outfit-director

Generates multi-outfit collage first-frame prompts and timed outfit-change video scripts for people and pets.

74|14|Updated Aug 3, 2026
One-click install
npx skills add https://github.com/liyue-aigc/outfit-director --skill outfit-director-liyue-aigc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: outfit-director
Source: https://github.com/liyue-aigc/outfit-director/tree/main/skill
Command: npx skills add https://github.com/liyue-aigc/outfit-director --skill outfit-director-liyue-aigc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Creating consistent outfit-change videos requires coordinating a collage first frame, a precise second-by-second timeline, and identity-consistent prompts across multiple outfits. This Skill turns a single outfit-change request into a complete, interlinked production package covering routing, parameter locking, image prompts, video timelines, and negative constraints. ## Core Features & Use Cases - Subject Routing: Automatically routes requests to female, male, pet, or mixed-collection tracks, locking identity anchors such as face, hairstyle, fur pattern, or breed features. - Two Video Modes: Beat-synced mode K (8-10 seconds, 5 outfits) and dance mode D (fixed 15 seconds, exactly 7 outfits with six side stickers that activate and permanently disappear). - Linked Outputs: Produces a first-frame collage image prompt, a full video timeline with 0.1-second change points, a condensed video prompt, and negative prompts enforcing identity consistency. - Use Case: Ask for a 15-second seven-outfit dance video for an original Chinese adult male, and receive the collage first-frame prompt plus a continuous choreography timeline with six outfit swaps at exact beat points. ## Quick Start Use $outfit-director to design seven urban outfits for a Chinese adult male and generate the 15-second continuous outfit-change dance first frame and video script.

Frequently Asked Questions about outfit-director

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a 15-second outfit-change dance video prompt?▼

Invoke the skill with a dance request and it generates a seven-outfit plan: one central outfit plus three side stickers on each side. It outputs a first-frame collage prompt and a continuous timeline with six outfit swaps at 2.0, 4.1, 6.3, 8.5, 10.8, and 13.0 seconds.

How do I generate outfit-change videos for pets?▼

The pet route locks species, breed, life stage, coat color, and other identification anchors instead of human features. Outfits must be lightweight and safe, never covering eyes, nose, or mouth, and dance is interpreted as natural movements like spins, paw lifts, and head tilts.

What is the difference between beat-synced mode and dance mode?▼

Beat-synced mode K runs 8-10 seconds with five outfits and four change points, defaulting to 9 seconds. Dance mode D is fixed at 15 seconds with exactly seven outfits, six side stickers that activate in order and permanently disappear, and continuous choreography.

Can I use a reference image to keep character identity consistent?▼

Yes, reference images take priority over default settings. The skill extracts visible identity anchors, outfits, pose, and lighting from the image, and all outfit versions in the video must remain the same subject with only clothing and pose changing.

What transition mechanisms are available for outfit changes?▼

The transition library includes 13 mechanisms such as click-and-fly-in (M1), sleeve occlusion (M2), fan遮挡 (M5), sticker flip (M8), beat-peak swaps (M10), and side-sticker activation (M13). A single video should mix no more than two main mechanisms.

When should I not use the seven-outfit dance format?▼

Avoid dance mode when the user wants quick beat-synced changes under 10 seconds or fewer than seven outfits; use mode K instead. Mixed collections of humans and pets should be split into separate videos rather than forced into one timeline.