photo-skills

Reusable Codex skills for photographic and editorial image transformations.

This repository is the public source and installation entry point. GitHub-hosted skills are not automatically listed in Codex's global Skills search. Install a skill from its direct GitHub folder URL, or share/publish it from the Skills page in your Codex/ChatGPT workspace.

Find and install a skill

Search terms: Codex skills, Agent Skills, photo editing, image generation, Genshin-inspired anime photo, Hong Kong multiverse photo, cyberpunk city art, liminal film photography, torn-paper cartoon, travel sketchbook, graphite watercolor, cinnabar journey line, vintage fashion poster.

Included skills

HK Multiverse Photo

Transforms supplied Hong Kong, city, street, vehicle, skyline, or harbor photographs into bold multiverse-collision stills. It preserves the photographed geometry while combining a clear primary world, multiple incoming universes, flat 2D animation, dimensional 3D/cel rendering, and a real-photo anchor.

The skill includes five original rendering systems, source-hook collision layouts for wheels, vehicles, trams, streets, and harbor reflections, deterministic HK//VERSE SVG annotations, and a portable raster-dimension validator. It does not bundle or copy official characters, logos, or film frames.

Sketch Zeriol

Transforms any supplied photograph into a fixed vertical travel-sketchbook poster language: warm flax-ivory paper, truthful graphite construction, restrained source-colored washes, natural erased transitions, one scene-following cinnabar journey line, and one quiet typewriter caption.

People are redrawn completely, distinctive landscape colors remain visible but controlled, and every correction returns to the original photograph. The skill explicitly rejects torn paper, collage windows, hard frames, photo blocks, and post-processed vector lines. Once installed, attaching a photo without further instructions is treated as a request to generate this style.

Torn Cartoon Window

Transforms a supplied photograph into a tactile virtual-real composition: one natural torn or peeled paper opening reveals a coordinated hand-drawn 2D cartoon world underneath.

The skill preserves the real photograph outside the tear, anchors the opening to a photo-specific contour or structure, and uses believable paper fibers, thickness, curl, wrinkles, and contact shadows.

Vintage Fashion Plate Poster

Transforms a supplied model, outfit, portrait, or fashion concept into a refined vertical vintage fashion-plate poster with aged paper, centered editorial composition, restrained frame lines, controlled typography, muted print colors, and flat scanned-catalog texture.

The skill preserves identity, pose, garment construction, silhouette, accessories, and styling while removing modern photographic backgrounds and commercial gloss.

Liminal Journey Film Photo

Transforms supplied photographs or written scene concepts into quiet, wistful journey photography built around believable viewpoints: windows, mirrors, doorways, fog, distant figures, foreground occlusion, roads, trains, ferries, and open landscapes.

The skill preserves identity and scene geometry while using restrained gray-teal color, selective warm accents, natural optical softness, film grain, halation, and generous negative space. Explicit prompts such as “colder,” “dreamier,” “closer,” or “preserve the original framing” override its defaults.

Genshin-Inspired Photo

Transforms supplied people, landscapes, architecture, food, vehicles, pets, and objects into original nation-aware elemental anime open-world imagery while preserving the source identity, pose, function, composition, terrain, or construction.

The skill supports Mondstadt, Liyue, Inazuma, Sumeru, Fontaine, Natlan, and a cautious northern direction. It includes dimensional 3D anime character rendering, function-preserving clothing and equipment redesign, landscape anchor-strength controls, and a thumbnail recognition gate that prevents important world-building from disappearing at normal viewing size. No official artwork or private calibration images are bundled.

Install

Install one skill directly

Copy one of the Codex prompts from the table above. This is the recommended method when another user only needs one skill.

Install the full collection manually

Clone this repository:

git clone https://github.com/1173206772/photo-skills.git

Copy the skill into your personal Codex skills directory:

mkdir -p ~/.codex/skills
cp -R photo-skills/skills/torn-cartoon-window ~/.codex/skills/
cp -R photo-skills/skills/vintage-fashion-plate-poster ~/.codex/skills/
cp -R photo-skills/skills/liminal-journey-film-photo ~/.codex/skills/
cp -R photo-skills/skills/genshin-inspired-photo ~/.codex/skills/
cp -R photo-skills/skills/hk-multiverse-photo ~/.codex/skills/
cp -R photo-skills/skills/sketch-zeriol ~/.codex/skills/

Restart Codex if the skill does not appear immediately.

Usage

Attach a photograph and ask:

Use $torn-cartoon-window to tear a natural opening in this photo and reveal a coordinated cartoon world underneath.

The skill works best when the photograph contains a strong visual hook such as a pose, road, ridge, cloud, shadow, wheel, window, sign, facade, or motion line.

For a vintage fashion plate, attach a model or outfit photograph and ask:

Use $vintage-fashion-plate-poster to turn this model photo into a refined vintage fashion-plate poster.

For a quiet journey-film treatment, attach a photograph and ask:

Use $liminal-journey-film-photo to preserve this composition and make the atmosphere colder and more solitary.

For the fixed Zeriol travel-sketchbook treatment, install the skill and simply attach a photograph, or ask explicitly:

Use $sketch-zeriol to turn this photo into the signature Zeriol travel-sketchbook poster.

For an original nation-specific anime open-world conversion, attach a photograph and ask:

Use $genshin-inspired-photo to preserve this composition and transform it into a readable Mondstadt-inspired 3D anime open world. If the landscape feels too restrained, use Strong anchor strength.

For a bold Hong Kong multiverse still-image series, attach one or more city photographs and ask:

Use $hk-multiverse-photo to preserve these locations and create a coherent series with one strong primary universe, visibly different 2D and 3D incoming fields, and deterministic HK//VERSE annotations. Generate still images only.

Repository structure

skills/
├── genshin-inspired-photo/
│   ├── SKILL.md
│   ├── agents/
│   │   └── openai.yaml
│   └── references/
│       ├── character-rendering.md
│       ├── genshin-anchor-system.md
│       ├── nation-atlas.md
│       ├── prompt-recipes.md
│       ├── subject-recipes.md
│       └── visual-language.md
├── hk-multiverse-photo/
│   ├── SKILL.md
│   ├── agents/
│   │   └── openai.yaml
│   ├── references/
│   │   ├── annotation-system.md
│   │   ├── collision-layouts.md
│   │   └── universe-atlas.md
│   └── scripts/
│       ├── make_annotation_svg.py
│       └── validate_image.py
├── liminal-journey-film-photo/
│   ├── SKILL.md
│   ├── agents/
│   │   └── openai.yaml
│   └── references/
│       ├── prompt-recipes.md
│       └── style-language.md
├── sketch-zeriol/
│   ├── SKILL.md
│   ├── agents/
│   │   └── openai.yaml
│   ├── assets/
│   │   └── style-anchors/
│   │       ├── blue-salt-lake.png
│   │       ├── danxia-road.png
│   │       ├── dune-wind.png
│   │       └── grassland-gully.png
│   └── references/
│       └── style-system.md
├── torn-cartoon-window/
│   ├── SKILL.md
│   ├── agents/
│   │   └── openai.yaml
│   └── references/
│       └── tear-system.md
└── vintage-fashion-plate-poster/
    ├── SKILL.md
    ├── agents/
    │   └── openai.yaml
    └── references/
        └── visual-system.md