My attempt to create a Wan 2.2 concept lora. Its scope is pretty limited, but I think it works well enough.
I'll be honest. Video generation is still pretty arcane to me, so I don't really have advice for. It's a lot more difficult to experiment, and prompting seems to be ancillary to a good starting image, and the right Loras.
Description
FAQ
Comments (7)
Hell yeahhhhh!!!, minimax please 🙏, I will pay you 10,000 buzz 😭.
I'm looking into it, you don't have to pay me anything, clearly there are a lot of people who want it
I think I've been waiting for this LoRA ever since I signed up—thanks a lot.
Model looks great. I wonder if this would work well for POV.
I don't know. The only perspective in the dataset I didn't add to the showcase was leaning back while kneeling. But it's in the dataset, and it would be the closest to a POV I think, so it could be possible
LORA is bullshit. She takes over all the movement, and the video turns out to be crappy if there are at least three characters in it who are doing something.
Multiple people is always going to be difficult, as for who is doing the motion I don't know what the problem might be. Looking at the videos I uploaded you can see both