Hunyuan Video LoRA, training from 15 short videos (480*480*41, ar length=3, not square resolution)， using RTX 4090 and diffusion-pipe.
This LoRA is not overfitting and can be used with other Face Loras.
blow_kiss_epoch60_accmulation2.safetensors
Same model published on other platforms. May have additional downloads or version variants.