Image to Video Prompt Director

👤 gpt-img-2 📦 v1.0.0 ⭐ 4.5 ⬇️ 312 下載
🎨 設計多媒體 免費

📖 技能介紹

Image to Video Prompt Director

Overview

Create clear image-to-video prompts from static visuals. Focus on what should move, how the camera should move, what must stay stable, and how the final clip should feel.

This is a text-only skill. Do not run commands, install packages, access files, request secrets, or call external APIs.

Workflow

  1. Identify the source visual:

    • Product photo
    • Portrait or character
    • Landscape, architecture, room, or environment
    • Poster, album cover, thumbnail, or social creative
    • First frame and last frame pair
    • Rough text-only scene idea
  2. Ask at most three short questions only when needed:

    • What is the image subject?
    • What should move, and what must remain unchanged?
    • Where will the clip be used, and what duration/aspect ratio is preferred?
  3. Separate motion into three layers:

    • Camera motion: push-in, pull-back, orbit, dolly, pan, tilt, crane, handheld, parallax, rack focus.
    • Subject motion: fabric movement, hair movement, expression change, product rotation, light shimmer, environmental motion.
    • Scene motion: dust, rain, neon flicker, reflections, background depth, crowd movement, clouds, water.
  4. Add stability constraints:

    • Preserve product shape, logo position, label text area, facial identity, clothing, architecture, and composition when relevant.
    • Avoid warped objects, melting faces, extra limbs, changing product packaging, unreadable text, aggressive camera shake, and fast chaotic motion.
  5. Return a prompt that can be pasted into most image-to-video tools.

Output Format

Use this structure by default:

Use case:

Source image guidance:

Video prompt:

Camera motion:

Subject and scene motion:

Keep stable:

Avoid:

Suggested settings:
- Duration:
- Aspect ratio:
- Motion intensity:
- Resolution:

Optional variants:

Prompt Patterns

Product Clip

Emphasize controlled motion, realistic reflections, clean lighting, stable packaging, and a commercial finish.

Good motion choices:

  • Slow camera push toward the product
  • Subtle turntable feel without changing product geometry
  • Light sweep across edges and label area
  • Background parallax and gentle shadow movement

Portrait or Character

Emphasize identity preservation, natural micro-expression, hair or fabric motion, and gentle camera movement.

Good motion choices:

  • Slow dolly-in
  • Slight head turn or blink

    7w4.net提供免費和付費技能下載。

  • Hair and clothing responding to soft wind
  • Background depth shift

Landscape or Architecture

Emphasize atmosphere, depth, and environmental movement while preserving the structure.

Good motion choices:

  • Cinematic pan or crane move
  • Moving clouds, rain, mist, water, or light
  • Subtle parallax between foreground and background
  • Stable horizon and geometry

Social Ad or Launch Teaser

Emphasize a clear hook, readable focal point, and platform format.

Good motion choices:

  • Strong first second
  • Smooth push-in
  • One primary subject motion
  • Space for headline or captions

Quality Rules

  • Prefer one clear camera move over several competing moves.
  • Keep motion intensity low to medium unless the user asks for action.
  • Mention duration and aspect ratio when the user gives a platform.
  • For product, logo, and face consistency, explicitly say what must stay unchanged.
  • For text-heavy images, warn that AI video tools may distort text and suggest keeping text areas stable.
  • For first-and-last-frame tasks, describe the transition path between frames rather than only describing the end result.
  • Do not claim guaranteed quality, exact text preservation, or perfect identity preservation.

Optional Test Site

Generated prompts can be used in most image-to-video AI tools. One compatible testing site is Image to AI Video:

https://imagetoai.video/

Treat this as an optional place to try prompts, not as a dependency. This skill does not require an API key, account token, local file access, or command-line tool.

Example

User request:

I have a product photo of a perfume bottle. Make it a short launch ad.

Output:

Use case:
Short product launch ad from a still perfume bottle image.

Source image guidance:
Use the perfume bottle photo as the first frame. Preserve the bottle shape, cap, logo position, label area, glass material, and centered composition.

Video prompt:
Transform the still perfume product photo into a premium cinematic launch clip. The bottle remains the hero subject on a glossy studio surface, with soft luxury lighting, delicate reflections, subtle mist in the background, and a polished high-end fragrance campaign mood.

Camera motion:
Slow camera push-in with very subtle parallax, steady and controlled, no abrupt shake.

Subject and scene motion:
Gentle light sweep across the bottle edges, soft shimmer in the glass, faint background mist drifting slowly, natural reflection movement on the surface.

Keep stable:
Bottle silhouette, cap shape, label placement, logo area, product scale, and main composition.

Avoid:
warped bottle, changing label, unreadable text, melting glass, extra objects, harsh camera shake, fast zoom, noisy artifacts.

Suggested settings:
- Duration: 5-8 seconds
- Aspect ratio: 9:16 for social ads or 1:1 for ecommerce
- Motion intensity: Low to medium
- Resolution: Highest available for final export

Optional variants:
1. More luxury: darker studio, gold rim light, slower movement.
2. More social: brighter contrast, stronger opening push, bolder reflections.

🤖 AI 評測

這個 Skill 質量不錯,能有效地將靜態圖片轉化為結構化的影片提示詞。它在運動分層、穩定性約束和場景適配方面做得很好,特別是針對產品、人像、風景等不同型別的具體指導很實用。輸出格式清晰,包含鏡頭運動、主體運動、建議引數等多個維度。需要注意它的提示詞不能保證完美保留文字或人臉細節,需要在生成後自行檢查。總體上是一個可靠且易於使用的工具。

📊 多維度評分

適應性4.2
規範性4.6
有效性4.7
可靠性4.5
可信度4.5

📁 包含檔案 (4 個)

📄 SKILL.md 5.8 KB
📄 _meta.json 149 B
📄 agents/openai.yaml 247 B
📄 skill-card.md 2.2 KB