⚠️Requires 9 custom node packs (10 custom node packs if using MMAudio)
⚠️ Uses GGUF Quantized Wan2.2 base model (for light weight or low VRAM)
⚠️ Uses Lightx2v distill loras for low steps generations
⚠️ Uses Subgraph
⚠️ Dense backend node processes
⚠️ Only tested in ComfyUI Desktop version
✅ UI Oriented workflow
✅ Switch between i2v/FFLF or t2v in one workflow
✅ 4-8 steps generations
✅ No Spaghettification
✅ Comes with other utility workflows
‼️Custom node ComfyUI-Swwan has been reported to cause problems in the workflow. (Make sure to uninstalled it, it has conflict with some of the switches in the workflow) thanks to all the users that help to troubleshoot this.
‼️Disable "Node 2.0" if you are using the latest ComfyUI Version.
🔆The latest ComfyUI Version have memory offloading of VRAM to RAM during video generation like block swap. (Blockswap currently not available on the latest version of ComfyUI for native nodes)
🍥Click Here(CivitAI Article) for Model/Lora Download links + Detailed Guide
Video posted above includes Embedded Workflow. (Download the video, drag into ComyfUI)
Merged videos will not load my main workflow.
Need ComfyUI-VideoHelpSuite custom node to open up workflow from video.
20/12/25 - 🟢V1.1
Fix Manual Scale - previously not scaling to the right resolution in megapixel for certain aspect ratio
Changed Video Upload format in Video Merger and Post Processing workflow from "Wan" to "AnimateDiff" - which will cause missing frames if frames are trimmed off.
Thanks to @Ifvin for discovering the above.
Fix Non-GGUF Workflows Default Model in Loader - previously accidentally swap i2v - t2v low models, causing i2v to generate different video from image used.
Added Volume Adjust to MMAudio subgraphs
All workflows will be included in one download files.
6/12/25 - Added 🎼 t2v/i2v/FFLF v1 + 🎼 i2v Only (+Batch) v1.0
Added MMAudio to main WF in separate download page.
Added/Updated 🎼i2v Only (+Batch) v1.0 with MMAudio WF. ("🎼i2v Only" downloads now contains WF with MMAudio and without )
Added MMAudio to Video 🛠️Post Processing WF.
Updated/Added Instructions for MMAudio installation/download/setup in (CivitAI Article).
Added Non-GGUF WF to all Downloads.
3/12/25 - Added i2v Only (+Batch) v1.0 Workflow - (⚠️ i2v Only)
Capable of generating batch videos with different images and different dimensions or aspect ratios from a folder.
Does both i2v with single image input or i2v batch images from folder.
Single input image i2v function like normal with after generation 🛠️Post Processing options editing following usage guide.
⚠️ Must set up all 🛠️Post Processing options before i2v batch videos generation. (Cannot edit or change 🛠️Post Processing options once generated)
2/12/25 - Added Extend t2v/i2v V0.01 Beta Workflow (WIP - Not perfect)
Capable of generating video and continuing in segments
⚠️t2v uses i2v model to extend/continue the video - Lora for t2v will not be able to work, as you will need to pass in i2v loras for continuation which are not implemented yet. unless the t2v loras are usable on both i2v & t2v.
⚠️More extensions/segments = more degradation to character/faces of the video (less coherent over segments).
⚠️FFLF is removed from the workflow due to conflict
⚠️ Only can pass in a single prompt/lora stack and durations are fixed at the moment.
⌨️Usage for 🟢Main t2v/i2v/FFLF v1:
Select t2v or i2v/FFLF Mode.
t2v: Adjust resolution for video
i2v: Drag and drop in the 1st image loader, choose to manually scale down image or disable it to use original image dimension
FFLF: Enable i2v mode and Enable FFLF and drag and drop image in the 1st & 2nd image loaders, choose to manually scale down image or disable it to use original image dimension (it will use 1st image dimension)
Edit/Add 🍏Prompts, add ✨Loras if needed.
Adjust ⌚Duration or just leave it at 5 seconds.
Set 🚶➡️Total Steps & 👞Split Steps. (4 Total/ 2 Split for Fast, 6 Total/ 3 Split for Good Quality, 8 Total/ 4 Split for Higher Quality)
Click on "New Fixed Random" in 🎲Seed Node.
Generate Video (▷RUN).
Repeat step 4 & 5 until you get desire video shown in 🎥Preview.
While using the same 🎲Seed and without changing anything that is not in 🛠️Post Processing, change/edit 🛠️Post Processing Options, Generate Video (▷RUN) again as many times and skip through KSampler.
Video generated will be in your output folder - 🗂️ComfyUI/Output
🌀To start on a new video project, disable everything in 🛠️Post Processing and set 📺Final Video Mode to 🎥Preview.
💡If you have lower VRAM
Click on the "ComyfUI icon" for the menu
Go to "Settings"
On left bar, go to "Server-Config"
Scroll to "Memory"
Look for "VRAM management mode" option
Select "lowvram"
Restart ComfyUI
📱The Workflows includes:
GGUF model loaders
Sage Attention
Lora Stackers
WanNAG ( to strengthen negative prompt when CFG is 1)
Auto input for FPS (based on Frame Interpolation and speed)
Post Processing:
Frame Interpolation
Upscaler with Model
Frame Trimming
Video Speed adjustment
Manual Adjust FPS (overrides auto fps inputs and speed adjustment)
Video Sharpening
Add Logo/Watermark
Save Last Frame
Frame Select
MMAudio (WF with MMAudio)
💽Download Files includes:
Main i2v/FFLF/t2v Workflow
Postprocessing for video (non-interpolated video)
Videos Merger/Joiner
Simple Megapixel Calculator
🧩Custom Node:
(All Custom Nodes are available in Custom Node Manager)
Open workflow, open up 🧩Manager, click on "Install Missing Custom Nodes", Select all and install, restart ComfyUI.
ComfyUI-GGUF (Manually search in the Customer Node Manager and install this, if it does not show up in “Install Missing Custom Nodes”)
rgthree-comfy
ComfyUI-East-Use
ComfyUI-KJNodes
ComfyUI-VideoHelpSuite
ComfyUI-essentials
ComfyUI-Frame-Interpolation
ComfyUI-mxToolkit
WhiteRabbit
ComfyUI-MMAudio (only for workflow that uses it)
🪞Extra Utility Custom Node: (not required)
ComfyUI-Crystools (add real-time graph in ComfyUI to monitor %usage of CPU, GPU, RAM, VRAM)
🚧 Progression on Video Extend on hold.
🖥️ My Hardware Spec:
♠️ RTX 3090 Ti 24GB - 64GB RAM
📽️ My Video Generation Stats:
🚨 Sage Attention Enabled
🖼️ 720*960 (0.66 megapixels - Scaled down from 1440*1920)
🚶➡️ 8 Total Steps + 👞 4 Split Steps
💠 Per iterations/steps ≈ 47-49 secs (5 sec video)
⏱️ 5 seconds video / 81 Frames ≈ 7.5-8 mins (455 - 480 secs)
🛠️ 💎Sharpen + 🦄Logo Overlay + 4x 🎞️Interpolation + 1.5x 📐Upscale ≈ 2mins (120 secs)
For 🎼MMAudio Reference:
🔗(NSFW) Dead-Simple MMAudio + RIFE Interpolation Setup for WAN 2.2 I2V
😀SeoulSeeker(CivitAI)
Description
20/12/25
Fix Manual Scale - previously not scaling to the right resolution in megapixel for certain aspect ratio
Fix Non-GGUF Workflows Default Model in Loader - previously accidentally swap i2v - t2v low models, causing i2v to generate different video from image used.
Changed Video Upload format in Video Merger and Post Processing workflow from "Wan" to "AnimateDiff" - which will cause missing frames if frames are trimmed off.
Added Volume Adjust to MMAudio subgraphs
All workflows will be included in one download files.
FAQ
Comments (31)
Супер. Спасибо большое.
hi, thanks for the workflow. all the switchers on my side is missing (📀t2v - 💿i2v Switch for example). tried different versions from the manager and nothing helped, if someone find a fix please let me know <3
Check if you have "ComfyUI-Swwan" custom node installed in your comfyui. This custom node cause the switches to not appear properly due to conflict.
I solved this problem with deleting "Mixlab nodes" custom node that claude ai suggested to me recently
Thanks for your workflow.
What would be the way to reduce the duration at 1 frame to make image to image ?
Yes you can, in the "Backend" below, look for the group "Duration / Number of Frames" on the left, there are 3 nodes there " Get_duration - Math - Set_frameNum ". Expand the "Math" node, replace the math line "(a*16)+1" to "1". When generating, select "SaveLastFrame" in "Other Options" in the "Post Processing" Group.
Thanks for creating this, but how are you supposed to use this for WAN models that already have lightning steps baked in?
you can go into the loader subgraph (click on the box with an arrow on the top right of the loader subgraph node). Inside, you will see "⚡️Lightx2v Distilled Lightning Lora " nodes, click those and hit "ctrl+b" and they will be disable.
Thoughts on integrating this for better motion? https://github.com/princepainter/ComfyUI-PainterI2V
i tested it before. when using the new lightning lora, motion were pretty great already. Also its an additional custom node that is not readily available in comfyui extension manager. I could integrate it into the workflow in a separate download, for people who wants to try it.
@Lannfield Thanks for the fast answer. Yea, that would be great! This one also seems interesting: https://github.com/XT-404/XT-404_SKYNET Maybe also worth trying. Apparently it helps improving the clarity even more.
@vellane i will think about, i'm also trying not to add too much custom nodes, especially custom node that are not readily available in the comfyui manager or commonly use. but thanks for the suggestion.
The most comprehensive and user-friendly I2V workflow I've tested.
It has everything, from video to audio.
The question now is: how can we make it even better!?
Incredible.
ty :)
+1 incredible face consistency with this workflow. any chance you will make for wan 2.2 animate?
Hi, love your project, it's one of the few I can get my head around.
How do we do batch images, is there a switch or something? I filled the node with my path, but it keeps using the single image node. Thanks
there are instruction in the batch workflow. there is a switch. in "🖼️📚I2v Mode Select"
@Lannfield Thank you for your quick responce. Sorry for not getting it, I see the '🖼️📚I2v Mode Select' but I don't see a switch: https://ibb.co/wFy95wLG
@BiscuitTinMan you probably have some conflicting custom node that causes it to be blank. Comfyui-Swann is one of them. you need to disable those custom nodes and restart comfyui
@Lannfield Thank you, I'll give that ago.
@Lannfield So deleing the Swvan did work to bring up the switch controls, but after it finished the Rife VFI, I get this msg:
VHS_MergeImages
'NoneType' object has no attribute 'shape'
Sorry for the hassle, but thank you for the help so far.
@BiscuitTinMan can you let me know which batch workflow you are using ? i will look at it tmr. its getting late here. sorry
@Lannfield Yes, it's the 🍥LF i2v(Only) +Batch WF V1.1, and sorry, I didn't think about the time difference. Thank you for all your help. G.Night
@BiscuitTinMan no worries, i will dm you, seems working on my end.
Anyway to add this via Pinokio in the Wan 2.2 setup or is it meant to be utilized with comfyui only
Sorry, I have never use pinokio before, so I’m not sure.
Hi, I love your workflow but I have one question. I cannot fix the mmaudio that show Failed to find the following ComfyRegistry list. I try to delete node and clone again but got the same issue. or update from comfy manager that got the same.
Try this after you clone( make sure you close you comfyui):
Go to your ComfyUI MMAudio custom node directory.
🗂️ ComfyUI\custom_nodes\ ComfyUI-MMAudio\
Right-click in the directory and select "Open Terminal"
Enter(with “):
"C:\Your\Directory\ComfyUI\.venv\Scripts\activate.ps1"
Then:
pip install -r requirements.txt
Or
If you are using portable:
Go to your ComfyUI, "ComfyUI_windows_portable" folder (where your run_nvidia_gpu.bat file is).
Right-click in the directory and select "Open Terminal"
Enter:
python_embeded\python.exe -m pip install -r ComfyUI\custom_nodes\ComfyUI-MMAudio\requirements.txt
I really appreciate it. that many help to me. <3
Thanks for the workflow, works great for i2V.
But there is an issue at least for me is that the SaveLastFrame and Frame Select is not working, enabling them didn't save any file to disk and didn't show any frame I can select. I have tried genrating preview and upscale (final) version of the same clip, those frame options are not working?
Anyone has the same issue?
wonderful workflows. i was wondering is there a way to disable/bypass the lighting loras? i couldn't find a setting to do that. cause i have unets that have them build in already i would like the use ( for example Dasiwa BoundBite models).