I conceived the ‘Statua Fluida’ prompt on June 1st, 2023, while exploring the potential of DALL·E 2, a generative image artificial intelligence developed by OpenAI.
Earlier In May 2023, I took photos in the streets of Dijon (France), between Rue d’Auxonne and the Chartreuse de Champmol, two significant places from my youth. In the following days, on my computer, I decided to use DALL·E 2 to perform inpainting on each of these photos.
How does it work
On the photograph I upload to OpenAI’s DALL·E site, I erase the pixels I want to modify and place DALL·E’s generating frame over the erased area. I then input a prompt to describe what I want to appear, and the AI generates four drawing proposals for me to choose from.
30 street photos of Dijon modified by inpainting with DALL·E
On Dijon #23, dated June 1st, 2023, you can see the first drawings generated using the ‘Statua Fluida’ prompt.
On Dijon #24, dated June 2nd, 2023, the AI creates a truly strange statuary group, which inspires me to work with the ‘Statua Fluida’ prompt in outpainting.
The results obtained with the ‘Statua Fluida’ prompt in inpainting were promising. The outcome proved to be even more accomplished with outpainting and the exquisite corpse technique, as you can see in the Hall of Frescoes. I realized that the AI reached a greater freedom of expression when it was no longer constrained by integration into a complex photograph.
4 photos taken at the Musée des Beaux-Arts of Dijon and processed for outpainting with DALL·E
Outpainting involves extending the image beyond its original edges. In the 4 images below, I wanted to explore DALL·E’s ability to extend downward the photographs of artworks seen at the Musée des Beaux-Arts of Dijon.
The following 4 photos show the screen of my computer while I am drawing with DALL·E 2 :
Come and see more behind-the-scenes photos in my Pensieve.