PROJECT STATEMENT IN CONVERSATION WITH ChatGPT-4
The Artist: When I ask DALL·E AI to generate a photograph from the original captions of the 10 most iconic photographs, what happens and why?
ChatGPT-4: First, it's important to understand what DALL·E is and what it does. DALL·E is a variant of GPT-3 developed by OpenAI, designed specifically to generate images from text descriptions. However, as of my knowledge cutoff in September 2021, DALL·E was only capable of generating novel, creative images based on text inputs - not exact replications of existing photographs or precise real-world objects. The technology essentially imagines a new image based on the description it's given.
Now, let's consider what would happen if you asked DALL·E to generate a photograph from the original captions of the 10 most iconic photographs.
Given the captions of these photographs, DALL·E would attempt to generate images that it interprets from the text. However, due to the limited description in a caption, it's highly unlikely the generated images would closely resemble the original iconic photos. This is because DALL·E creates images based on the information provided in the description and does not have the ability to recall or reproduce specific photographs or images from its training data. Additionally, many iconic photos are recognized not only for the objects and subjects they depict, but also for their context, moment, and artistic composition – elements that may not be fully captured in a caption.
The results would therefore be DALL·E's creative interpretations of those captions and not exact replicas of the iconic photographs. The generated images would depend on how detailed and descriptive the captions are, and how DALL·E interprets this text.
It's important to note that AI systems like DALL·E are trained on large datasets and are not given specific information or have access to specific databases of known photographs. They don't "know" the world in the way humans do, but instead generate outputs based on patterns and structures they've learned during their training.
The Artist: So, what if I gave it some more context, like author, time, etc.?
ChatGPT-4: Adding more context to the captions could help guide DALL·E's output, but remember that DALL·E, like other AI models, doesn't have direct access to real-world data, a knowledge base, or any specific database of images or historical context. It generates images based on the patterns and structures it learned during its training, and its knowledge is limited to what it learned from the data it was trained on, with a cut-off point and no updates after that.
If you add more context such as the author, time, and other descriptive details, DALL·E will generate an image that tries to match that description as closely as possible. However, it won't necessarily be an accurate depiction of the real photograph.
For example, if you asked it to generate an image based on the description "a black and white photograph taken by Robert Capa during the D-Day landings showing a soldier in the water", it would try to create an image that matches this description. But since it doesn't "know" the specific photograph by Robert Capa, it can't recreate it exactly.
The resulting image would be an interpretation based on the description given and could vary widely based on how DALL·E has been trained to understand the text. The image might capture the broad themes but miss the nuances and specific details of the original photograph, or it might interpret the description in an unexpected way.
–––
View from the Window at Gras, DALL·E, 1.6.2023
DALL·E generated image based on: “This is the very first photograph ever taken. The scientist Nicéphore Niépce (1765-1833) was enthralled by the burgeoning craze for Lithography and decided to place polished pewter plates coated with a light sensitive chemical called bitumen of Judea inside a camera obscura. Niépce took this famous picture from his house in Saône-et-Loire. The exposure took eight hours. Once the plate was removed from the camera he used a mixture of lavender oil and white petroleum to develop the image.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “The French photographer Henri Cartier-Bresson took this famous photo behind the Saint-Lazare train station in Paris. Cartier-Bresson is known for coining the term “the decisive moment,” which refers to a photographer’s capacity to capture a subject at precisely the right instant and in doing so shattering normal life to retrieve a transcendent moment. The man jumping over the water can be compared to a dancer whose silhouette is reflected most gracefully in a puddle. An enigmatic, grey background juxtaposes this elegant figure. The image also perfectly reflects Roland Barthes’ theory about the punctum and stadium, which he explains in his book Camera Lucida: Reflections on Photography (1980). In this seminal book he defines punctum as the features of a photograph that produce or convey a meaning without invoking any recognizable symbolic system. This kind of meaning is unique to the response of the individual viewer of the image. The punctum punctuates the stadium - historical, social or cultural meanings extracted via semiotic analysis - and as a result pierces its viewer. To allow the punctum effect, the viewer must repudiate all knowledge. Barthes insists that the punctum is not simply the sum of desires projected into the photograph. Instead, it arises from details that are unintended or uncontrolled by the photographer, namely the detail that attracts the eye and the general impression that emanates from a photograph” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “We don’t know the names of the subjects in this famous photo or the name of the photographer. This image of workers taking a daredevil’s lunch break at the top of a skyscraper certainly makes the viewer dizzy but it also brings their attention to the very risky life the workers building the Rockefeller Center lead. In the first half of the 20th century, dozens of workers died after fatal falls during the construction of various skyscrapers.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “In the aftermath of World War II, Alfred Eisenstaedt took to the streets of New York's Time Square to immortalize celebrations marking the end of the war. This is one of the most famous photos of the 20th century; however while some critics believe it to be capturing a loving reunion between a sailor and a nurse others see it as sexual assault. The latter interpretation resonates with the present day #MeToo movement.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “Margaret Bourke-White, an iconic female photographer from the 20th century, immortalized Gandhi several times and is well known for this famous photo that was taken two years before his death. This picture was first published in the American magazine Life. After Gandhi's assassination, on January 30, 1948, Bourke-White’s photographs of Gandhi skyrocketed in price and took an important place in art history.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “On June 11, 1963, in a street in Saigon, Vietnam, the monk Thich Quang Duc immolated himself as an act of protest over discrimination toward Buddhists by the South Vietnamese government. During a demonstration, he asked to be doused with gasoline and demanded that he be set on fire. Associated Press photographer Malcolm Browne was on the scene at the time and captured a stunning image, a world-famous photo that also won a Pulitzer Prize. The American fusion-rap band Rage Against The Machine used it for the cover of their 1992 self-titled album.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “The Vietnamese photographer Nick Ut captured this chilling photograph on June 8, 1972. During the Vietnam conflict, the American army often used napalm during their missions, however this time one of the planes mistakenly hit the village of Trang Bang, wounding many civilians. Nick Ut brought the 9-year-old girl in this photograph to an American hospital where she was saved. After 17 skin grafts, the young girl went on to live a full life and even had two children.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “Steve McCurry took this famous photograph during a trip to Afghanistan in 1984 to document the migration of Afghan refugees. This picture of 17-year-old Sharbat Gula was taken in the refugee camp of Nasir Bagh. It became the symbol of the Afghan struggle and was featured on the front page of the National Geographic newspaper in June 1985.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: “This famous photo of a young Chinese man standing in front of tanks during the 1989 Tiananmen Square uprising made Jeff Widener the most famous photojournalist of the late 20th century. The day before taking this photograph Widener had been injured by a stone and was told to stay in his hotel, while all the other American and European journalists took refuge at the airport. Widener took the opportunity to photograph the uprising from his hotel window. He ran out of film and asked an Australian tourist staying in the hotel to borrow a roll of film. Widener used this roll to take this famous photograph, which is now widely held to be one of the most recognized photos ever taken and won the Pulitzer Prize in 1990.” (Source: https://www.artalistic.com/en/blog/10-famous-photos)
DALL·E generated image based on: Javier Bauluz was awarded the Pulitzer Prize for this famous photograph, which underlines the large disparity between social classes in the world. This image depicts two tourists passively sitting in front of the lifeless body of a migrant who was trying to reach Europe. (Source: https://www.artalistic.com/en/blog/10-famous-photos)