GenAI x Painter

In collaboration with Chilean painter Luis Gomez, this project explores how traditional artists can use AI as a tool to supplement and expand their creative practice. Combining Stable Diffusion, Python, and oil on canvas, the process was a dialogue between human creativity and machine-generated outputs, iteratively blending the two into a cohesive final piece.

Luis starting the initial sketch.

Initial sketch made by Luis.

I conceptualized the project and set up the tools, running Python code on a Google Colab notebook to generate AI outputs using Stable Diffusion’s image-to-image model. The process began with Luis creating an initial sketch, which I photographed and passed through the AI model. Together, we refined the outputs over multiple iterations until we arrived at a result that balanced Luis’ vision with the AI’s generative potential.

Initial outputs from Stable Diffusion.

Initial outputs from Stable Diffusion.

Once we finalized an AI-generated image, we projected it onto the canvas, allowing Luis to paint over it while maintaining his unique style. As the painting progressed, Luis added new elements, color-blocking sections for further development. To ensure the AI outputs complemented rather than replaced his work, I manually masked specific areas of the painting, generating new outputs only in the desired sections. This iterative process continued until the painting was complete, blending traditional and digital techniques into a single artwork.

Painting in process.

Painting in process.

Binary mask used for inpainting.

Color blocks with masked fish.

Final stage of the painting.

Masking.

Final Stable Diffusion output

Final piece