✨ WAN2.2 — Image to video — Simple Workflow
A clean, all-in-one WAN image-to-video workflow built entirely with the UmeAiRT Toolkit for ComfyUI.
Only 12 nodes. No spaghetti wires. Just load your model, write your prompt, and hit generate.
⚠️ IMPORTANT — Nodes 2.0 Required
This workflow is built for the Nodes 2.0 (Vue) interface of ComfyUI. If you don't enable it, the workflow may have display problems.
How to activate Nodes 2.0:
Open ComfyUI
Go to Settings (⚙️ icon, bottom-left)
Find "Use Nodes V2 (Vue)" and toggle it ON
Refresh the page
Load the workflow
If you prefer the classic interface, check out my Legacy version of this workflow instead (link).
🎯 Features
Text-to-Image generation
Automatic download of models in auto version
Built-in SeedVR2 upscaler — high-quality tiled upscaling (toggleable on/off) Slower than a classic upscaler, but significantly better quality
Full metadata embedding — your images are saved with all generation parameters, ready for online publishing and remixing
3 LoRA slots — with individual on/off toggles and strength control and you can connect as many other lora modules to each other for as many LoRA as you want.
📦 Custom Node Required
Only one custom node to install:
Install via ComfyUI Manager (search "UmeAiRT") or use the UmeAiRT Auto-Installer.
The Toolkit packages everything internally — upscaler, face detailer, metadata saver. No other custom nodes needed.
📂 Files you need (in manual version)
For base version
I2V Model : wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors and wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors
In models/diffusion_models
CLIP: umt5_xxl_fp8_e4m3fn_scaled.safetensors
in models/clip
For GGUF version
I2V Quant Model : wan2.2_i2v_high_noise_14B_QX.gguf and wan2.2_i2v_low_noise_14B_QX.gguf
In models/unet
Quant CLIP: umt5-xxl-encoder-QX.gguf
in models/clip
For speed version
lightning LoRA : Wan2.2-Lightning_I2V-A14B-4steps-lora_HIGH_fp16.safetensors and Wan2.2-Lightning_I2V-A14B-4steps-lora_LOW_fp16.safetensors
For AIO version
I2V FAST AIO Model : wan2.2-i2v-rapid-aio.safetensors
In models/checkpoint
VAE: wan_2.1_vae.safetensors
in models/vae
ANY upscale model:
Realistic : RealESRGAN_x4plus.pth
Anime : RealESRGAN_x4plus_anime_6B.pth
in models/upscale_models
Description
🔄 Full Nodes 2.0 rewrite
Rebuilt entirely with the UmeAiRT Toolkit pipeline (only 1 custom node needed)
SeedVR2 Upscaler and metadata saver now built-in with on/off toggles
12 nodes total — clean, visual, beginner-friendly
⚠️ Requires "Nodes V2 (Vue)" enabled in ComfyUI settings
FAQ
Comments (35)
This has to be a really stupid question, but... I'm confused! In the new x3 LoRA slot, how do I set HIGH/LOW?
This is a simplification to apply LoRa to both
so how exactly do you set the lightning lora to "high and low" respectively when it applies to both models? i feel they really should be seperated. simplicity is nice, but this breaks some loras that need very specific high/low values. i think the lightning lora's i have are not even being applied, as it takes me 200 secodns to generate a 3 second video on the new workflow and 140 seconds of generation on the legacy workflow.
@Sirkura Yes it’s why I have created a dedicated node for Lightning Lora that patch only the corresponding model
I'm trying to use only native ComfUI functions, and sometimes they're slightly less efficient than the older custom nodes. I need time to implement all the optimizations.
@UmeAiRT how exactly do you set the changes to the lightning loras? i have about 8 i use for varius things and couldn't figure it out. the defaults you included usually are the best, but i don't know how to set them to be used, unless the installer put them in the right location automatically?
@Sirkura I only had one Lightning lora ; if you have others, I will add them to the dropdown menu in a future update. Send me the links.
@UmeAiRT Here is a bunch here i like Dyno. https://huggingface.co/lightx2v/Wan2.2-Lightning/tree/main. then i have the high and low lightning loras from your first "legacy" wan. wouldn't it be easier to set it to check a folder for the drop down and auto populate it there? additionally i think seperating lora's to high and low would be useful, since some work better as high/low models respectively. maybe on the manual version? i have some i use that create artifacts if applied wrong.
@Sirkura I'm going to create some “advanced” settings so you can choose all of that. I'll take care of it this weekend—I'm currently on a business trip.
@Sirkura I'm having a little trouble figuring out how to integrate this into the UI. I'm hesitating between creating a LoRa node where you can choose high/low noise or a Lightning node with a choice of models.
So if I have both low and high versions of a lora, do I apply both? Just one? If so, which?
@UmeAiRT i'd say the easiest way is just to add 2 nodes for high/low loras wouldn't it? since lightning lora's fall in that anyway, 2 lora nodes for high and low would tackle both of those issues. for now i've just been using the legacy version since it has that option, half the lora's i have don't work with the new worflow unfortunately.
@Sirkura @DoctorAculaMD I've released a new version where you can choose which noise is applied to each LoRa.
@UmeAiRT I'll test it when i can, the latest update to comfy broke almost all my nodes. so i'll have to fix and repatch them for my build.
@Sirkura ComfyUI's updates are sometimes truly awful ^^" That's why I ended up creating my own installer and node toolkit. I hope things work out for you.
@UmeAiRT I got everything working again, luckily it was just one of the junction links i made breaking with the update, my models weren't being found. I had to make sure everythign worked for the workflow i'm making for flux. Some of your nodes have actually been really helpful, dropping the node count from 77 to 25.
Back on topic: The good news is that it works! I actually like that you have the lora target selection on the node. it worked out well. the lightning lora node for me doesn't help me, but it might help others. i just have a high and low lora i use for it which i just put in the normal loras instead. Also i noticed your auto updater for workflows grabbed the newest 2.1 workflow, but it didn't grab the manual 2.1 workflows. i downloaded it off the website.
@Sirkura I hesitated quite a bit about the best solution for LoRa, and I think a selector like this is the best compromise. For now, my workflow synchronization tool only downloads the automatic versions. I'm waiting to finish migrating all my old workflows before figuring out how to organize everything, because right now there are over a hundred workflows! ^^"
@UmeAiRT Oh i found another minor bug in the workflow. i tried to drag and drop one of the previous videos i created with the workflow to recover my settings for it and it didn't have the metadata embeded in it, despite having embed metadata toggled on in video output, it was missing the workflow data.
Great ! how can we added custom Wan ? (DaSiWa or else) ?
The manual version allow you to choose the files, he dont work?
Hi there, thank you so much for sharing your workflow! I really appreciate all the hard work you’ve put into it.
I’ve run into an issue, though: even after installing the UmeAiRT Toolkit, I still get an error saying that the node UmeAiRT_VideoFrameInterpolation is missing and needs to be installed.
Could you please let me know if there’s anything else I need to do to get this working?
Thanks again for your help!
It’s strange, you have the lastest version of the toolkit?
@UmeAiRT Probably so. I tried installing UmeAiRT-Toolkit via git clone. Maybe I should use ComfyUI Manager instead. I'm still figuring this stuff out, total newbie here haha.
@UmeAiRT Just wanted to update you on my situation! I tried installing the node again via ComfyUI-Manager, but still ran into the same error. After some troubleshooting, I updated my ComfyUI version, and now everything works perfectly. Thank you so much for your help!
@xanxan942 i didnt see you have same problem with me but you already solve this. Can you explain how to fix this error? Im already update my comfyui and everything fresh but still show UmeAiRT_VideoFrameInterpolation error
@Horiiiii To be honest, I’m not entirely sure myself. I installed version 1.5.0 of the UmeAiRT-Toolkit via ComfyUI-Manager, but it still kept showing that “missing node” error. Then I updated my ComfyUI version, restarted everything, and the error was gone. I don’t fully understand why that fixed it, though.
@Horiiiii I forgot to mention that I always test my workflows on the latest version of ComfyUI available on the day they're released. The interpolation uses a fairly recent native function, and if ComfyUI is too old to detect it, it won't load.
That's probably the cause of your problem.
helloo, im facing a problem. My UmeAiRT_VideoFrameInterpolation has error. what should i do for fixing this issue?
Node 'ID #14' has no class_type. The workflow may be corrupted or a custom node is missing.
@Horiiiii Interpolation uses a new feature in ComfyUI. Have you updated ComfyUI and the toolkit?
@UmeAiRT my bad, yess im running on my old comfyui. Now run perfectly, thank you for help!
Just a minor typo i found. This file is named "image2image." instead of "image2video" and the text to video is named "text2image" instead of "text2video"
it also doesn't specify if its wan 2.1 or 2.2 in the name, which can be confusing if you have a folder of workflows.
Thank you, I must have made a mistake with the copy-paste. Regarding the file name, the secret is that they work for WAN 2.1 and 2.2. Since I coded the toolkit, I can make it adapt automatically, which I couldn't do before.
I accidentally broke my ComfyUI environment a while back when trying to update it, so seeing your auto-installer back up and running is a welcome sight to say the least.
The newest version especially seems even more resource efficient and faster compared to previous iterations.
Any plans on making a workflow with start image/end image input as well in the future?
That's the only thing I can think of that would be nice to have.
Yes, I'm updating each workflow little by little, but since I'm now coding all the nodes myself, it's taking me a bit of time