A new study from Google AI Research examines how diffusion models generate novel imagery. The team analyzed the relationship between training data and output to quantify creative drift. This research provides a framework for understanding when models innovate versus when they simply interpolate. Practitioners can use these metrics to tune model diversity.