Just fooling around with style adapters and different ways to combine conditionings. The model clearly has good responsivity to these kind of things.