My ComfyUI workflows for using Wan 2.2
This workflows are used by me to create my art.
They are optimized for my checkpoints and created of my latest knowledge to enhance the outcome.
"If this workflow leveled up your day, I'd purr-eciate a like! 😻"
Versions & Information👇👇👇👇👇👇👇👇
👉 Please read below and the file descriptions "About this version" for more info's.
🎬🐈Click me! HowTo + Video
💡Some WAN 2.2 versions use high+low checkpoints, other like S2V use only a single checkpoint - make sure to read the descriptions and use the correct checkpoints.
⚠️ Do not use the workflows with the "Nodes 2.0 beta" from ComfyUi or it will mess up things.
👇👇👇👇👇👇👇👇
What you get from the comfy workflows:
♨️ Easy controls
✅ As less as possible dependencies
🪧 Detailed documentation
⛓️ Highly automatic logic
✨ Optimized results
🎬 Fully automated resolution logic
🔖 Bookmark-Shortcuts with number keys
Types of workflows
FastFidelity C-AiO
🖼️ I2V and FLF2V
🧩 Automatic aspect‑ratio calculation and fitting
✨ Multiple Upscalers
Torchlanc (very fast, color correct, low VRAM)
Upscale with Model (additional detail, high quality)
RTX Super Resolution (ultra fast, very accurate)
🤝 Video resolution matching - Fully automatic scaling and resolution calculations
🕠 Length automation - Fully automatic calculation of frame count
🫥 Watermark option
🧮 Color match feature
👾 MiniMeme feature - Create small gif's
🎭 NAG - Negative prompting with CFG1
🪄 Interpolation feature
🌀 Perfect loop feature
🃏 Last Frame Extraction
🔖 Bookmark-Shortcuts - with number keys
FastFidelity C-SVI
🖼️ SVI (SVI 2.0 up to 10 samplers)
🧩 Automatic aspect‑ratio calculation and fitting
✨ Ultra‑fast, color‑correct upscaling (torchlanc) OR Upscale with Model
🤝 Video resolution matching - Fully automatic scaling and resolution calculations
🕠 Length automation - Fully automatic calculation of frame count
🫥 Watermark support
🪄 Interpolation feature
🔖 Bookmark-Shortcuts - with number keys
Swarm Basic
Absolute basic SwarmUI preset
Not recommended over ComfyUI for video generation
Backend Test
If this does not create the example video your ComfyUI backend is broken.
If this works, but your other workflow not, the other workflow is broken or missing dependencies
🩻 Known issues and advice's
⚠️ Some workflows may set on webp av1 encoding (VHS node) - If your computer/setup missing drivers use any other like H265 or H264!
Install ffmpeg!
Update Comfyui and custom_nodes!
Update pytorch 2.9+cu128 or higher
Make sure to read where files/models should be placed inside the workflow
Check if the filepath for model/clip/vae match your system like Linux/Windows
The plugin ComfyUI-DD-Translation can break node connection (avoid)
All older Versions are available inside my GitHub Repo.
Spacial thanks to @Abyss_Games for a really good idea and help with the "loop"!
I got some really good ideas from @Gladas workflows!
YOU are responsible for outputs as always! If you make ToS violating content and I get aware I WILL report this.
Description
Changes:
Centralised Sigma-Shift to the other "Setting" nodes.
Optimised for new checkpoint "LureNoir"
deleted/substituted mtb and easy-use dependencies (partly broken with latest comfyui)
Substituted features with standard comfyui nodes -> lowered dependencies
Changed Sampler for streamlining with MoE-KSampler
30+ Tests showed same or batter quality as 2x KSamplers
Less Nodes
No speed penalty
Reintroduced separation of VAE and CLIP for more control for offloading
Slightly less memory consumption
Minor adjustments
Requirements:
FAQ
Comments (22)
I like the workflow; I was able to work with it, but I noticed something. I’ve been working with Daxamur’s workflow https://civitai.com/models/1853617?modelVersionId=2165705 which uses three KSamplers,He told me that it helps improve motion, and it really does. With his workflow, I’m able to create 9-second videos at resolutions between 820p and 1080p in less than 10 minutes. However, that workflow isn’t very friendly with fluids — it tends to make them disappear.
Now, in his setup, the upscale and interpolate processes run at the end, once the x-second clip has already been generated. In your workflow, however, the upscale happens while the clip is still being generated, which makes the generation process take much longer.
I’m not an expert, but is there a way for the interpolate and upscale steps to be done at the end, after the video has already been generated?
Thanks for the infos. I'll have a look :) Maybe I can learn something :)
What I can say, atm, that I'll not do the 3*KSampler thing. Thats has one benefit of more motion but ... many downsides. Maybe I'll explain that in my guide...
@darksidewalker thanks !!!! looking forward to it :)
I had a look into his WF. It is not bad, but it has many flaws from my view of things:
1) Loading clip_vision_h - just not usable with wan 2.2, waste of resources
2) Using a realesgan upscaler as pth - potential security risk (?) and adding up on resources (slow)
3) Much more dependencies (backend and nodes)
4) The first KSampler will add steps and kill some details, operating out of the distilled range. Also it will not utilise the option to get the optimal high-low-noise ratio flip.
5) It uses KJNodes - which are not my cup of tea
The good things may be the face tracking and color correction, but I did not see any real advantage from my testing. Maybe even not wanted in terms of anime/abstract videos.
My interpolation also runs past video gen. What I could improve is the sequence of upscaling, that could possibly a minor improvement.
My goal is just as less dependencies as possible and easy usage.
Immaculate workflow! Appreciate the very detailed notes and for embedding the upscaler in the workflow. Blessed you with 10k buzz for Early Access to your LureNoir High and Low models.
As a note: your workflow works right out of the box for your RadiantCrush High and Low models, but when I switched out the models for the LureNoir checkpoints, the workflow threw a ClipTextEncode error ("ERROR: Clip input is invalid: none"). I am still new to ComfyUI, but I finally figured out that the workflow will not let you have a blank/empty Negative prompt box for that model. I just threw in something like "deformed" and finally got the workflow to run.
Keep up the great work! 👌
Thanks for sharing this workflow! Even someone like me who's clueless with ComfyUI can use it and generate results without any issues. I usually stick with A1111 for image gen, but this worked great!
Thank you and thx for the buzz!
Using 6 steps; it uses 2 steps for high and 4 for low. Is this working correctly?
FLF1.1? If so, yes. MoE-KSampler always gets the shifting steps based on the correct latent-to-noise ratio. Not the steps.
If you want to 50%/50%, you may use the 1.0 with the standard KSamplers.
version 1.1 Missing 32 fps interpolation... 16FPS is not smooth
It is not missing. Just enable it? 🤔
You're goated ngl. Bless you.
I left a comment on the wrong model, oops.
Is there anyway to include WhiteRabbit's InterpLoop into this somehow? When I use FLF with the same picture to create a loopable animation, I get dirty frames as it tries to overlay the last frame over the first. WhiteRabbit's workflow has a couple nodes that resolve that, however I'm a beginner at Comfy and I have no idea how to connect those specific nodes to this workflow...
This is a limitation of WAN 2.2 it seems. He made a massive workflow to prevent the dirty frames if you use the same image as 1st and last frame. I can not interoperate this just easy into my workflow. You have to do it or ask WhiteRabbit to do a WF for WAN 2.2 14B i2v.
I did add a basic loop feature, maybe not as powerful like the one from whiterabbit, but easy to use. You could check out my latest workflow.
@darksidewalker Awesome. I was able to create something myself that works decently well too.
I'll have a look if I can tweak your workflow to improve it as well, if possible.
@Abyss_Games Feel free and if you find something useful I would love to hear that!
Any thoughts? i keep getting this error everytime.
WanMoeKSamplerAdvanced
mat1 and mat2 shapes cannot be multiplied (77x768 and 4096x5120)
I would guess your chooses a incompatible resolution?
I've got the same error. Fixed it by switching CLIP. Most likely, you're using the wrong CLIP model. You're probably using CLIP from lurenoir's dependencies (umt5-xxl-enc-fp8_e4m3fn), which has a mismatch in matrix sizes with the model. Use umt5_xxl_fp8_e4m3fn_scaled instead.