For the past 15 years, I’ve used self-portraits as a way to share my inner world. Traditionally, I would set up my camera, arrange the lighting, and then dive into Photoshop to immerse myself in surreal environments.

Now, I’ve been leveraging LoRAs (Low-Rank Adaptation) to train image generation models using a curated set of my portraits. This allows me to produce high-quality, consistent outputs that depict me in virtually any scenario.

Below are some of my explorations using this technique.