MIRAGES of MATONGE / MATONGE MEMORIES
For my street photography work, the Matonge quarter in Brussels is one of my favorite locations to roam randomly with my camera, at least once a week you can bump into me in the streets there. It is known for its large community of people of African origin, primarily Congo. It is a place defined by its diversity, colors, scents, crowds, street life, wigs, exotic fruit and vegetable stores, religion, food, bars, and so much beauty. I always wanted to make a series about the diverse beauty of the women I encountered by chance during my photography walks. Many times, whenever I asked women on the street whether I could portray them, regrettably, most of them declined for different reasons, ranging from not feeling comfortable about being photographed, to having a bad hair day, often linked to the belief that photographs could seize the soul, ... all valid reasons, that I respect. Then I noticed when talking to a friend about my encounters with different women in Matonge that my memory preserved many vivid details of the persons I met, from the expression of their eyes to the outlines of their faces, clothes, and hairstyles, to small details like the jewelry or small tattoos that were visible. That gave me the idea to create a series of portraits, turning my memory accounts into images. To achieve that, I started digging into generative artificial intelligence techniques and practices that allow me to create photo-realistic images based on my detailed remembrance. The result is a series of detailed portraits of women resembling those I met in Matonge. Important to mention is that the selection of women is not a cross-section of the women in Matonge, it is about specific women, that caught my eye, because of their beauty, expression, a certain aura, a particular look, anything that caught my attention. However, they still are entirely fictive, generated portraits.
"MIRAGES of MATONGE should not be considered a realistic account, but a faux-documentary as a tribute to the beauty and diversity of the women of Matonge."
My thoughts on the usage of generative artificial intelligence for image creation
AI provides a very intriguing palette of opportunities for image creators. Like the emergence of photography has changed the art of painting to a large extent, AI will also hugely impact photography. As a photographer, I did find it an enriching and challenging experience, starting with the conception of images using language, configurations, references, and tagging. I also found a lot of restrictions, limitations, shortcomings, and biases within the AI framework I used (Adobe Firefly), and it required a very intensive and sometimes extremely annoying iterative approach to refine the results to reach the outcome I had in mind. It was a time-consuming process, getting all the details right, from the general pose approach, position of the head, chin, hands, spots on the skin, folds and patterns in the clothes, the structure of the nose, the texture of the lips, the catchlights and the reflections in the eyes, the hair details, the jewelry, .. every single little detail to ensure it matched first of all my memory and also to create a convincing photo-realistic portrait that could pass for a portrayal of an actual person. In addition, I also had to translate all my photographic requirements into the image rendering, including elements like depth-of-field, shutter speed, light strength and angle, shadow usage, dynamic range, reflection of light on the skin, background lighting, ... etc. A real portrait approach would have been much faster and much more fun I have to say, because that is my known habitat, and also nothing replaces the real human interaction when creating portraits. I took the project very seriously and fully committed myself to the generative creation process to ensure that the result would meet my vision as well as the quality standards I defined. Nonetheless, I also consider it an experiment, a learning experience that would give me a taste and feel of what AI can bring from my perspective as a photographer. In that sense, I enjoyed this journey of discovery, making it crystal clear that I am not an AI expert and that the core of my passion lies with real people and real images.
Some thoughts from a social & cultural perspective
When developing the images using generative AI, I could not help noticing the cultural bias deeply rooted within the AI's frame of reference. In my creation process, I spent plenty of time, fighting AI-generated stereotypes, to the extent that I had to explicitly counter it constantly to avoid being dragged into the stereotypical representations that AI proposed. A simple example is that when I would use the simple notion of 'a woman' and let AI generate an image for that notion, the result was an image of a young Caucasian woman with long blonde hair, large blue eyes, and a skin softness that seemed extreme even to current social media filter standards. In the case of women from the Matonge project, I had to deal with many ethnic stereotypes. I consistently had to correct and provide feedback to educate AI not to generate a kind of semi-colonial image of women. It forced me to correct and educate AI, being very severe and strict rejecting more than 95% of the AI-generated images, and refining on a detailed level what I wanted and especially what I would not accept. This is certainly a worrying aspect of AI, considering the time and effort I had to invest in countering the social, cultural, and ethnic bias of the AI engine. It reflects a mainstream Anglo-Saxon/western-centricity that is very much present.
I also faced limitations and restrictions of AI regarding the human body, for instance when I wanted to portray a woman with a bare shoulder, it needed some creativity to make that happen. It seems very similar to the anxiety of most social media platforms regarding human nudity and always reminds me of a news article I read about an American school that classified Michelangelo's masterpiece 'David' as pornography. That preposterous puritanism is damaging in many ways, especially regarding freedom of expression.
The root cause for the bias issues has to be sought in the data sources on which the AI engine relies, so hopefully, the data bias issues can be tackled and improved to provide a more nuanced and open framework, as more people will also provide feedback on these issues. Improvement is certainly needed.
Finally, the purpose of the experiment was to create photorealistic portraits of unreal people and raises questions about the concept of identity, and the boundaries between reality and fiction when AI-generated people resemble real people to varying degrees.
Response from peers in the creative domain
I noticed a lot of negativity and hostility from my peers in the creative domain towards the usage of AI, people seem to feel threatened by AI. At first, I shared those reservations, prejudice, and resistance towards using AI, a textbook resistance reflex to change and the potential threat of a new medium. On the other hand, I was curious and eager to see it in action in the field of photography, and that curiosity helped me to look at AI with an open mind. The Matonge portraits project was the perfect opportunity to embark on the AI learning journey, as it was a project with a clear purpose and an approach combining my memory, photography vision, and generative AI. In that threefold approach, it was helpful I felt quite confident about the first two parts.
Finally, I also realize that my approach to photography makes it probably easier to be open to AI. I always considered my photography equipment, camera, lenses, lights, ... etc. as tools to realize or create an image that reflects my vision or ideas. And of course, in photography to deliver the photos you envision, a lot of skill, experience, and mastery are needed. However, the only thing that I find relevant is the outcome, and there are and should not be any restrictions on ways to achieve that. Photography or any form of art is about expression, not boundaries or rules that make no sense. I like to explore the limits where photography and painting meet and to achieve that I will embrace any form of creativity that will contribute to that end. That freedom is for me an essential part of creativity.
In summary, I don't think there is right or wrong when it comes to the use of AI, as long as there is clarity and the relevance of that will largely depend on its purpose and framing. For instance, when generating AI images of a Siberian tiger, you cannot sell that to National Geographic as a real wildlife photo. However, if you integrate the same AI-generated Siberian tiger into an imaginary artwork, you can sell it as a piece of art.
Concluding note
On a personal note, impressed as I am by the vast possibilities of AI, I will probably continue to build on my current photography approach, in which real people are essential, for which I will still largely rely on real camera equipment. And whenever I feel that AI contributes to improving whatever I want to express, I will not shy away from it.
The series is a GOLD WINNER on the NEW YORK PHOTOGRAPHY AWARDS 2024
in the category 'AI-generated images'
THE IMAGES BELOW ARE GENERATED USING AI POWERED BY
