AI Image Gen Prompt Engineering — Detailed Fashion + Pose Control

• URL: https://www.threads.com/share/BBCYQHb_i0/
• Date Saved: 2026-08-09
• Source: Threads
• Tags: creative-ai

Summary

By @akira06240. An extremely detailed AI image generation prompt demonstrating advanced techniques for photorealistic fashion photography with precise control over clothing replacement, pose, and anatomy.

What the prompt does

Generates a “Follow Me / 回頭拉手” POV photo where a person walks ahead, reaches back to hold the viewer’s hand, and looks back at the camera — wearing a specific ivory crochet fringe off-shoulder crop top.

Prompt Structure (reusable pattern)

1. Identity preservation — reference photo for face/body, explicit “do not change person”
2. Clothing replacement (highest priority) — detailed material, color, construction method, specific garment features, multiple language descriptions (Chinese + English)
3. Core action — precise body mechanics (which parts face which direction), natural motion cues
4. Hand anatomy (highest priority) — explicit correct anatomy rules + exhaustive list of common AI failures to avoid (extra fingers, fusion, wrong wrist angles)
5. Camera/perspective — POV type, distance, focal length constraints, foreground/midground/background composition
6. Style keywords — aesthetic direction (bohemian chic, resort style, festival style)
7. Negative prompts as prose — extensive “do not generate” list for clothing confusion (no bikini, no poncho, no cape, etc.)
8. Quality anchors — photorealistic, natural anatomy, editorial quality

Key Techniques

• Bilingual prompting (Chinese + English) for emphasis
• Repeating core instruction in multiple phrasings
• Explicit “this is NOT…” clarifications
• Prioritization markers (最高優先)
• Physics/gravity awareness for fabric and hair
• Anatomy-first approach for hands