tbh we have had the hunyuan (bit i have spelled that wrong) model for a while now. Though that can not do image to video. Looking at the size of the model files I rather doubt this will work on my 3090 let alone standard consumer gpus
Tried image to video 480p on my 3090. the vram was filled fast. but it managed 10fps 3 second video (the default settings) in 411 seconds (that includes load time.. i think...)
I guess yes it does run on consumer hardware, its not as friendly as hunyuan...hanyuan?, but image to video is cool and the quality was surprising. I did a simple cybernetic character turn and smile with teeth at the camera (that creates a sinister smile)
Pro tip: swap out the god awful webp save node for video combine when using the templates. Not even vlc could play the webp format video (what on earth is that format?)
Just ran the t2i workflow on an RTX2060 with 6GB and it worked.
WAN please
Great work! But where to find the 'SaveWEBM' node in the workflows?
为了庆祝官方支持wan，我注册了！
希望官方可以原生支持的模型越来越多，也越来越好~
tbh we have had the hunyuan (bit i have spelled that wrong) model for a while now. Though that can not do image to video. Looking at the size of the model files I rather doubt this will work on my 3090 let alone standard consumer gpus
Tried image to video 480p on my 3090. the vram was filled fast. but it managed 10fps 3 second video (the default settings) in 411 seconds (that includes load time.. i think...)
I guess yes it does run on consumer hardware, its not as friendly as hunyuan...hanyuan?, but image to video is cool and the quality was surprising. I did a simple cybernetic character turn and smile with teeth at the camera (that creates a sinister smile)
Pro tip: swap out the god awful webp save node for video combine when using the templates. Not even vlc could play the webp format video (what on earth is that format?)
consumer friendly? = need 40GB VRAM for the good models… yeah sure
You can actually run on 8G VRAM if you have enough ram to offload, the post is a bit misleading
Just usual consumer chinese used 4090 with modded 48 gb =)
牛掰！我是第一个评论的吗？