Innovate Futures @ Benji
patreon
(ComfyUI Workflow)Wan VACE 14B 5 In 1 Features Workflow Version 20250518
🕑 Added 2025-05-17 19:01:49 +0000 UTChey!
Just tested 14B last 2 days and updated VACE workflows. And I was thinking, how about just put it in one workflow, with all the fundamental features that VACE have, and just share to you all.
It's pretty interesting and things come with good timing, we just talk about CausVid and MoviiGen in the last video. Find out that CausVid 14B Lora model do help a lot for loading VACE 14B locally, and within 4-6 steps it does generate something like a normal Wan 2.1 video. Plus, we loaded with Wan 14B Text2Vid model, it do perform even better.
I think you will delete that VACE 1.3B once you have try this. :)
More information and tutorial will post later. I just want to share the workflow here first hand as this is done.
Have fun :)
Comments
Benjamin Law
Mix_Color_by_Mask is coming from https://github.com/BadCafeCode/masquerade-nodes-comfyui. And Video_upscale_with_model https://github.com/ShmuelRonen/ComfyUI-VideoUpscale_WithModel This upscale node are doing pretty fast , and better than SD upscaler
Synthius
Mix_Color_by_Mask and Video_upscale_with_model can't be found after i installed Comfyui-WanVideoWrapper. Any idea how to resolve or replace those two nodes with something else?
Benjamin Law
try reduce screen size . width, height . or reduce frame length.
Carl Nimbus
Any chance of getting this to run on a 4080super with 16gb? I'm getting OOM when trying the ControlNet2Video.
