MINRILL ANIMATION
Hunyuan Video version of my Minrill LoRA, trained on the original dataset and some additional gens from the LoRA itself. Works great for a clean animation style.
Description
FAQ
Comments (10)
👍🏻👍🏻👍🏻
Thanks, will try it out for sure. Your video's look super high quality, thanks for providing the prompts. They also look like they contain a lot of high quality :D
Curious to see if you managed to get good quality out of it as well.
Can you tell us how you was your dataset prepared for this LoRa? Only pictures of the art style ? With minimal captioning?, i might wanna try to do something like this but different style, i guess working with pictures is a bit easier than videos
Just images yes, and not the most extensive captioning, just generally descriptive. Like 5,6 lines per image.
This is still the best style lora for HV. Would you mind sharing a bit about training - resolution, #images, learning rate, epochs? Thanks for the great work!
254 images @1216*832 res, 16 epochs, 5e-4 learning rate
Thank you! I tried 512 and 768... guess I need to be more patient!
good job! will there be a wan version?
Thanks. I haven't set up WAN for LoRA training yet, but I might try it one day, as I am using it a lot more than Hunyuan currently.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.