This model was trained on over 50 uncaptioned abstract images of widely differing styles until realistic prompts were affected. I absolutely do not make any guarantees regarding your results. You're rolling the dice with this one!