WAN T2V, I2V and V2V - A Refresher (Patreon)

With WAN 2.2 released, before diving straight in to the new things, I figured a quick refresher course on where WAN 2.1 is at the moment would help people transition to the new model. I know it's certainly refreshed my memory!

Video - https://youtu.be/OJM582CAyBk

Introduction

As we've seen of the past few years, the previous generations get various additions and we also get newer models even while that is happening! WAN is no exception here, and it's worth taking a look at the combination of WAN2.1 + PusaV1 + LightX2V LoRAs as they bring the execution time of WAN down to around 2 mins (on a 3090), all thanks to the low CFG and reduced step count requirement.

This combination gives great quality videos in just 6 steps, though you can go higher, play around with the LoRA strengths etc...

For the NEW WAN2.2 you'll obviously need to have the very latest ComfyUI, but before diving straight in to that one, it's worth taking another look at WAN 2.1 with these new LoRAs.

But before going headfirst into that, with all these different AI things going on at once, I figured a refresher on WAN 2.1 would be a good place to start from!

Workflows

There are 2 workflows attached, one with an LLM option and one without. They will likely also work with WAN2.2 as well, although I haven't tested yet and the LoRAs may or may not work as expected. Only one way to find out ;)

They include:

Text to Video

Image to Video

Video to Video

In all a single workflow, with RG3 options for fast bypass.

The LLM version requires Ollama running locally - https://ollama.com/

The model I used is IBM Granite - https://ollama.com/library/granite3.3

Links & Custom Node Information

The workflows have notes with links to the required models, as I think it's easier to click things in ComfyUI than having to refer back to this Patreon post, but I've included the main requires here also.

4 custom nodes are required for the non-LLM version, and can be installed using ComfyUI Manager:

ComfyUI Wan Video Wrapper - https://github.com/kijai/ComfyUI-WanVideoWrapper

KJ Nodes - https://github.com/kijai/ComfyUI-KJNodes

Video Helper Suite - https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite

RGthree - https://github.com/rgthree/rgthree-comfy

For the LLM-Enhanced version, you'll also need Comfy Roll Studio / Comfy Roll Custom Nodes, pretty old now but still seems to work.

This is the one I'm using - https://github.com/Suzie1/ComfyUI_Comfyroll_CustomNodes

Have fun! :)