Iso Scene logo

Iso Scene

Organization
vm0-ai
iso-scene

Isometric editorial-magazine scene illustration in a locked flat-vector style — ultra-fine hairline outlines, monochromatic saturated background filling the canvas, and a single composed scene whose props themselves embody the theme. Trigger when users say /iso-scene, ask for an "isometric editorial illustration", a "scene illustration in the editorial machine style", or brief with palette + scene archetype + complexity.

Overview

Publishervm0-ai
Repositoryvm0-skills
Skill nameiso-scene
Stars
76
Forks
18
Bundled files
4
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.

  • 4 bundled files

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

  • Open source

    Published by vm0-ai on GitHub. Read the source before you install it.

Installation

Install the Iso Scene 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.
  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/vm0-ai/vm0-skills.git /tmp/vm0-skills
mkdir -p .claude/skills
cp -r /tmp/vm0-skills/illustration-template/iso-scene .claude/skills/iso-scene
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Iso Scene 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 Iso Scene 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 Iso Scene 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.

/iso-scene — Isometric Editorial Scene Illustration

Generates a single composed isometric scene in a locked house style. Each invocation varies the palette, scene archetype, and complexity; everything else stays locked. The defining commitment is that the scene IS the metaphor — never a generic filing-cabinet box with the theme labelled on folder tabs.

Prompt interpretation

The user supplies a short brief — typically a scene archetype (e.g. "city construction", "sky garden", "castle in the sky") plus an optional palette. It is your job to flesh that brief out into a fully composed scene with theme-native props. Do not stop to ask. Do not default to a generic container shape.

For each invocation:

  1. Take the brief as a seed. A two-word theme is enough to start.
  2. Choose a scene archetype whose structural parts come from the theme itself (see archetype catalog below). The shape of the scene must read as that theme to a viewer who can't see any labels.
  3. Resolve the palette to four colour roles (BG / BODY / FOLIAGE / ACCENT). Pick from presets or accept custom hex.
  4. Set complexity (L1 / L2 / L3) — see complexity scale below.
  5. Compose the scene as one self-contained object centred on a flat saturated background. Vary silhouette with optional modules (cranes, smokestacks, ladders, airships, antennas) so the machine/scene never reads as a plain rectangular box.
  6. Render at 1024×1024 with a high-fidelity text-to-image model that adheres tightly to detailed prompts.

Locked style axes (NEVER vary)

Perspective

  • 3D isometric, axonometric — no vanishing-point perspective
  • Centred composition with generous monochromatic negative space on all sides

Linework

  • Ultra-fine, delicate, wispy hairline strokes — uniform weight, ~0.5pt ink-pen feel
  • NEVER bold, NEVER chunky, NEVER variable-weight
  • Refined architectural-drawing or fashion-illustration sensibility — the colour blocks do the heavy lifting, the lines are structural guides only
  • If the output reads as "graphic / poster / sticker", the linework is too heavy — regenerate thinner

Fills

  • Flat colour fills, no gradients, no airbrush shading, no inner highlights
  • Subtle drop-shadow on the ground beneath the scene is allowed; no other shading

Background

  • The ENTIRE canvas, edge-to-edge, is filled with one solid flat saturated colour (the BG role of the palette)
  • No white. No transparency. No vignette. No background scenery.
  • If the output puts the scene on white, the prompt failed — regenerate with stronger BG-fill instruction (capitalise CRITICAL, repeat the hex)

Aesthetic

  • Editorial magazine-print quality — Monocle / Kinfolk / The New Yorker / Bloomberg / Wes Anderson sensibility
  • Sophisticated, dusty, slightly muted but saturated
  • NEVER neon, NEVER fluorescent, NEVER blacklight, NEVER bubblegum, NEVER cartoon-saccharine
  • Words to avoid in the prompt: "neon", "fluorescent", "glow", "blacklight", "kawaii", "cute", "bold outlines", "thick strokes"

CRITICAL: the scene IS the metaphor

Each theme produces a scene whose structural parts come from the theme itself.

Anti-pattern: A generic rectangular box-machine with a file-drawer on one side, a megaphone funnel on the other, and folder tabs relabeled per theme (PLANS/LABOR/FUNDS, EMAIL/ADS/SOCIAL, DRAFT/EDIT/PUBLISH). This makes every output look like the same accounting machine with different stickers. Do not do this.

Correct approach: Pick props native to the theme so the silhouette itself reads as the subject.

Archetype catalog (extend liberally — these are seeds, not a closed list):

ThemeScene archetype
City constructionHalf-built tower + crane + cement mixer + scaffold + bricks + drafting board + hard-hat figure
MarketingBroadcast tower + signal waves + envelopes radiating + studio booth at the base
HiringConveyor belt of empty chairs + résumé sheets + "HIRED" stamp + sorting station
PublishingPrinting press with rollers + ink wells + paper feed + finished book stack
Product launchRocket on launchpad + steam jets + mission-control console + countdown timer
Sky gardenTerraced garden tower + hanging vines + gazebo + fountain + gardener with watering can
Floating island parkIsland chunk with grassy top + gazebo + benches + paths + lamp posts + dock + rowboat
Castle in the skyFloating rocky island with hanging roots + multi-tower castle with flying buttresses + banners + airship + clouds + crystal at the underside
Customer supportTicket-roll dispenser + headset + queue line of envelopes + bell on counter
Financial closeVault door + ledger book stand + abacus + calculator tape + coin stacks
OnboardingWelcome desk + kiosk + red carpet + swag boxes + door with tiny figure stepping through
Lighthouse at duskRocky cliff + tall lighthouse + lantern beam + waves at the base + a single boat

If the user names a theme not in the table, invent the archetype. The rule is always: the silhouette must read as the theme to a viewer who can't see any labels.

Customisable axes

Axis 1 — Palette (4 colour roles within one hue family)

RoleFunction
BGSaturated mid-tone, fills the whole canvas
BODYDeeper shade — main structural elements (stone, metal, wood)
FOLIAGE / secondaryMid-shade for plants or secondary structures (optional, used in plant-heavy scenes)
ACCENTWarm complement (cream / butter / sunflower / yellow) — small elements, fabrics, fixtures, lit windows
PAPERWhite — highlights, water, small details, cloud-puffs

Presets:

NameBGBODYFOLIAGEACCENT
green#5BD43A grass#1F5C2E forest#F4C430 sunflower
blue#3A7BD5 cobalt#1B3B7A navy#F4C430 sunflower
coral#F4A0B0 dusty pink#7A2E4E berry#C76B89 mauve#F5E4C8 cream
yellow#F2C94C mustard#8B5A2B cinnamon#9E8C2F olive#F8EDB5 cream
terracotta#D27D4A clay#5C2A14 dark brown#F5E4C8 cream
lavender#C7A7E0 lavender#4A2B5C aubergine#F4E4B8 butter
mint#A8E0C8 mint#1E5C50 teal#F4B58A peach
sunset#F2A07B peach#8C3A2A brick#F5E4C8 cream
sky#A8D0F2 sky-blue#2B4A7A steel-navy#F8EDB5 cream

The user may also supply a custom triplet: palette=#xxxxxx/#xxxxxx/#xxxxxx.

Axis 2 — Scene archetype

User-supplied theme (see catalog above). Resolve to a list of theme-native props, materials, and a small figure or two when the scene calls for it.

Axis 3 — Complexity

LevelElement countDescription
L1 Minimal4–6 visible elementsOne central object + 1–2 supporting props
L2 Standard (default)8–12 elementsCentral object + supporting props + small figure + atmospheric details (clouds, birds, small flora)
L3 Rich15+ elementsPopulated scene with multiple sub-vignettes, multiple figures, hovering airships or clouds, ambient props

Axis 4 — Optional modules (silhouette variety)

For machine-style themes especially, pick 2–4 extra modules so the outline is varied and asymmetric, never a plain rectangle:

  • Mechanical: gears, pipes, conveyor belt, crane-arm, lever, pressure-gauge
  • Atmospheric: smokestack with paper-puff smoke, steam-vent, antenna/dish, signal-light bulb
  • Structural: wheels, tripod legs, side ladder, scaffolding, periscope, fuel tank
  • Human-scale: tiny operator figure, gardener, hard-hat worker, traveller
  • Aerial: hovering airship, floating crystal, dangling root cluster, cloud-puffs

Reusable prompt template

Isometric editorial vector illustration of a stylized miniature {SCENE_NAME} — a small, self-contained scene as a single composed object.

BACKGROUND — CRITICAL: The ENTIRE canvas, edge to edge, is filled with one solid flat saturated {BG_NAME} ({BG_HEX}). No white. No transparent areas. The {BG_NAME} IS the canvas. The scene sits centred with generous {BG_NAME} negative space around it.

LINE STYLE: All outlines are ULTRA-FINE, delicate, wispy hairline strokes — 0.5pt ink pen weight, uniform line weight throughout, no variation, NEVER bold strokes. Refined architectural-drawing feel — colours do the heavy lifting, lines are structural guides only.

COLOUR & FILL: Flat colour fills, no gradients. Strictly {HUE_FAMILY}-family palette:
- Canvas background: saturated {BG_NAME} ({BG_HEX}) covering the WHOLE image
- Main structural body: {BODY_NAME} ({BODY_HEX})
- Foliage / secondary structures: {FOLIAGE_NAME} ({FOLIAGE_HEX}) [if applicable]
- Accent surfaces: {ACCENT_NAME} ({ACCENT_HEX})
- Highlights (water, clouds, small details): white
Editorial magazine-print quality, sophisticated and dusty — NOT neon, NOT saccharine, NOT cartoon. Monocle / Kinfolk / Wes Anderson.

COMPOSITION: 3D isometric perspective, centred, with the saturated {BG_NAME} filling every corner of the canvas.

The scene contains:
{SCENE_PROPS_LIST — 4–15 props with materials and colours, drawn from the theme itself, never relabelled-folder-tabs}

Soft drop-shadow on the {BG_NAME} ground beneath the scene. No background scenery beyond the flat {BG_NAME} fill. No text. No filing drawers, no megaphone funnels, no labelled folder tabs (unless the scene IS literally about filing or broadcasting). Editorial magazine illustration aesthetic — Monocle / Kinfolk / New Yorker spot-illustration sensibility, ultra-fine linework.

Generation notes

  • Model recommendation: a high-fidelity text-to-image model with strong long-prompt adherence (gpt-image-2 is the current best for this style — supersedes gpt-image-1.5 with stronger long-prompt adherence and cleaner flat-vector output). Image-to-image is not required — the locked style is fully describable in text.
  • Output: 1024×1024 PNG, high quality
  • Cost / determinism: since this is text-to-image, output varies slightly between runs even with the same prompt. For exact reproductions, pin a seed if the underlying model supports it.

Failure modes and how to recover

FailureSymptomFix
Background dropped to whiteScene floats on white instead of the BG colourRe-prompt with BACKGROUND — CRITICAL: … capitalised, repeat hex 2–3×, mention "fills edge to edge"
Linework too boldLooks like a poster / stickerRe-prompt emphasising "ultra-fine", "0.5pt ink pen", "colours do the heavy lifting"
Generic filing-cabinet shapeOutput looks like the accounting-machine reference with relabelled tabsExplicitly forbid "filing drawer", "folder tabs", "megaphone funnel"; describe the theme-native silhouette concretely
Saccharine / neon paletteOutput reads cartoonAdd "sophisticated, dusty, NOT neon, NOT saccharine, Wes Anderson palette"
Foliage reads as a different hue familyE.g. green leaves in a pink sceneSpecify the foliage hex explicitly within the monochromatic family

Reference outputs

The four images below are user-approved generations that anchor the style:

FilePaletteScene
references/sky-castle.pngskyFloating rocky island + multi-tower castle ruin + airship + clouds + crystal
references/coral-hanging-garden.pngcoralThree-tiered terraced garden tower + cascading vines + gazebo + fountain + gardener
references/yellow-floating-island-park.pngyellowFloating island park + gazebo + benches + lamp posts + dock + rowboat + visitors
references/blue-city-construction.pngblueHalf-built skyscraper + crane + cement mixer + scaffold + drafting board + hard-hat figure

Use these as the canonical look. New scenes should match their linework weight, background saturation, and "scene-as-metaphor" composition.

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 Iso Scene AI skill do?

Isometric editorial-magazine scene illustration in a locked flat-vector style — ultra-fine hairline outlines, monochromatic saturated background filling the canvas, and a single composed scene whose props themselves embody the theme. Trigger when users say /iso-scene, ask for an "isometric editorial illustration", a "scene illustration in the editorial machine style", or brief with palette + scene archetype + complexity.

Why use Iso Scene on TypingMind?

Because you install it once and use it with any model. Iso Scene 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 Iso Scene in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/vm0-ai/vm0-skills/tree/main/illustration-template/iso-scene. 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 Iso Scene?

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 Iso Scene?

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

Is the Iso Scene AI skill free?

It is published on GitHub by vm0-ai. Check the repository for licensing terms. 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 👇