Home
Creators
Search Recent Missing Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

(For SOTAI Squad) Wan Animate Is Dead? SCAIL Motion Transfer in ComfyUI (Low VRAM Workflow) (Patreon)

Published:
2025-12-23 16:26:40
Imported:
2026-06
haven't archived this post yet. have a subscription? use the importer!

Downloads

  • SCAIL - Day1 - SOTAI (fixed JWInteger issue).json

Missing 1 full-res photo, 1 file, as this post has not yet been imported.

Files

Previews only

Wan Animate Is Dead? SCAIL Motion Transfer in ComfyUI (Low VRAM Workflow)

Wan Animate Is Dead? SCAIL Motion Transfer in ComfyUI (Low VRAM Workflow) Wan Animate is great, but have you ever gone crazy trying to fix "mutated" limbs or characters changing faces during fast motion? Today, I’m bringing you the ultimate solution to fix that once and for all: SCAIL. Coming straight from the research team at Tsinghua University, SCAIL is currently rated by the community as the most powerful Motion Control tool for the Wan Model. From complex camera movements to multi-character interactions, SCAIL handles it all with shocking smoothness. 🔥 In this video, we will: Discover what SCAIL is and why it completely outperforms Wan Animate. Step-by-step guide to building the SCAIL Workflow in ComfyUI. How to run animations on 12GB VRAM (or less) Real-world Tests: From 3D characters and real humans to difficult shots like couple dancing or moving cameras. Don't forget to Smash that Like button, Subscribe, and hit the bell so you don't miss the latest AI tools! 🔗 Related Videos: - BindWeave workflow Day 1: [https://youtu.be/QsG0CRAxIHw] - Wan2.2 Animate workflow: [https://youtu.be/tkGVhU5hhiI] - Wan2.2 low VRAM : [https://youtu.be/WU9rr04_D4Y?si=pYVwjFfPoF6W8Kut] - ComfyUI tutorials: [ https://www.youtube.com/playlist?list=PLPb4yHX_Xwzf4oY77qD1zxW_zExbh3bXw] 📌 All Workflow: https://www.patreon.com/c/SOTAI #SCAIL #ComfyUI #WanModel #AIVideo

Comments

george jones

getting this error "line 82, in sageattn assert headdim in [64, 96, 128], "headdim should be in [64, 96, 128]."" any idea how to fix?

Stephen zumstein
>>197603151

try changing the attention mode in the wanVideo Model loader to sdpa. you will have to instal sage to use it. its meant to be faster but sdp[a is more stable for WAN

Douglas Chiu

the "JWInteger" custom nodes cannot be import, try several time still got missing

Stephanie Lawrence

help please. I am running into some issues with the workflow after installing missing nodes 95 LayerUtility: CropByMask V2 Conflict with ComfyUI_Swwan 96 LayerUtility: CropByMask V3 Conflict with ComfyUI_Swwan 129 LayerUtility: ImageScaleByAspectRatio V2 Conflict with ComfyUI_Swwan 151 LayerUtility: RestoreCropBox Conflict with ComfyUI_Swwan

FintuBoi
>>198090494

There is an error in the ComfyUI_Swwan node. Please delete it completely and reinstall it along with all its dependencies. This may solve the problem...

TheD1C

Edit: Nevermind, it was me. I should watch the video before trying to get my prereqs done. Fantastic context btw.

Buns McSmurf

I don't have a "detection" folder in my ComfyUI models folder. Do I just make a folder and name it "detection"?

Chris Ella

Has anyone had any luck with this on 8GB ? I'm stuck on "Sampling 121 frames at 928x544 with 8 steps" and that has been on 0% for 5-10mins and not moved, though my GPU is doing something. UPDATE: I found this - https://github.com/kijai/ComfyUI-WanVideoWrapper/issues/676 Basically increasing the blocks_to_swap and disabling the use_non_blocking. I'm testing it now, currently it's still slow BUT the memory usage pattern seems more positive so I'll leave it a while and see.

Natan Santos

Hello, I'm a new subscriber and I'm enjoying the content. I intend to continue my subscription. Could you tell me if, in this new workflow, it's possible to extend the video duration without compromising the resolution? I've already seen the video (FINALLY! ComfyUI WAN 2.2 Animate: Longer Videos + Low VRAM Workflow Guide), but it's a different workflow from this one. Could you help me? Is this possible in this new workflow? I'm still a beginner at creating nodes :(

SOTAI
>>200892119

Hello, yes, it can definitely be applied to this new workflow. However, it’s a bit more complex because the node parameters don’t support it directly, and you need to control the transitions to keep them smooth. I still recommend using the low-VRAM techniques I’ve shared so you can generate longer videos. With this workflow, I can generate an 20-second video on a 3090. Besides lowering the resolution, you can also review this video: https://youtu.be/sVzQosJZWk8

Rohan Furries

I have been trying to get this working but i get a prompt execution error, the KepStringLiteral, and onnxdetectionmodelloader seems to be missing. any idea how to fix that? i installed the necessary nodes on the comfyui manager

Killerpixeel

can't find onnxdetectionmodelloader and it didn't show up in the comfyui manger.

waleed gamer

invalid prompt: {'type': 'missing_node_type', 'message': "Node 'ID #243' has no class_type. The workflow may be corrupted or a custom node is missing.", 'details': "Node ID '#243'", 'extra_info': {'node_id': '243', 'class_type': None, 'node_title': None}} i hove 3 errors like this with different numbers how i can fix them

EVILWez
>>200190003

Do you have the settings you used. (could you paste them) I tried using the link you used, but nothing, I keep getting an error on the "onnxruntime_inference_collection.py:123" part. Specs: Total VRAM 8188 MB, total RAM 65273 MB pytorch version: 2.10.0+cu130 Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA GeForce RTX 4070 Laptop GPU : cudaMallocAsync Using async weight offloading with 2 streams Enabled pinned memory 29372.0 working around nvidia conv3d memory bug. Using pytorch attention Python version: 3.13.11 (tags/v3.13.11:6278944, Dec 5 2025, 16:26:58) [MSC v.1944 64 bit (AMD64)] ComfyUI version: 0.13.0 ComfyUI frontend version: 1.38.13 Windows 11 25H2 i notice one error: C:\ComfyUI\ComfyUI_windows_portable\venv\Lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py:123: UserWarning: Specified provider 'CUDAExecutionProvider' is not in available provider names.Available providers: 'AzureExecutionProvider, CPUExecutionProvider' warnings.warn( however i followed the guides on youtube to change the different architectures but it just hangs at Hangs here GPU 90% - I did go down on the model to Q2 even. 121 frames at 544x928 (Input sequence length: 61132) with 8 steps Generated new RoPE frequencies in the json = confyui interface, it highlighted green in the "wanvideowrapper"

justcheckingthings

I am getting this error with an RTX 5090 and CU12.8 Rope function: comfy Using SCAIL embeddings: ref_latent_pos: torch.Size([20, 1, 160, 92]) ref_latent_neg: torch.Size([20, 1, 160, 92]) ref_start_percent: 0.0 ref_end_percent: 1.0 pose_latent: torch.Size([20, 75, 80, 46]) pose_strength: 1.0 pose_start_percent: 0.0 pose_end_percent: 1.0 ---------- Sampling start ---------- 297 frames at 736x1280 (Input sequence length: 276000) with 8 steps Generated new RoPE frequencies 0%| | 0/8 [00:00

Hib00 Gaming

Hello, I have a big problem. Step 1 works without any issues. Step 2 generates the model perfectly, but in step 3, nothing happens; no video appears. It seems to skip the step completely, and I haven't been able to solve the problem. Thank you for explaining.

SOTAI
>>204151229

Hello, sorry for missing your message earlier. The issue you described sounds quite unusual. You might want to try restarting ComfyUI to see if that resolves it. Have you been able to fix it yet?

SOTAI
>>202705623

Sorry for missing your message earlier. Have you installed ComfyUI-WanAnimateProcess in your custom nodes yet?