Update: Anima version is now up with mostly the same dataset used from the Illustrious version with a different trigger word. It will get the character's look fine, but sometimes it'll make her armless and have her wings be the arms itself for whatever reason especially if you generate her flying in air (I'm starting to see how Anima can have it's downsides now lol). Put "armless, legless" in the negative prompts to help mitigate that from happening. The tail won't come out correct too sometimes. Illustrious will be the better option for this lora versus Anima. Everything else works fine from my testing.

Light Fury aka Nubless from the DreamWorks movie How to Train Your Dragon 3: The Hidden World. All of the image training was done taking about 40 4K image screencaps from the movie focusing only on her. No fan art was trained, so it be true to the movie's art style. If the eyes appear a bit weird sometimes it's because a lot of her close-up scenes have her eyes reflecting what she's seeing from her pupils. If you give her narrowed eyes or don't do close-up shots, then this issue doesn't really pop up. Lowering down the weight a bit or combining it with an art style lora of your choice will help with that issue too. The tail might also not generate the most accurate due to her not having much scenes showing her tail in full without any motion blur the movie has which would affect the training badly if I were to include the motion blur shots.

Positive Prompts: LightFuryHTTYD, furry, feral, dragon, scalie, light-blue sclera, black eyes, white body, white scales, light-blue body, light-blue scales, two tone body, two tone scales, dragon wings, long tail, light-blue tail frill

Extra: angry

Optional: forked tongue