nuyoah-ai-works/nuyoah-xiezhen-prompt

What it solves

This project provides a specialized prompt engineering system for portrait photography and AI image generation. It solves the problem of maintaining consistency in character identity and photographic style when recreating a photography plan across different shots or characters, preventing the common issue where AI models over-polish images or ignore the specific lighting and focus details of a reference photo.

How it works

The system operates as a "Skill" that can be integrated into an AI agent. It uses a structured workflow to separate character identity (facial anchors) from the specific expression and action of a scene. When recreating a photography plan, it reconstructs the observable photographic elements—such as light sources, camera angles, exposure, and background hierarchy—rather than simply swapping faces. It ensures that new prompts are compiled from the original reference and feedback, rather than relying on the previous generation's output to avoid style drift.

Who it’s for

It is designed for AI artists and photographers who need high-fidelity control over the composition, lighting, and texture of AI-generated portraits, specifically those wanting to replicate a professional photography "plan" or "shoot" with different characters or new angles.

Highlights

• Photography Plan Recreation: Allows users to swap characters while maintaining the same photographic style and conceptual plan across multiple new shots.
• Identity and Expression Separation: Decouples character facial features from the scene's emotional and physical reactions for better control.
• Style Fidelity: Specifically prevents "commercial polishing" (like unwanted rim lighting or hero compositions) when the original reference is intentionally under-exposed or out-of-focus.
• Prompt-First Delivery: Delivers full prompts, quality status, and traceability records directly in the conversation, treating image generation as an optional step based on user request.