Identity lock
SCAIL‑2 reference mask + CLIP Vision keep the face, body and outfit consistent across the whole 30–60 fps clip. No morphing.
NEW · WAN 2.1 14B SCAIL + SAM3 · v2 release
Copy any video's motion onto your AI model.
One ComfyUI file. Drop in a reference clip + one photo of your model — out comes a 30–60 fps video with the same face, same outfit, every move matched.
- 2‑minute setup — drag the JSON in, pick 5 models, run
- Prompt pack + video tutorial included
- Me on Discord 24/7 until your first render works
4.9/5 from 300+ creators
🔒 Secure Stripe checkout
⚡ Instant download
NEW · WAN 2.1 14B SCAIL + SAM3 · v2 release
OUTPUT · 30–60 fps · mp4
rendering
0%
30–60fps output
2 minto set up
"Insane results"
the workflow · drag to rotate
drag the graph
scroll
Reference videos
Made with this workflow.
Real outputs — one photo of the model, one driving clip, zero manual keyframes.
Built on the strongest open video stack
Without it
Hours lost fighting nodes.
- Missing custom nodes, red errors, wrong model versions
- Motion drifts, hands melt, face changes every frame
- Reference video and model never line up
- Tutorials on YouTube are outdated in 2 weeks
→
With our workflow
Drag. Drop. Run. Done.
- One JSON — every node, model path & setting pre‑wired
- SAM3 + SCAIL‑2 mask keeps identity locked the whole clip
- Pose strength, frame count & resolution ready to tweak
- Tutorial + prompt pack + me on call 24/7
How it works
From purchase to first video
From purchase to first video
in under 5 minutes.
01
Checkout in 30 seconds
Pay securely with Stripe (card, Apple Pay, Google Pay). You land on your private download page instantly.
02
Drop the JSON into ComfyUI
Drag icymark_motioncontrol.v2.json onto the canvas. Follow the 2‑minute tutorial for model paths. That's the whole setup.
03
Upload video + model image
Any TikTok / dance / talking video as the driver, one photo of your AI model as the target. Paste a prompt from the pack.
04
Hit Run
She mirrors the motion exactly at 30–60 fps — head, hands, hips, hair. Export, post, repeat. Unlimited generations, forever.
The workflow
28 nodes. Pre‑wired.
28 nodes. Pre‑wired.
Zero guesswork.
This is the exact graph you receive — the same one running on our RunPod pods every day. Hover the hotspots to see what each block does. (Labels are blurred on purpose — you get the readable file after checkout.)
🔒 Preview intentionally blurred — the full, readable graph ships with your purchase
Under the hood
Three inputs. One perfect output.
● REC
Driving videoany motion source
Model image1 photo, any AI model
"A blonde hair woman dancing"(depends on your model — I can send you a prompt tailored to your model when you contact me on Discord)
drag to rotate · live pose tracking
ICYMARKpose tracking · identity lock
30 fps or 60 fps
Why creators switch
Everything that makes motion control
Everything that makes motion control
actually work.
Frame‑perfect motion
Dual SAM3 Video Track pipelines extract the human pose from the driver and the target, so every move maps 1:1.
4× faster renders
lightx2v step‑distill LoRA + fp8 weights. Full 30–60 fps clips render in minutes on a single RunPod GPU instead of an hour.
One‑knob control
pose_strength, length, shift, resolution — the only sliders you'll ever touch are labelled and explained.
Prompt pack included
Tested positive prompts for dance, talking, fashion & product shots, plus the exact negative prompt that removes artifacts.
24/7 human support
Stuck on a node? Message me any time — I set this up with buyers personally until the first video renders.
Reviews
Creators who stopped fighting
Creators who stopped fighting
ComfyUI and started posting.
4.9average · 312 verified buyers
JD
Jake D.AI model agency · UK
Spent 3 weeks trying to get SCAIL working myself. This ran on the first try. The mask node setup alone is worth the money.
Verified purchaseMP
Marek P.Content creator · Slovakia
Insane results. My model copies TikTok dances 1:1 now, face stays identical the whole clip. Support answered me at 2am lol.
Verified purchaseRK
Rahul K.Freelance video
The lightx2v LoRA config cut my render from 40 min to 9. I recouped the price with one client job.
Verified purchaseTS
Tomáš S.AI model agency · Czech Republic
We run 6 models. Bought once, now every VA on the team uses the same JSON. Setup tutorial is literally 2 minutes.
Verified purchaseTW
Tom W.RunPod user
Works perfectly on a 4090 pod. The prompt pack negative prompt fixed the melting hands I had for months.
Verified purchaseMK
Max K.Fashion brand
Used it for product try‑on videos. Motion is clean, no jitter between frames thanks to the previous_frames setting.
Verified purchaseCB
Chris B.Agency owner
Honestly expected a random JSON. Got a full system with links to every model and a human who replies instantly.
Verified purchaseLR
Luca R.Creator
Smooth 60 fps output in one go is a game changer. My old workflow stuttered and stitched badly.
Verified purchaseDM
Dan M.Video editor
Best €69 I've spent on AI tooling. Period. Lifetime updates too — got v2 for free.
Verified purchaseNH
Nick H.AI influencer studio
Identity lock is real. Same face, same body across every clip. Clients can't tell it's AI.
Verified purchase
Pricing
One payment. Unlimited videos.
Launch price · limited
ICYMARK Motion Control Workflow v2
ComfyUI · WAN 2.1 14B · SAM3 · lifetime license
€129
€68.99
one‑time · no subscription
10% off applied · ends in 30:00
- Full workflow JSON (28 nodes, pre‑wired)
- 2‑minute video setup tutorial
- Prompt pack (positive + negative)
- Model download list with direct links
- 24/7 personal support until it runs
- Lifetime updates — every new version free
- Commercial use allowed — unlimited generations
Secure checkout by Stripe. Instant download page.
Digital product delivered instantly — by paying you consent to immediate delivery and waive the right of withdrawal. Refund policy
Digital product delivered instantly — by paying you consent to immediate delivery and waive the right of withdrawal. Refund policy
✓
Works‑or‑I‑fix‑it guarantee. If it doesn't render on your setup, I'll personally get it running — 24/7.
What exactly do I receive after payment?
A ZIP with icymark_motioncontrol.v2.json (the ComfyUI workflow), the prompt pack, a model download list with direct links, and the 2‑minute setup video. Plus my direct contact for 24/7 support. It appears on your download page right after Stripe checkout.
What hardware do I need?
ComfyUI with a GPU that has 24 GB+ VRAM (RTX 3090 / 4090 / A5000+). No GPU? It runs perfectly on a RunPod pod for about $0.40/hour — the tutorial shows the exact pod template we use.
Which models does the workflow use?
It runs on the open WAN 2.1 14B video family with SAM3 for body tracking, plus a speed LoRA, VAE, text and vision encoders — all free & open. The exact files, versions and direct download links are in the model list you get after checkout.
Is the setup really 2 minutes?
Yes — if the models are already downloaded. Drag the JSON in, install missing nodes with ComfyUI Manager (one click), select the models in 5 dropdowns, done. Downloading the models the first time takes longer depending on your connection.
Can I use any video as the driver?
Any clip with one visible person — TikTok, dance, talking‑head, walking, product demos. Best results with a clear full‑body or upper‑body shot at 30 fps. The workflow caps at 220 frames (≈7 s) per run; chain runs for longer videos.
Does the AI model's face stay consistent?
That's the entire point of the SCAIL‑2 reference mask + CLIP Vision path. Identity is locked from your start frame and held across all frames. Buyers use it for the same AI model across hundreds of videos.
Can I use the videos commercially?
Yes. Unlimited generations, commercial use included, no watermark, no monthly fee. The only thing you can't do is resell or redistribute the workflow file itself.
What if it doesn't work for me?
Message me. I'll jump on it 24/7 and personally get it rendering on your setup — that's the guarantee. Digital products are non‑refundable once downloaded, so I make sure it works instead.
Your model. Any motion.
Tonight.
Stop rebuilding graphs. Drop in one JSON and post your first motion‑controlled video today.