WAN support

[pflky]

I already have other ways of running WAN, but I hope this model gets supported as an option, and I would assume LORAs would be supported eventually. But right now Hunyuan is essentially dead, ever since WAN was released it took a major backseat amongst the video generation community because WAN is better in virtually every way. GGUF WAN is also very small in size, and even the lowest quant of WAN GGUF is capable of delivering great results, especially when combined with LORAs. Hunyuan just doesn't hold a candle to it. The only bonus of Hunyuan is that it's a bit faster than WAN, but WAN's file size and memory usage is also much easier to manage, and its results are more than worth the extra processing time, but with stuff like teacache and sage and triton, WAN is significantly sped up too, closer to Hunyuan.

[garbageplaidskids]

Wait so you are telling me this doesn't support any models and loras unlike normal image generation forge etc? I was so hyped about this damn.. glad i didn't waste my time with this.

[LindezaBlue]

Would be a waste to not try. It works really well despite not using Wan 2.1 models. And I am sure support will eventually come.
But as with most things new, they won't always have all the "bells and whistles" be patient, this was given to us for free.

[garbageplaidskids]

I didn't said the otherwise and idk why i get 3 downvote lol. Y'all need to chill, i am already patient and i'm grateful that Illyasviel still giving us free stuff.

[xSemi]

i saw Wan Mentions in the paper but still not a released models i hope for an Update to Support the Wan Model in FramePack and will agree on Addion Lora Support would be Great too. But i would prefer endless Wan Img2vid outputs with Framepack when supported