This model was trained on a large dataset of images generated by using negative embeddings as positive prompts. Expect a tendency towards grey and minimalism.