from Adobe Stock, lack the refinement expected from a publication with a substantial archive, and their imperfections make them easily recognizable as AI-generated.

In one recent instance, WWD illustrated an article about retail strategy with the image of woman with her back to the camera carrying shopping bags, the configuration of the bag on the right and the jacket belt reveal it as being AI-generated, as proved by the user's portfolio on Adobe Stock. The image is clearly marked on the site as "Generated with AI", but WWD doesn't mention it.

The portfolio of this particular user also includes images of workers in different contexts, some of them, like an image of men working in a sweatshop, generated with the assistance of Artificial Intelligence. It is easy to recognise this image as AI-generated as Artificial Intelligence generates distortions in the depiction of sewing machines and tends to create pseudo-industrial sewing tools.

Hopefully, publications such as WWD will refrain from utilizing such AI-generated images, particularly in serious reports addressing human rights and exploitation in the fashion industry. The use of such images has indeed the potential to compromise the authenticity and credibility of the reporting.

The question that arises is whether these prominent publications should be more transparent about the use of AI-generated images especially when depicting realistic subjects (as opposed to illustrations). The lack of clarity raises indeed concerns about how readers perceive these visuals - will they assume they are produced with AI or will they think they are real?

Though AI offers cost savings, the ethical implications of using low-quality or easily recognizable AI-generated images without proper disclosure pose a threat to the quality of esteemed publications. This risk further extends to eroding the trust of their audience, something particularly noteworthy considering the subscription fees associated with accessing their articles.

Can Midjourney's /Blend Function Be Used as a Fashion Tool for Educational Purposes (or for a Fun Game)?

Disclaimer: Please note, this post explores the possibilities of the /blend function in Midjourney for educational purposes in fashion; we're not using it to steal images or designs and we are not encouraging fashion design students or designers to do the same. But we are exploring how an AI system behaves when it collages inspirations from existing collections or from famous designers. So, take this post as an educational exercise, almost a game to spot specific references in AI remixes.

A decade ago, I was tasked with compiling a set of fashion-themed questions for a Vogue Italia game. Some questions were straightforward, while others demanded a deeper knowledge of fashion history. Over time, a few questions lost relevance as they were tied to specific trends of that era, but those exploring the history of fashion remained pertinent and potentially educational.

Pondering on those questions led me to consider how the landscape of the fashion quiz has evolved.

Would people still engage with a fashion game centered on questions and answers? Could there be room for a board game dedicated to fashion? Designers Ferrán Renalias, Eloi Pujadas, and Malen Company recently answered this question by releasing a board game inspired by the Battle of Versailles, the legendary 1973 face-off between French and American designers, with the latter emerging victorious. The game does not revolve around fashion questions, but mirrors the dynamics and rules of the fashion battle between the two groups, making it an ideal budget-friendly Christmas gift for fashion enthusiasts. However, in a society dominated by visuals, could AI be used instead to play a fashion game?

In yesterday's post while exploring how easy (or frustrating) it can be to design a fashion collection with Midjourney, we also mentioned the /blend function.

This feature collides multiple images users upload into the system and creates something entirely new, producing unpredictable and often quirky results: at times, the system borrows the pose seen in one picture with a remixed ensemble from the other picture, so the results aren't often a perfect blend of two designs. 

Could we use this option to generate unique images for a fashion guessing game? Let's find out! Mark one point for each designer you guess in these blends. 

Let's start with something easy: the colors and geometrical configuration of this sheath dress and the hairstyle, should give you enough clues to guess where it comes from.

It’s a blend of Audrey Hepburn as Holly Golightly in Breakfast at Tiffany's and Yves Saint Laurent's Mondrian Dress (the colors and the grid point at it, but the hairstyle is Hepburn's).

Though passable and useful for our game, fashion-wise this is definitely not that interesting, especially because the Mondrian grid gives it away almost too easily.

Let's try another one: look at this model in a 1930s jacket and graffiti sprayed skirt: does it remind you of anything in particular?

No idea? Invert the colors, use the graffiti for the model's top and match it with a black skirt. Is this clearer?

Some of you may have spotted a clue in the the graffiti: that's the remix of Shalom Harlow's spray-painted dress from Alexander McQueen's "No. 13" catwalk show (S/S 1999). The jacket, though, represents an anachronism and it is indeed taken from Schiaparelli.

Now let's change one of the elements of the equation, let's keep McQueen's dress and combine it with something else, let's see what we have: the graffiti motif seems to have been filtered through the Ikat technique, forming almost a staircase motif on a dark emerald dress.

What's the second equation in this blend? Cinzia Ruggeri's "Homage to Lévi Strauss" dress. Now let's feed into the /blend function two designs by the same creator.

The following tiered dress may have a bit of the Christmas tree appeal as it seems to sprout red lights and it is donned by a model with a spiky crown of red lights.

Where could this come from? That's a blend of Ruggeri's LED dress and Lévi Strauss dress.

Would she have been offended by this experiment? I don't think so: Ruggeri was very much into the future and she would have laughed at this vision provided by a machine.

Talking about the future, looking for something more bizarre and maybe cutting edge? What do you think about this design with oversized partially padded portholes matched with a tortellino hat?

I'm not too sure about it, but it's rather extravagant, and it's a blend of Jean Charles de Castelbajac's teddy bear coat and Pierre Cardin's '60s kimono coat.

Let's keep Cardin's design and recombine it with something else to generate something even more dystopian: check this futuristic yet monastic hooded tunic matched with black gloves with appliqued silver metal fingers. Could this be a costume for a character in a sci-fi film? Maybe.

What's the second design employed for this blend? Can you guess it from the fingers? It's Mugler's Fall 1995 robotic suit.

If you're looking for evening wear and in particular for something more formal or even apt for the Christmas festivities, try this column dress with pleats that open up in a fringe.

Any guesses? Who did finely pleated designs? No, it's not Issey Miyake, try again, maybe the velvet swim cap will help you spotting one of the references?

Despite looking a bit like an Armani design, this is a blend of a Fortuny gown and an ensemble from Prada's S/S 24 collection with a tinsel curtain-like silver skirt.

So what's your score? Could you identify at least one designer in these AI blends or did you fail miserably? Do you think that Midjourney succeed in breaking their aesthetic and recombining it (well, you may argue that in most cases it came up with designs that aren't desperately desirable...).

You decide, but, remember: there's just one way to improve your skills in identifying design origins in fashion collections IRL and in images generated by AI  - growing more fashion-savvy.

AI Design Diary 2: New Developments in Creating a Fashion Collection with Midjourney - "Be the Crane" for Atelier Blancmange

The year is drawing to a close, so let's take a retrospective glance at some of the stories we looked at months ago to update their content and do some necessary follow-ups.

In April, for example, we looked at the developments that Artificial Intelligence (AI) text-to-image application Midjourney went through in the previous three months to see how easy it was to design a fashion collection employing this application.

Midjourney seemed easy to use, but not perfect: as you may remember from that previous post, a prompt to generate a fashion collection inspired by Renaissance costumes easily confused the system that, rather than using the reference as prints or surface details for the designs, started mixing fashion models with paintings, at times creating blends of humans and canvases with uncanny results.

If you remember that post, hands, fingers and other details such as eyes and lips were also imperfect and the best solution at the time seemed to try and come up with models with their hands in their pockets to avoid seeing undesired distortions or fingers in excess.

The system also exhibited discrimination, primarily generating models with white features. When prompted to create Black models, it often resorted to pasting the head of a Black model onto a body, resulting in a loss of the fluidity observed in the generation of white models.

Generating crowds was also a nightmare, so prompting the system to come up with the front row at a fashion show ended up producing guests with multiple distortions, with several people sprouting additional limbs or combining in one head the features of two different people.

In mid-March, after Midjourney v5 was released, things radically improved: everything was better defined, there seemed to be more diversity when it came to models and some images produced may have almost passed for real ones.

Yet, even the best attempts at producing a coherent fashion collection were plagued by tricks that Artificial Intelligence played on users.

When a user employs an AI application, the latter starts responding to the prompts and learns what the user likes or dislikes, eventually understanding which is the user's favored image in response to a specific prompt.

This means that, from then on, it will also start suggesting other users that specific aesthetic in response to prompts that may contain the same words.

In a previous post, for example, we have seen how in Midjourney the word "monster" in the prompt - combined with terms such as "kawaii", "stuffed animals", "plush toys", "knitwear" and "fun moods" - generates colorful jumpers with conglomerations of undefined colorful monsters.

We also learnt the perils of mode collapse: in the context of generative models, this definition refers to a situation where a model fails to capture the diversity present in the training data and instead produces limited, repetitive, or biased outputs. This phenomenon is often associated with generative adversarial networks (GANs), a popular class of models used for generating realistic synthetic data.

In a GAN, there are two main components: a generator and a discriminator. The generator creates synthetic data samples, and the discriminator evaluates whether a given sample is real (from the training data) or fake (generated by the generator).

The generator aims to improve its output to fool the discriminator, while the discriminator tries to become better at distinguishing real from fake samples.

Mode collapse occurs when the generator manages to produce samples that fool the discriminator, but these samples are limited in diversity. Instead of capturing the entire range of variability in the training data, the generator focuses on a subset of patterns or modes. As a result, the generator might produce repetitive or unrealistic outputs that lack the richness and diversity of the true data distribution.

To address mode collapse, researchers and practitioners employ various techniques, such as adjusting the model architecture, optimizing hyperparameters, using different loss functions, or incorporating regularization methods.

Additionally, there have been advancements in GAN variants and training strategies designed to mitigate mode collapse and improve the overall stability and performance of generative models.

As months passed, more featured were introduced in Midjourney – such as the Vary Strong, Vary Subtle, Vary Region and Zoom Out options, that allow to introduce more variations in a generated image. More complex prompts and using other functions such as /describe and /blend, can help adding more variety.

Is it easier or more satisfying to create a fashion collection now using Midjourney? Well, I attempted to do a new one in October for Atelier Blancmange, so let's have a look at what went on behind the scenes (please note, the draft images for the collection go from Image 1 in this post to Image 18; the following image opens instead the real collection).

The collection - entitled "Be the Crane" - was showcased during the second edition of Artificial Intelligence Fashion Week (AIFW).

The starting point for the collection were Taipei's claw machine arcades, combined with SpongeBob SquarePants's episode "Skill Crane" (Season 4).

In this episode, Squidward attempts to win a prize from the claw crane that Mr. Krabs installs in the Krusty Krab.

He eventually manages to do so after SpongeBob tells him that the secret to win a prize is to "be the crane", even though there will be a rather bizarre twist in Squidward's fate.

I had in mind a playful and imaginative collection inspired by this whimsical world that moved from the vibrant colors of soft toys and of plastic figurines from gashapon vending machines as well, hoping to come up with designs that encapsulated the carefree mood of childhood fun.

So the main moods of the collection had to be joyful, possibly overloaded with sensorial stimuli with textile and surface embellishments and some soft elements that pointed at cute plush toys, but also at other items, including hard plastic figurines or food items such as snacks and candies, also distributed by such machines.

The first experiments were not successful: I dreamt of a proper runway on which models walked among claw machines, but the system was confused about what a claw machine is and mainly created glass cabinets filled with toys.

Only several trials later I managed to create a runway that featured some claw machines on one side and the audience on the other, but no models.

For what regarded the clothes, in most cases Midjourney created editorial images of models with a pile of plush toys in the background or wearing designs made with plush toys or integrating them.

See the first few images in this post and you will get an idea: as you can see they aren't that new at all fashion-wise, but resemble more colourful and also more infantile re-editions of Jean Charles de Castelbajac's (View this photo) and Moschino's (View this photo) late '80s teddy bear coats (the idea was later borrowed again by Jeremy Scott for Moschino's S/S 2014 collection).

This wasn't the effect I was trying to achieve, so I discarded most images and attempted to come up with knitted designs.

Yet also knitwear wasn't a great idea as the system produced a series of jumpers that looked like conglomerations of cute monsters.

Eye-catching and colorful, most of these jumpers were too similar to kids' garments designed by Midjourney and currently advertised and sold on sites trying to scam consumers who order the AI designed garment, but then receive something completely different.

Further attempts at insisting along these lines produced more childish designs, kind of cute, but verging more towards the Japanese "decora" aesthetic and style, in bright and bold shades, yet not necessarily what I was looking for.

Besides, most of these images didn't show models on runways, but they looked like shoots done in a studio.

After working along these lines for a few days in October I completely abandoned the idea and started thinking about how to use my main inspiration in other ways.

Eventually I decided to shift perspective and place the models inside the claw machine, as if they were surrounded by cute toys.

This seemed to open up another world with more options and finally Midjourney started generating images that seemed to fit my vision.

Midjourney overloaded the set of the runway with toys and plush animals, and started flattening the toys integrated in the garments, presenting them as quilted or padded elements (and in some cases as prints), features that in a real collection may provide a sense of comfort and care, further enhancing the wearer's experience.

Models at times still displayed a look aligned with the "decora" style, but, rather than in super bright Crayola-like shades, the system satisfied my request for a softer palette, verging more towards dusty pink and lilac, faded yellow, pale blue and pastel mint.

For the final section of the runway I abstracted the colorful cloud motif that decorated the set, magnified it and came up with a sort of squishy or melting motif.

Oversized, whimsical abstract shapes, puffed sleeves, balloon and cloud-like elements are employed in this section to reflect the playful aesthetic of the main theme.

The same motifs were replicated in a few menswear items (mainly jackets and tops), while a bomber kimono jacket combined the padded toy theme with the abstract cloud motif.

Shoes and bags were inspired by gashapon machines and decorated with plastic spheres that pointed at the capsules dispensed by these machines, or embellished with small figurines and toys.

The gashapon bag was a bit tricky as the system has learnt to create handbags filled with plastic balls (View this photo) and tends to recreate this aesthetics when in the prompt you use words such as "handbag", "plastic spheres" or "capsules" (View this photo - not rare to spot this version on Instagram accounts by other AI artists).

Yet that wasn't what I had in mind; I was indeed trying to achieve another configuration that vaguely evoked a pétanque ball carrier and in the end I managed to.

For what regards backstage images, front rows and street styles (complementary images requested by the rules of AIFW), I didn't have any of the problems I had in February/March when generating images with Midjourney.

Midjourney seemed to respond well to my vision and there were only minor imperfections (almost undetectable in some cases) when it came to hands, eyes or lips.  

The level of details that the system managed to produce when prompted to create whimsical makeup and hair styles was definitely  impressive.

Street style images also proved fun to make and easily combined the aesthetic of a street in a Taipei alley, with some futuristic and whimsical twists.

I found problematic instead coming up with Black female models for this collection: the system lacked indeed consistency when it came to creating Black models that fit with this pastel fantasy world and kept on generating either young Black women in "decora"-inspired designs or wearing tracksuits decorated with bright elements that didn't fit with the rest of the collection.

So, after a year of experimenting with Midjourney, what do I think, is it worth designing a fashion collection with the support of AI? Well, it is fun to do it, but it can still be frustrating and there are still several issues to consider.

First and foremost, if you're not feeding into the system your own images, the system may be stealing and collaging other people's work (yes, I know what you're thinking, there are designers in the real fashion industry copying and pasting things from other designers' works and nobody seems to care…point taken, but in this case we are considering the highs and lows of AI).

This means that, when you create a fashion collection with AI, you must keep your feet firmly on the ground and not think you have created something previously unseen and 100% new as, in the same instant you may be feeling triumphant, the system may have generated a very similar image for another user in another part of the world (consider how the designs I have generated with quilted and padded clouds look like the more sophisticated versions of some kids' clothes generated by Midjourney and sold on scam sites...).

If you want to create your fashion collection with AI, remember that a vast fashion history knowledge may help you spotting copies or derivative designs (think about the early designs Midjourney generated with my first prompts that seemed to borrow from JCDC and Moschino, as stated above).

So, rather than thinking you're a great fashion designer, marvel at the polished and sophisticated image you have obtained and wonder how you can transform, enhance or reinvent it (if you're using Midjourney play around with Vary Strong, Vary Subtle, Vary Region, Zoom Out and so on, as highlighted above).

Last but not least - and this is probably the most important point to take into consideration if you want to take part in an AI fashion competition - be careful about the monster you're feeding.

The more you train it with your vision, the more the system will learn to respond to those specific words in your prompt with that vision, which means that, at some point, it will learn to respond in more or less the same way to another user.

If you want something more exclusive and unique, start working on your idea, leave it for a while, and then return to it again in a more intensive way, but for a shorter period of time, maybe closer to the deadline of your competition.

In this way the system may not be able to immediately borrow from the latest knowledge it has gathered from you to generate similar images for another user.

That said, while creating something, you may already be borrowing from the results of another user's training session, so once you obtain an image, do a research on the Internet to see if that specific picture you have generated looks like someone else's and it's already popular on social media.

From this list of suggestions and advice you can easily realize that doing a fashion collection with Midjourney is not that fast, after all. 

Sure, AI text-to-image applications allow us to quickly visualize something and, while they do allow us to save money on sets, props and teams, it takes several passages and stages to refine your vision and come up with something that in some ways mirrors what you had in mind.

Besides, consistency and coherence are vital if you are working on a collection and you must try and create different designs that should look as taken from the same show. So, try to focus on how to generate a variety of creations that may fit well in the same imaginary show.

Once you're done, expand your vision, adding AI videos, maybe an AI ad campaign, or imagine your brand ambassador, influencer or spokeperson as an AI twin, like Atelier Blancmange's Harumi K, one of the models from the "Be the Crane" collection.

Trained to talk about the collection and its inspirations, Harumi K is on Twise and you can interact with her vocally or type your questions and thoughts in a chatbox (mind you, at times, she still hallucinates...perils of AI). 

So, will AI fashion be a reality at some point? Well, Maison Meta's AIFW seems to be a reality at the moment and the winners get the chance to enter the Maison Meta x Revolve AI Fashion Incubator and get their collections produced. Yet, even though AI proceeds at a fast pace, it will probably take a while before a prominent fashion house or brand designs and manufactures an entire collection using Artificial Intelligence. 

Till then, just have fun with these systems or, as Alan Warburton states in his video essay "The Wizard of AI", "show me some AI candy!"