- Home
- Skills
- Agents & Orchestration
- muse video prompting
Works with the AI Agents you already use
muse video prompting
Generate drift-proof, cinematically consistent video prompts for Meta Muse Video using Reference-Lock formulas.
Free
muse video prompting
Example session with this skill installed
Animate this photo of a woman in a blue jacket. She should look at a coffee cup, pick it up, and take a sip. Keep her face exactly the same. 16:9 ratio.
- Read your context and instructions
- Compiled the muse video prompting
I have generated a Reference-Lock prompt for Muse Video. It uses a static framing to maximize identity stability while she interacts with the prop.
Muse Video Prompt
Use the woman in the reference image exactly as she appears, do not redesign her. She is a woman with pale skin, a sharp jawline, shoulder-length blonde hair parted on the left, thin eyebrows, and a small silver nose stud. She is wearing the exact same navy blue quilted jacket from the reference image, same fit, same texture, same collar. She sits at a wooden cafe table in soft morning light coming from frame-right, as in the photo. With her right hand, she reaches for a white ceramic coffee cup at frame-center, lifts it to her lips, and takes a slow sip, keeping the cup from covering her eyes. The camera stays on a static tripod at eye level, framing her in a medium shot with a blurred cafe interior in the background. We hear the ambient clink of spoons and low chatter of a coffee shop. Keep her face, hairstyle, skin tone, jacket, and nose stud identical to the reference image throughout the entire video. Do not morph her face, do not age her, do not change her body shape, do not swap her outfit. She ends with the cup at her lips, looking down toward the table. Wide 16:9 cinematic video. STOP.
Next steps
- Upload the original reference photo alongside this text prompt in your Muse interface.
- If the jacket texture begins to drift, add the specific material (e.g., "polyester weave") to the Wardrobe Lock.
- To extend this into a sequence, let me know if you want a reverse shot of what she is looking at.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop video models from redesigning your characters mid-shot. This skill provides a specialized prompting framework for Meta's Muse Video and similar reference-conditioned models. It solves the common pain of identity drift, wardrobe mutation, and broken physics by applying a strict Reference-Lock Formula that transcribes visual data into model-readable constraints.
What it does
- Identity Locking creates a descriptive anchor for facial features, skin tone, and hair to prevent character morphing.
- Cinematic Continuity enforces the 180-degree rule, eyeline matches, and screen direction across multi-shot sequences.
- Drift Prevention uses specific syntax like the "STOP" rule to force clean hard cuts and prevent shot blending.
- Audio-Video Coordination utilizes Muse's native audio capabilities for synchronized foley, ambience, and short dialogue.
- Motion Control scaffolds action using a "shot ladder" to ensure physics remain stable during movement.
How it works
- Transcribe the reference photo into a five-point identity lock sentence.
- Define the screen geography, naming which side of the frame the character occupies and where they face.
- Sequence the shots using the
STOP. HARD CUT.syntax to maintain distinct framing. - Apply the Return-to-Anchor ending to ensure the final frame matches the original reference.
Frameworks & tools
Specifically designed for Meta Muse Video and Muse Image reference-conditioning. Compatible with any video model that supports text-based identity locks layered over image inputs.
Why this beats prompting it yourself
Generic prompts often lead to characters "melting" or changing clothes between cuts. This skill applies professional film grammar and technical syntax that forces the model to respect spatial logic and visual consistency that naive prompting ignores.
Use cases
- Character Consistency keeping the same person and outfit identical across a 10-second multi-shot clip.
- Physics Stabilization animating realistic interactions with props like fruit or phones without face occlusion.
- Cinematic Storytelling building complex scenes that respect the 30-degree rule and eyeline matches.
- Drift Diagnostics identifying why a face morphed and applying specific technical fixes.
Known limitations
Works best with slow, simple physics. Fast motion like running or dancing is prone to artifacts. Dialogue is limited to short phrases due to current sync gaps in Muse Video.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Optimized for slow motion and 2-beat actions. Fast motion and long dialogue may cause sync issues. Requires high-quality reference photos for best results.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 5 days ago
- Free with an Agensi account
Permissions
No special permissions declared or detected