If you want the best result do NOT use tags like masterpiece, best quality and so on. Those prompts act as a style and will heavily influence the outcome! Except you want to mix it, then do whatever you want.
Who else is happy that this crazy Anime is back after 15 years? 😏
Just had to do the styles because the animation is so crisp now.
Anima:
Unlike the IL version, there is no additional trigger for the limbs. The outcome of the style also depends a lot on the used checkpoint and LoRAs you mix in.
Illustrious:
I tried to make it in a way that you can push the arm/leg style. Since the dataset included background characters and ofc Garter, the body types vary much and in a test run the outcome was random. So if you want the arms/legs more like our main characters add chibi as a tag. You can see the difference in picture 1 (with chibi) and picture 2 (without).