ref-to-image

Generate new scenes while preserving facial identity using reference-based anchoring.

Updated May 25, 2026
One-click install
npx skills add https://github.com/fuyucn/skills --skill ref-to-image
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ref-to-image
Source: https://github.com/fuyucn/skills/tree/main/ref-to-image
Command: npx skills add https://github.com/fuyucn/skills --skill ref-to-image

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill solves the common AI image generation issue where characters lose their identity, face, or unique features when placed in new scenes, styles, or multi-person compositions.

Core Features & Use Cases

  • Identity Anchoring: Uses specific P1/P2/PN identifiers to ensure the same person is generated across different prompts.
  • Multi-Person Isolation: Prevents face-swapping or identity mixing in group photos by enforcing strict reference-to-subject mapping.
  • Anti-Beautification: Overrides AI's tendency to create generic, plastic, or overly symmetrical faces, preserving natural human features and subtle asymmetries.

Quick Start

Use the ref-to-image skill to generate a new scene for the person in the attached photo while strictly preserving their facial identity and natural features.

Frequently Asked Questions about ref-to-image

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

FAQPage Schema
How do I maintain character identity consistency across different AI-generated scenes?

To maintain character identity consistency across AI-generated scenes, use reference-based anchoring with specific P1/P2/PN identifiers to ensure the same person is generated across different prompts while preserving their facial structure.

How can I prevent face-swapping or identity mixing in multi-person AI group photo compositions?

To prevent face-swapping or identity mixing in multi-person AI group photo compositions, enforce strict reference-to-subject mapping. This multi-subject identity isolation ensures each individual retains their unique facial features without bleeding into other characters.

Can I use a single photo reference for both portrait transformations and artistic style transfers?

Yes, you can use a single photo reference for both single-person portrait transformations and artistic style transfers. The reference-based anchoring mechanism strictly preserves the subject's facial identity even when applying diverse artistic styles to the generated output.

Why does my AI character's face change when I place them in a new scene or style?

AI characters lose their identity in new scenes because the generation model lacks strict facial structure preservation. Applying reference-based anchoring with specific identifiers solves this by locking the subject's unique features to the original input photo.