First try at a wan video LORA, so I thought why not use a character I've already trained.

This a T2V version, 14b.

I used the same dataset as the PONY lora for the training, and I only put the trigger word and the physical description in the captions.

Like I said, first time training a wan lora (and I'm still learning), so there's definitely room for improvement (but I already like what it can do). Let me know what you think, or if you have tips to improve it. I already have some ideas, but your opinion matters.