<div align="center">

Distil a photograph into calm, clear and quietly vivid geometry

<a href="README.md">简体中文</a> · <strong>English</strong> · <a href="README.ja.md">日本語</a> · <a href="README.ko.md">한국어</a> · <a href="README.ar.md">العربية</a>

</div>

Sample gallery

16:9 landscape left–right samples

| sample-05 | sample-06 | |---|---| | | | | | |

3:4 portrait top–bottom samples

| sample-09 | sample-10 | |---|---| | | | | | |

The eight works above comprise four 16:9 landscape left–right samples and four 3:4 portrait top–bottom samples. Every work was generated independently by Panel 102 from its own original brief; no artwork from another numbered Panel is reused. The samples use intelligent English copy.

<!-- xxd-human-intro:start -->

What it solves

Panel 102 is for preserving a photograph's recognisable subject while escaping cluttered backgrounds, literal redraws, and cheap filters. It reduces the subject's defining contour, structure, pose, and relationships to a few geometric colour fields, creating a quiet, modern, and vividly healing editorial poster within clear whitespace.

Best for

• making a busy photograph read clearly by retaining only the most recognisable contour, structure, pose, and relationship;
• achieving a modern premium feel without coldness through positive–negative shape, hierarchy, and generous whitespace;
• replacing muddy or cheap filter colour with bright, pure, warm hues derived from the source;
• top-bottom, left-right, design-only, multi-ratio, wallpaper, and directory-batch delivery;

Quick start

Use XXD Panel 102 on this image and recommend the most suitable composition and size first.

<!-- xxd-human-intro:end -->

<!-- xxd-panel-benefit:start -->

Quick fit check

| What you need to know | What this style gives you | |---|---| | What you get | A modern editorial poster that carries a photographic memory through a few geometric shapes, clear colour, and rational whitespace | | Recognisable signature | geometrised subject, flat colour fields, clear contours, fine micro-annotations, pale ground, and expansive whitespace | | How it respects the source | The object's defining features, pose, and relationships remain intact, so the design still maps immediately to the current photograph | | Where it works | Brand content, exhibition and architectural posters, portrait or pet keepsakes, covers, social content, and device wallpapers | <!-- xxd-panel-benefit:end -->

Usage tips

• Start with one clear photo: choose a source whose subject, action, and relationships are easy to recognize before choosing the delivery format.
• Join the parameters in one sentence: say “top-bottom / left-right / design-only + 16:9 / 3:4 / phone wallpaper”; you can also name desktop, tablet, or smartwatch sizes.
• State what must stay: identify the people, objects, actions, relationships, and copy to preserve, while leaving room for the style to design the layout.
• Choose a text mode: let the model write from the image, lock exact wording with --text exact --copy, or remove text completely with --text none.
• Clarify reality and design regions: for top-bottom or left-right, say which region keeps the photograph and which region is redesigned; for design-only and wallpapers, say that the whole canvas is redesigned.
• Test one image before batching: confirm mode, ratio, text, and language on one source, then reuse the settings for a folder; change one variable per iteration.

Get started

git clone https://github.com/nevertoday/xxd-panel-102.git
mkdir -p ~/.codex/skills
ln -s "$(pwd)/xxd-panel-102" ~/.codex/skills/xxd-panel-102

You can also install it directly with npx skills:

npx skills add https://github.com/nevertoday/xxd-panel-102 --skill xxd-panel-102

The command fetches the repository from GitHub and installs the same-named Skill for the current agent. To install it in the user-level Codex Skills directory, append --global --agent codex --yes.

Claude Code users may link the same folder under ~/.claude/skills/xxd-panel-102. Restart the agent session after installation.

$xxd-panel-102
Use this photograph, ask me for the modes and copy setting, then generate fresh raster outputs.

Full specifications: Skill workflow · source archive · English runtime adapter · Chinese runtime adapter

Original prompt · Five languages

[简体中文](references/

…