Our work on continuous, subject-specific attribute control in text-to-image models is available online.