"Neural Fad" merges AI with Tokyo's rich fashion history. Can you share what initially drew you to explore the intersection of AI and fashion, particularly focusing on post-war Tokyo subcultures?
Emi Kusano: In the beginning, around 2017, I wanted to create artwork using AI, so I incorporated it into music videos. However, the outcomes did not meet the expectations I desired. Early AI was very abstract, which wasn't what I was looking for. However, around 2022, when I tried using Midjourney, I was amazed by how realistic the photos were. From that point on, I started using Midjourney daily, followed by generating images with Stable Diffusion. At that time, I didn’t have a specific theme in mind, but I was deeply knowledgeable and passionate about fashion, especially niche, archival Japanese fashion. As I became more absorbed in creating these works, "Neural Fad" was born. When I used to take photos as a fashion street photographer, there wasn’t a strong trend or movement like the era from the 1940s to the 2000s depicted in "Neural Fad." During the time of globalization around 2008-2009, everyone was wearing brands like Zara and H&M, but I had an obsession with the domestically influenced fashion culture back in the days when we were heavily influenced by television.
Your work reimagines fashion subcultures that never actually existed. How do you approach the creative process of blending historical accuracy with imaginative future projections? What sources of inspiration do you draw from?
Emi Kusano: While creating "Neural Fad," I realized that large language models (LLMs) are biased towards Western inputs, making it difficult to generate niche trends from Japan's 80s that lasted for a short span of 5-10 years. Therefore, I found it fascinating to prompt AI by drawing hints from details like the makeup and hairstyles of, for example, the Takenoko-zoku tribe, which might have originated from American glam rock, or satin jumpsuits that resemble kimonos. It was during this creative process of generating scenes of niche Japanese fashion that I became aware of cultural bias within AI.
How did you utilize AI in the creation of "Neural Fad"? Can you elaborate on the specific technologies and tools you employed, and how they helped you achieve your artistic vision?
Emi Kusano: "Neural Fad" consists of 100 NFT artworks created using Midjourney, while "Morphing Memory of Neural Fad" transforms these images, along with newly generated ones, into immersive videos.
The process involves creating a model based on my generated data and using a tool called Deforum to handle camera direction vertically and horizontally. Additionally, I trained the model on my face, so similar facial features appear in some of the characters shown in the "Morphing Memory of Neural Fad." I also created the music featured in this work, generating tracks by training free sound sources from each era with a song I previously composed called "Fake Memory."
You worked with video director Kanji Kyoda on this project. How did your collaboration influence the final outcome of "Neural Fad"? What was the creative dynamic like between you and Kanji?
Emi Kusano: My collaborator, Kanji Kyoda, an AI fashion researcher, helped me effectively express the specific morphing aesthetic I wanted to convey in "Morphing of Neural Fad" to AI.
For this project, Kanji-san participated as a video director. His extensive knowledge of fashion enabled him to grasp the imagery, details, and Japanese fashion references I wanted to communicate. We collected images from each era and discussed how to present the created data. In the final work, the visuals in the video transform at a speed that the human eye can’t keep up with, but that spiraling motion creates an immersive feeling for the viewers.
The collection evokes a sense of nostalgia while imagining an alternative reality. How do you ensure that your AI-generated art retains emotional depth and resonance with viewers?
Emi Kusano: No matter how much AI is involved, there's always a human behind the machine, providing their perspective and life experiences. Without these elements, the result would be empty, bizarre, and either creepy or merely amusing. I believe AI’s advantage is its ability to narrate previously untold stories in various expressions. It allows for the expression of impossible compositions in photos and enables people who can’t hire models or visit specific locations to create unique works. The infinite parallel universes in which AI generation takes place are also incredibly intriguing. I think what captures human interest the most are the emotions someone felt while creating their work, the narrative behind it, and their relationship with it.
Finally, what message or experience do you hope viewers take away from "Neural Fad"? How do you envision your work influencing their perception of fashion, history, and technology?
Emi Kusano: There are many AI-generated images that resemble Japanese fashion, but in my case, they are created by someone who is a nerd of each depicted fashion culture. I want people who become interested in Japanese fashion culture through my work to explore archives and books to research the cultural background of that time. I think both Western influences on Japan and vice versa show that culture evolves like a game of catch. As mentioned earlier, the current AI is heavily biased towards the West. For instance, inputting a prompt like “beautiful person” often results in predominantly white faces. The work I’ve created hacks this bias. I hope it sparks even a little interest in the background of niche subcultures. I also want people to immerse themselves and enjoy the work. Viewing it on a large screen offers a sense of immersion. If there are any opportunities to exhibit my works immersively, I welcome any invitations.
コメント