Not so Simple (or is it?) Wan 2.1/2.2 Lightspeed I2V Workflow
Details
Download Files
About this version
Model description
The model it is meant to work with the Lightspeed version of Wan 2.1 and 2.2 by DarkSideWalker
This workflow is very basic in what it does:
Image to Video
Extract last frame from video
Combine Video
Even my 5060ti (16GB) can generate decent videos locally now. My generation time varies from 2 to 5 minutes for a 480x640 5-second video. YMMV.
There are more advanced and probably nicer workflows out there. This was just something I wanted to try and I like my results.
I don't plan on adding much more to this workflow. My goal was to keep it relatively simple and not require a ton of additional custom nodes. It currently has more than I wanted already.
If for some reason you are getting OOM:
You can try adding "Clean VRAM used" nodes from comfyui-easy-use or whatever you prefer for that function.
Those weren't helpful for me, so I will not be including them.
Regarding Upscaling Video:
There seem to be some nice options out there for upscaling videos, but unfortunately the one I liked didn't like my version of pytorch and I don't plan on downgrading or having multiple versions of anything python related. I'm too lazy for that.
These workflows contain nodes from some or all of these: