Must read: The quality terms of this model are:score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up   In simple terms, it can be understood as the scoring of the image quality, there is no need to add additional quality terms. Scores 4-6 can be removed and added to the negative terms, positive terms generally use scores 7-9 for quality ratings.

'source_pony', 'source_furry', 'source_cartoon', and 'source_anime' can also be added to positive or negative. For example, if it's needed to create anime-style images, add 'source_anime' to the positive side; if it's real people or 2.5D, it can be added to the negative side.

This model does not necessarily require negative terms, but it is best to add: NSFW when generating images on the platform.

The most prominent feature of this model is its text parsing ability, in simple terms, it can draw whatever you say. It excels in natural language processing, surpassing most models, and crucially, it complements sdxl's biggest shortcoming!

This model is trained based on mistoonAnime. I think mistoonAnime itself looks great, but I found the background a bit dull in my opinion. Therefore, I enhanced the background and color representation, added more detailed elements, and strengthened the animal elements of the furry characters. This model is adjusted based on my personal preferences and the original model by the expert is already excellent.