Vijing the Mona Lisa with AI

One of the ABCs of playing with generative AI is transforming Leonardo da Vinci’s Mona Lisa. Painted in the early 16th century, during the height of the Italian Renaissance, it has since become a universal pop icon—just as iconic as Warhol’s Marilyn. And precisely for that reason, attempting something even moderately innovative with her poses challenges on several levels.

On a technical level, it allows me to explore which AI tools are more advanced when it comes to achieving realistic results, and which ones simply don’t allow meaningful interaction with the painting without turning it into an illustration. Let alone preserving her “unreachable smile” and the characteristic technique of sfumato, whose optical effect is inevitably compromised when you try to set the painting in motion. Trial and error gradually narrows down the set of tools I trust for specific tasks.

On a creative level, the real challenge was how to present her in a personal way. Once again, my focus is on bringing joy to the women without identity, portrayed by men with great names in art history. Here, what I bring to the table is my own identity: my love for felines, for coffee, and for vijing.

It’s her, but it’s also me.
And I like to see her dance.