Yo, Civitai just dropped Hunyuan and Wan2.1 in their generator, both with LoRA support.

资讯 / AI 快讯
1

Just saw Civitai added Hunyuan and Wan2.1 models for video generation, and both support LoRA — so you can actually customize the output instead of being stuck with the default style.

I usually make some dynamic product demos, and before this there were only a handful of video models to pick from. Now we’ve got two Chinese options, and you can slap your own LoRA on them too. I’m really curious if I can lock in a specific style. The official post says the detailed usage is in their video generation guide. Anyone here already tried them out? Which one gives more stable results, especially for text-to-video?

2

The fact that both support LoRA is such a game-changer, way more room for customization.

3

I’ve run Wan2.1 elsewhere, the motion consistency is pretty solid.

4

Yo anyone tried Hunyuan for generating characters? Is it more stable? Need a comparison pls.

5

You can run it directly on the site, saves you the hassle of setting up your own environment. Super convenient.

6

For product showcases, is the duration long enough?

7

I’ve run Wan2.1 too, motion is smooth, but slightly complex movement still gets blurry, simple actions are the most stable