Cringe
After more than 1 year since Wan2.1 and practically a year after Wan2.2 you are releasing a specialized model, with nothing novel except "dancing". It's a model on outdated architecture, without inbuilt sound, without any quality improvements.
You are simply mocking the community, which has helped Wan get attention in the first place and to feed upon the dependency.
Ah yes, you can already do audio-video sync in LTX, simply providing the latent audio channel. Up to ~30 seconds, on consumer hardware. I don't think your project here is something special.
LTX, Nvidia and Sber's Kandinsky are building audio-video models, the latter dropping at any moment now. This is the last chance to save your face.
Alibaba has already lost the local image market. It is only a matter of time before we permanently replace Wan.
This is the last chance to save your face.
u sure 100% positive u r not on drugs?
Want a Wan that can generate audio? Well, it already exists Mova-720p / Mova-360p. It's based on Wan2.2, and the audio part uses HunyuanVideo-Foley.
Please remove this cringe post 😟