Apr-10-25 hotfix:
Made some changes to the workflow to improve quality, see the "About this version" section for details.
This workflow implements SAM3 for character masking.
It uses looping to process long videos, and while it can theoretically generate infinitely long videos, in practice it is limited by VRAM and processing time.
It supports both Wan Animate modes: character animation and character swap.
It has been tested on 1-minute, 480p videos using an RTX 4090 with 64GB of RAM, though it supports 360p, 720p and 1080p resolutions too. My tests suggest you can generate up to 20 seconds at 1080p with an RTX 4090 using a video batch size of 80 frames (if you have the patience and the time for it).
Features:
Character animation:
Full reference image animation using a video reference.
Character swap:
Insert a character in-place in a video using high quality segmentation with SAM3.
Face swap:
Possible using "face" as "CHARACTER ISOLATION DESCRIPTION", but don't expect miracles. You can achieve better results by face swamping the face separately using some edit model like Flux Klein or Qwen edit and then use that new edited image as your Character Reference Image using the character animation feature. 
Detailed instructions are contained within the workflow itself:
- Yellow nodes are input and configuration nodes you can change to suit your needs.
- Red nodes are instructions and helpful notes
Description
Changed the VAE decode nodes for the tailed variant, for improved performance.
Add an option to choose from tiled or normal decoding.
Added instructions for face swap.
* Hotfix:
Removed a node that worsen quality.
Changed the "VAE Decode (Tiled)" node settings for better quality outputs.
FAQ
Comments (10)
Haven't tested yet, but someone explain to me.... Why people are so awesome :D
edit : It works great and it's amazing thanks. But is the a tweak I can use to modify face influence ?
Unfortunately no, what I personally do to get slightly better character consistency is use my wan2.1 character LoRA, the Animate result is pretty close most of the times, so the light effect of the LoRA does the rest. If not even that is enough, I do a second pass using Wan2.2 and the LoRA and that sets it.
@LatentHeart I see.. I'm asking because some other WF had the possibility to turn up/down the strength of the influence of the face animation. Which allow to use expression loras like wink, ahegao, etc
@hboxgames132 Oh I see, in this workflow what you can do is bypass the Set_POS_VID_FACE node in the "POSE ESTIMATION" group to completely disable the face expression part, or use the "POSE ESTIMATION ALT" instead, and enable "detect_face" field in the "DWPose Estimator" node, this group uses the native dwpose movent only, which has a less strong effect in the face expression.
@LatentHeart Ah great ! thanks for the tip !
I notice a halo effect or a transparent outline around the character. Do you have any idea what this might be? What exactly causes this halo effect or transparent outline to appear around the character? Thank you in advance.
I've been researching, I it seems to be a combination of high resolution reference image / high resolution video output / lighting LoRA / long video duration (long being more than 5 seconds) that triggers the effect. The alternative is to use low resolution images and a low resolution video output; I'm not 100% sure what exactly causes the effect though. This models are trained to behave for short 480p videos at a high step count (30+ steps), every twick we do to that base line (step reduction, higher resolutions, longer videos, etc) pushes the limits of what they where intended for.
https://www.reddit.com/r/comfyui/comments/1mxcxvg/comfyui_wan_22_video_ghosting/?show=original
My comfyui and wan is up to date yet two nodes shown missing.
PoseDetectionOneToAllAnimation,
WanCacheOptimizer.
Wan cache optimizer is optional, you don't really need it. Here's the github repo:
https://github.com/Jasonzzt/ComfyUI-CacheDiT
PoseDetectionOneToAllAnimation is part of the ComfyUI-WanAnimatePreprocess pack by KJ, you can find it here:
https://github.com/kijai/ComfyUI-WanAnimatePreprocess
Both custom nodes should appear in ComfyUI manager though, so just to be sure, update comfyui using both, the bat file and ComfyUI manager.
BEFORE INSTALLING ANYTHING, ALWAYS MAKE A BACKUP OF YOUR COMFYUI SETUP, SO YOU CAN ROLL BACK IF ANYTHING GOES WRONG.
@LatentHeart I already installed that still give me node missing error. Just checked again and switched from latest version to version 1.0.2 and now it works lol wth. Thanx for the repy tho :)