Ganassa Style Lora.

The artist's signature is prominent and appears almost always on the generated images.

Sidenote for this and any other model

You can have 100 people with 300 opinions on what the ethics of AI is or should be, but I'm not here to preach or tell any of you what's right or wrong, or what to think. Everyone is entitled to their opinion, but please let it at least be an informed one. The truth of the matter is that AI-generated images are a derivative work of an existing copyrighted material (copyright protection is received automatically when a creative work is created, aka an idea is "objectified", i.e. given tangible form). So unless strictly specified, any creative work is protected by copyright. Derivative works is a bit of a murky water, though. They can be made, provided the one who made them uses them personally, and that that usage is not commercial, aka they are not making money out of it. If you want to sell your derivatives of a copyrighted material you need a license from the copyright owner. This is the gist of the situation, and "what things should be", but we all know that this is not the current reality.

All this is to make you think. If an artist's signature bothers you much consider the above. If the above bores you, then consider this: a dataset (a good one) can include over 500 images. Clearing the artist signature from all of them is great effort for minimal/no reward.

So if a signature bothers you too much skip this model, or provide me (or someone else) with a signature free dataset for training.

Sidenote specifically for this model

You can use my Lora if you like this artist's style, or alternatively you can use "ausm" with no Loras in Pony and its merges in the prompt. I'll let you decide which is best for you, and, again, I'll not comment anything on the logic and ethics of having "hidden" tags in your model.