This is my first lora for Minimax H3 and in general for video base model.  The target was to get a lora that would always generate the same body with the same voice. I think that I achieved this target, unfortunately this lora has a strong dependency to generate platform shoes because I had a pair of those within the training dataset. But when you scale down the strength to 0.70 the base model will overwrite this. I normally generate my character Violet with her violet hair and golden eyes, but you can change those in the prompt. I hope you like this Lora and If you find anything which I can do improve don’t hesitate to tell me.

Violet for Minimax H3 (Character Lora with body and voice)

Changed the Dataset to remove the general generation of platform shoes and the influence of other clothes.  Additionally trained a Lora for text to video.

Files with SHA256: eafed7e777edeea61623f824ecc30b1b6a06a180ae32c1bfc35b046124a84023