Ascii Diorama logo

Ascii Diorama

Community
JDhruv14
ascii-diorama

Build tilt-shift ASCII miniatures, which are small, dense, living 3D worlds in Three.js drawn entirely as coloured monospace characters and seen from high above with the top and bottom of the view blurred like a scale model. Use when someone asks for an ASCII scene, ASCII shader or renderer, glyph-rendered or text-mode 3D, an ASCII diorama or miniature, a tilt-shift ASCII effect, or a dense little world (village, harbour, coast, forest path, shrine, city block) in this style. The result is a live web page that animates and loops.

Overview

PublisherJDhruv14
Repositoryascii-diorama
Skill nameascii-diorama
Stars
54
Forks
2
Bundled files
3
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 3 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by JDhruv14 on GitHub. Read the source before you install it.

Installation

Install the Ascii Diorama AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/JDhruv14/ascii-diorama/tree/main
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/JDhruv14/ascii-diorama.git \
  .claude/skills/ascii-diorama
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ascii Diorama in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Ascii Diorama on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Ascii Diorama is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

ASCII diorama

Build a small, dense, animated 3D world and draw it as coloured monospace glyphs, seen from high above through a long lens. Each glyph cell takes its colour from the 3D render, and its character from what the surface is, how bright it is and which way its edges run. The middle band is sharp text; above and below, it melts into a soft painted blur, like a tilt-shift photograph of a model.

The result is a live scene in the browser that animates and loops forever.

The look

  • Camera: about 52 to 58 degrees down, FOV about 23, around 70 to 80 units from the subject. It sweeps slowly back and forth over about 14 degrees of yaw with eased turns and a slight push in. Nothing cuts.

  • Composition: one leading line runs diagonally through the view, such as a stair path, shoreline, river or street. A single saturated focal colour sits against calmer greens, golds and creams. A strong focal colour is one like vermilion gates, striped parasols or a yellow tram. Something different fills the top band (a canopy, rooftops) and the bottom band (water, a plaza) so the blur has colour to work with.

  • Density: everything is many small things. That means tens of thousands of grass blades, thousands of leaf cards clustered into tree crowns with dark gaps, flower cards in noise-shaped patches, rows of repeated props (lanterns, steps, posts) and rocks along the edges. Sparse, smooth, low-poly scenes read as toys.

  • Life: wind moves the grass and leaves. Water rolls in bands of foam and pulls back up the shore. Petals or motes drift across. One small creature or figure walks through the focal area.

  • Glyphs:

    • grass in / | \ strokes that lean with the wind
    • foliage in . : * + o
    • stone and steps in — = #
    • water in - ~ =, with foam in * % #
    • creatures in $ &
    • strong edges switch to | / — \ along their direction

    The cell background is a dim, blurred copy of the render, so gaps between glyphs still carry colour.

Files

  • templates/scene/main.js: the complete starter miniature, which is red gates running down a green cliff to a turquoise sea. It includes stone steps, lanterns, gold and red trees, white flower patches, rocks with foam, a walking fox and drifting petals. Every system you need is in it: terrain bake, sea shader, instanced grass, flowers and leaves, path props, the glyph pass and tilt-shift. It runs live and loops; Space pauses.
  • scripts/fetch-assets.sh <scene-dir>: downloads the starter's textures and fox model.
  • scripts/serve.mjs <scene-dir> [port]: bundles the scene and serves it, rebuilding on every reload.

Workflow

  1. Start from the template and change the subject. Don't ship the template's subject unchanged. Keep the systems and swap what they build, so it becomes your own world. For example:

    • a fishing harbour with a breakwater path and boats
    • a hill village with a cobbled lane
    • a waterfall gorge with a footbridge
    • a lighthouse point
    • a desert oasis
    • a city rooftop garden
    bash
    mkdir -p <scene> && cp ~/.claude/skills/ascii-diorama/templates/scene/* <scene>/
    bash ~/.claude/skills/ascii-diorama/scripts/fetch-assets.sh <scene>

    It needs three and esbuild resolvable from the folder (a parent node_modules works).

  2. Know where things live in main.js:

    SectionWhat it controls
    coastZ, rawH, ROCKSthe land shape and shoreline
    PATHthe leading line, as control points from top to bottom of the view
    buildTerrainbakes ground colour and glyph class from real textures, zoned by height, path distance and noise
    buildSeathe water: depth colours, rolling foam, swash, caustics
    buildGrass, buildFlowers, buildTreesthe dense fill
    buildTunnel, buildLanterns, buildShrine, buildRocksthe props along and around the path; replace these with your subject's props
    CLS and the cls<... branches in asciiFragthe glyph rules per surface
    placeCamera, TARGET, DURATIONthe sweep and its period
    tiltFragthe focus band and blur strength
    COLS, ROWS, W, HHglyph grid and canvas size
  3. Gather assets with clear licences. Prefer CC0:

    • Poly Haven textures and HDRIs: https://dl.polyhaven.org/file/ph-assets/Textures/jpg/1k/<id>/<id>_diff_1k.jpg, with the list at https://api.polyhaven.com/assets?t=textures.
    • Khronos glTF sample models: read each README's Legal section.

    Put downloads in their own folder, never run anything from them, and keep a list of each asset's source and licence. Avoid copyrighted characters.

  4. Look at it while you build.

    bash
    node ~/.claude/skills/ascii-diorama/scripts/serve.mjs <scene>

    Open the printed URL, reload after each edit, and take screenshots to judge it. Check that the focal element and the leading line sit in the sharp band, that the top and bottom bands hold colour, and that large areas sit in the middle brightness range so their glyphs vary.

  5. Check motion, not just stills. Pause with Space at a few moments, or take several screenshots in quick succession. Glyphs flipping every frame read as strobing, so slow down the noise that drives them. The creature should step, not slide.

  6. Hand it over. The scene folder runs with serve.mjs. For a static copy, index.html, build/main.js and assets/ are everything the page needs, and they can go on any static host.

  7. Report: say how to view it, list each asset with its source, licence and any credit needed, and name any weak spots you saw.

Tuning that works

  • Grid 200×100 on a square canvas (cells of about 7×14 px). Sample the scene at the cell's mip level so each glyph averages its footprint.
  • Grade: tone curve, saturation 1.2 to 1.4, glyph colour × 1.3, cell background × 0.2 to 0.35. Raise the background for large bright surfaces such as sand or snow so they stay light.
  • Tilt-shift: focus around 0.42 to 0.47 of the height, blur starting about 0.27 away from it, up to 26 to 30 px on a golden-angle disc.
  • Light: a low-ish sun from the front-left with 4096 shadow maps, so canopies throw dappled shadow onto the grass and props cast readable shadows. Use a warm hemisphere fill.
  • Props on a path: space them at about the prop width, never tighter. Use a long gentle S rather than tight switchbacks.
  • Anything time-driven should loop cleanly or ping-pong, so the live view never jumps.

Pitfalls seen before

  • Props lying along the path instead of across it: with local x as the wide axis, use rotation.y = Math.atan2(tangent.x, tangent.z).
  • Tight switchbacks under wide props: they pile into a claw shape.
  • Black squares spreading through the image: NaN in the half-float target, which mipmapping spreads. Lit skinned glTF meshes are the usual cause, so use MeshBasicMaterial for small animated creatures.
  • Hulls or roofs black from above: flipped normals on extruded shapes. Use side: THREE.DoubleSide or unlit materials.
  • Dark tops on gates or roofs: these hide the focal colour from a steep camera, so colour the top surfaces.
  • Water blowing out to white glyphs: lower its base colours.
  • Orange or brown blob rocks: tint them cool grey and flatten them.
  • A glyph used in a rule but missing from GLYPHS: it produces undefined in the shader, so add it to the atlas string first.
  • Someone else's browser view: never navigate, resize or click in a browser someone is watching just to verify. Open your own tab or a separate headless browser instead.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Ascii Diorama AI skill do?

Build tilt-shift ASCII miniatures, which are small, dense, living 3D worlds in Three.js drawn entirely as coloured monospace characters and seen from high above with the top and bottom of the view blurred like a scale model. Use when someone asks for an ASCII scene, ASCII shader or renderer, glyph-rendered or text-mode 3D, an ASCII diorama or miniature, a tilt-shift ASCII effect, or a dense little world (village, harbour, coast, forest path, shrine, city block) in this style. The result is a live web page that animates and loops.

Why use Ascii Diorama on TypingMind?

Because you install it once and use it with any model. Ascii Diorama is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Ascii Diorama in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/JDhruv14/ascii-diorama/tree/main. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Ascii Diorama?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Ascii Diorama?

As many as you like. As long as a model supports skills, you can use Ascii Diorama with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Ascii Diorama AI skill free?

Yes. It is published on GitHub by JDhruv14 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇