Technically a preview, but the Frieren x Fern one was manual and not part of the set.
Made a big update to my app, so that now it's all completely modular and not strictly set to one prompt. So rather than one prompt including, the pose, action, effects, face, clothes, setting, etc, all of these are separate and are dynamically switched out between generations.
Today I made it so that facial expressions match the pose/action, and clothes match the setting/environment. Should make for much nicer consistency.