Version 2.0 - Updated this to work with latest version of Comfy.
You can now provide LORAs for each of the 4 video segments.
This workflow allows you to create long videos and also supply a reference face to perform a face swap to try to keep the face consistent through out the video.
I recommend using the face swap if you only have 1 person in the video as it can sometimes face swap the wrong person. yes, this is rather annoying.
If you are missing the Video User Input node, you should be able to get it from the manager,
You will need to make sure comfy has been updated to support "Subgraphs"
This workflow uses subgraphs
Using a resolution of 800 x 600 each video segment has been taking about 4 minutes on my PC.
WAN 2.2 LIGHTNING
https://huggingface.co/Kijai/WanVideo_comfy/tree/main/Wan22-Lightning
GGUF
https://huggingface.co/QuantStack/Wan2.2-I2V-A14B-GGUF/tree/main
Description
I simplified the workflow. Removed the reversal subgraphs .
This is a more straight forward workflow.
This should run a bit faster as well, since the reversal logic is removed.
II am trying to keep the interface as simple as possible to use.
FAQ
Comments (8)
I've been trying to get this to work but I don't have a clue about subgraphs, how do I open them to set up the models ,can anyone explain it to me please.
Ok I have worked out how to load the models but I get this error now when I run the workflow. ( KSamplerAdvanced SM89 kernel is not available. Make sure you GPUs with compute capability 8.9. ).
我很喜欢这个工作流，但提示缺少“TKVideoUserInputs”节点，管理器里也找不到这个节点。能不能告知节点下载地址？谢谢！
Thanks!
Is there any way to add checkpoints & loras to this?
Please simplify the workflow.
Hi, This is a clever Workflow.
Great Job. it turns square in your head.
thank you
Bye
SD
Excellent workflow.. Looking for where I can control the latent size -- it keeps clipping my 1024x1024 input image, impacting the result
This is pretty cool. I wish I knew how to apply different loras for each step... I'll figure it out eventually :-D
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.