Zero Two - Hunyuan only model (for now)
Since there is about 100 Zero Two models I thought I'd be the first to make a Zero Two hunyuan model.
This Lora is experimental, but it seems to do a good job. Mostly released it as a proof of concept since some of the examples looked as if they were officially animated.
Training was done with about 88 images (which is probably overkill) but I wanted to see how Hunyuan works with a similar dataset to what I train for on Pony models. The tagging was not done with natural language however the video examples are.
Description
FAQ
Comments (3)
Hi, the lora looks great, very good work, I wanted to ask you, with what graphics card do you train the loras for the huyuan video? and did you only use images from Zero Two or did you also include short videos of it?
4090 is what I use and sufficient with images. I probably could get more if I added some short videos which I might try and see if it makes any difference in terms of flexibility. Training animations is a lot heavier load though unless you have patience.
Hi, do you have a tutorial on how you create LORAs like this?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.