colorize-image
Add plausible, context-aware color to a black-and-white or grayscale image while preserving identity, content, tonal structure, texture, era cues, and damage. Use for archival-style, family, editorial
SKILL.mdskills/colorize-image/SKILL.md
GitHubColorize Image
Catalog
- Display name: Colorize Image
- Categories: Utility
- Status: Ready
- Route label: GPT Image 2 contextual colorization
- Tone: gold
- Short description: Add plausible color while preserving the original photograph.
What this skill does
- Uses whole-image context to infer coherent skin, fabric, object, and environment colors.
- Distinguishes documented color references from plausible creative inference.
How to use
Provide grayscale image, known colors or references, desired realism, and disclosure needs.
Colorize a grayscale photo
Colorize this grayscale 1950s-style family photo naturally; preserve faces, grain, contrast, clothing, and background.
Workflow
- Record known colors and protected monochrome evidence; mark all unsupported colors as inferred.
- Use
gpt-imagev1.0, GPT Image 2 Medium, source, original ratio,batchCount: 1. - Request restrained context-aware color without sharpening, restoration, face changes, object additions, or modernized styling.
- Compare geometry, identity, tonal hierarchy, grain, damage, edges, and era plausibility.
- Reject neon casts, uniform skin, color bleed, changed objects, cleaned damage, or false claims of historical certainty.
When this Skill selects GPT Image 2, use
quality: "medium"andimageSize: "2K"by default; change either only when the user explicitly requests another tier.
Every GPT Image 2 call in this Skill defaults to Medium at 2K; use another tier only when the user explicitly requests it.
User-facing output
- Media type: One colorized image
- Default quantity: 1
- Content per image: Source content with plausible color
- Default layout: Source composition
- Model policy: GPT Image 2 Medium
- Downstream use: Family, archive-style, and editorial imagery