Faces in Art

What happens when you teach a neural network five centuries of portraiture and ask it to dream?

Faces in Art is the result. It’s a visual exploration of the "latent space" of art history, a continuous, morphing journey where iconic portraits from the Renaissance to modern expressionism flow into each other. It’s not just an animation; it’s a machine’s interpretation of how the human face has been captured by artists over time.

Watch the full video:

The Concept

I wanted to see if I could use AI to bridge the gap between static art and dynamic music. By using StyleGAN, I used a trained model on thousands of curated classical and modern portraits. The goal wasn't just to generate new faces, but to find the paths between them, the organic transitions that make a Renaissance oil painting feel like it’s slowly breathing into a modern abstract piece.

Technical Poetry

The technical challenge here was temporal interpolation. If you move too fast or too randomly through the latent space, the transition feels digital and jarring. I spent hours fine-tuning the interpolation paths so the morphing felt natural, almost liquid. It’s a mix of raw matrix multiplication and artistic intuition.

Synchronized Atmosphere

The visuals only tell half the story. The entire sequence is precisely locked to my track Sombra. I treated the AI transitions as another instrument in the arrangement:

• Eras flow with phrases: Renaissance faces emerge during the melodic beginnings.
• A Unified Experience: Every visual "breath" is synced to the musical arc.

The Result

The final piece is where my two worlds collide: AI engineering and music production. Even though the origins are synthetic, the result feels deeply human. It’s a testament to the idea that the most interesting things happen when we use machines to explore our own artistic history.

Watch on YouTube | Listen to Sombra