A lot of artists use text-to-image generative models to explore creative ideas. However, in most times, the model misinterprets the artist's intention and it is hard to obtain satisfactory results, even after dozens of attempts. This project aims to provide personalised support for artists to use generative models.