TikTok Fashion Single-Person Video Skill
Purpose
Create high-converting TikTok Shop US women's fashion AI videos.
This skill handles the complete workflow:
Product analysis → Target customer analysis → Scene concept → Outfit styling → First-frame design → Video script → Seedance video prompt → Quality control
The final video should feel like authentic TikTok UGC rather than a traditional commercial.
1. Product Analysis
Before creating any content, analyze the provided clothing product.
Identify:
- Garment category
- Color
- Material appearance
- Silhouette
- Fit
- Neckline
- Sleeve design
- Hem
- Length
- Texture
- Construction details
- Visual selling points
- Functional selling points
- Styling possibilities
- Likely US target customer
Never invent product features that are not visible or provided.
Product accuracy has priority over creative styling.
2. Target Customer
Determine the most suitable US consumer persona for the product.
Consider:
- Age
- Lifestyle
- Body type
- Fashion preferences
- Typical use occasions
- Purchasing motivation
The model and environment should feel believable for this customer.
Avoid generic AI influencer aesthetics.
The character should look like a real person who could naturally buy and wear this product.
3. Character Rules
Only ONE speaking person may appear in the video.
The same character identity must remain consistent throughout the entire video.
Maintain consistency in:
- Face
- Age
- Ethnicity
- Body shape
- Hair
- Skin tone
Do not accidentally generate another speaking character.
Background people may only appear when necessary for environmental realism and must never speak or become the focus.
If a character reference image is supplied, preserve that identity as closely as possible.
4. Styling
Build a complete outfit around the product.
The product must remain the visual hero.
Styling may include:
- Pants
- Skirts
- Shoes
- Bags
- Earrings
- Necklaces
- Scarves
- Sunglasses
- Watches
- Other appropriate accessories
Accessories should increase realism and fashion appeal without distracting from the product.
Do not alter the original garment design to make the styling work.
Create meaningful variation between different styling concepts.
Changing only the background is NOT sufficient variation.
Different concepts should vary in:
- Outfit combination
- Accessories
- Location
- Lighting
- Activity
- Mood
- Camera composition
5. Scene Design
Choose environments that feel believable for the target customer.
Possible environments include:
- Home
- Bedroom
- Walk-in closet
- Living room
- Kitchen
- Backyard
- Porch
- Café
- Shopping district
- Downtown street
- Suburban neighborhood
- Vacation environment
- Other realistic US lifestyle locations
Avoid repetitive scenes.
Each scene should have a reason for existing and should help demonstrate the product.
The environment must look lived-in and natural rather than like an AI-generated showroom.
6. Visual Style
Default visual language:
Authentic TikTok UGC filmed on a modern smartphone.
The image should feel:
- Warm
- Bright
- Natural
- Approachable
- Contemporary
- Slightly imperfect
- Social-media native
The subject's face must receive flattering light.
Avoid:
- Excessive saturation
- Artificial HDR
- Plastic skin
- Heavy beauty filters
- Cinematic commercial lighting
- Unrealistic depth of field
- Overly perfect studio composition
The result should look like a real creator filming content for TikTok.
7. First Frame
Before generating the video prompt, design the first frame.
The first frame must establish:
- Character
- Product
- Outfit
- Environment
- Lighting
- Camera position
- Body position
- Emotional tone
The garment must be clearly visible.
Do not hide important product details behind:
- Hair
- Hands
- Bags
- Furniture
- Accessories
The first frame should provide a strong and stable starting point for video generation.
If the supplied first-frame reference is a sketch, pencil drawing, storyboard, or other non-photorealistic reference:
The generated VIDEO must NOT inherit the sketch style.
The first video frame must immediately appear as real smartphone footage.
8. Performance
The creator should behave like a confident TikTok fashion creator talking to a friend.
Performance should include:
- Clear facial expressions
- Natural hand gestures
- Vocal emphasis
- Genuine excitement
- Emotional peaks
- Direct camera interaction
The performance should feel energetic and expressive without becoming theatrical.
Do not mechanically force reaction phrases such as:
"OMG" "Wait" "Stop"
Use them only when naturally appropriate.
9. Product Demonstration
The video must clearly demonstrate the garment.
Use natural actions such as:
- Turning slightly
- Adjusting the hem
- Touching the fabric
- Showing the sleeve
- Showing the neckline
- Showing the side silhouette
- Showing the back when useful
- Walking toward or away from camera
- Moving closer to camera
Avoid actions that AI video models frequently generate poorly.
Do not create complicated hand-object interactions unless necessary.
10. Detail Shots / Zoom-In
Every sales video must include intentional close-up or zoom-in moments.
Use these moments to show relevant selling points such as:
- Fabric texture
- Stitching
- Neckline
- Sleeve construction
- Waist shaping
- Hem
- Stretch appearance
- Draping
- Cut
- Fit
The close-up must have a clear product-selling purpose.
Do not zoom randomly.
Product details must remain physically consistent during the zoom.
11. Camera Language
The video should feel like smartphone TikTok content.
Camera movement may include:
- Gentle handheld movement
- Push-in
- Pull-back
- Small reframing
- Natural creator-to-camera movement
- Product-detail zoom
Avoid overly cinematic camera moves.
Avoid impossible camera paths.
Camera movement should help sell the garment.
12. Dialogue
Dialogue should sound conversational and native to TikTok.
Prioritize:
- Hook
- Personal reaction
- Product benefit
- Specific selling point
- Styling/use scenario
- CTA
Avoid corporate advertising language.
Avoid excessive adjectives.
Avoid sounding like a product listing.
Only the main character speaks.
Never generate an off-screen second speaker.
13. CTA
Every sales video MUST contain a CTA.
The CTA should feel natural rather than aggressively scripted.
Examples of CTA intent:
- Check it out
- Grab it while available
- Find it in the shop
- Try this color/style
- Add it to your wardrobe
Adapt the CTA to the actual product and script.
14. Seedance Video Generation
When creating a Seedance prompt:
Use the supplied first-frame image as the visual starting reference.
The generated character must preserve the intended character identity.
The product must remain visually consistent throughout the video.
Preserve:
- Color
- Pattern
- Cut
- Length
- Neckline
- Sleeve
- Fabric appearance
- Decorative elements
Do not allow the garment to transform during movement.
Specify:
- Duration
- Aspect ratio
- Visual realism
- Camera behavior
- Character performance
- Product demonstration
- Dialogue
- Timing
- Zoom/detail moments
- Ending CTA
The prompt should describe temporal progression clearly.
15. AI Failure Prevention
Before finalizing, actively check for common AI-video failures.
Prevent:
- Extra fingers
- Broken hands
- Face changes
- Character identity drift
- Clothing transformation
- Pattern changes
- Color changes
- Disappearing accessories
- Duplicated people
- Unexpected second speakers
- Impossible body movement
- Sudden camera teleportation
- Background morphing
- Object deformation
- Unnatural fabric physics
Simplify actions when necessary to improve generation reliability.
16. Final Quality Check
Before delivering a final concept or prompt, verify:
Character
- One speaking character only
- Identity consistent
- Appropriate for target customer
Product
- Garment remains accurate
- Main selling points visible
- No invented construction details
Styling
- Complete outfit
- Accessories appropriate
- Product remains hero
Scene
- Believable US lifestyle environment
- Not repetitive
- Appropriate lighting
Performance
- Expressive
- Natural
- TikTok-native
- Not robotic
Camera
- Smartphone UGC feeling
- Useful product close-ups
- No impossible movements
Sales
- Strong hook
- Clear selling points
- Natural CTA
AI Reliability
- Actions are realistically generatable
- Product consistency protected
- No second speaker
If any important item fails, revise the concept before final output.
17. Default Workflow
When the user provides a new product, follow this order:
- Analyze the product.
- Identify its strongest selling points.
- Determine the target customer.
- Create the character direction.
- Design the complete outfit.
- Choose a suitable scene.
- Explain briefly why the concept fits the product.
- Design the first frame.
- Create the first-frame generation prompt when requested.
- Create the Seedance video prompt when requested.
- Run the quality checklist.
- Revise any detected problems.
Do NOT automatically generate every deliverable at once if the user is working step-by-step.
Follow the user's current production stage.
For example:
If the user asks for a scene concept, stop after concept development.
If the user asks for the first frame, produce the first-frame specification.
If the user asks for the video prompt, continue to the Seedance video stage.
Always preserve decisions already approved earlier in the workflow unless the user asks to change them.