Optimize Text Mesh Pro logo

Optimize Text Mesh Pro

OrganizationPopular
Unity-Technologies
optimize-text-mesh-pro

Optimizes TextMeshPro rendering, memory, and font setup, including font asset stacks, fallback atlases, SDF quality, and worldspace text. Use when the user asks about TMP fonts, CJK or multilingual text, or text performance.

Overview

PublisherUnity-Technologies
Repositoryskills
Skill nameoptimize-text-mesh-pro
Stars
1K
Forks
61
Bundled files
Instructions only
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.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by Unity-Technologies on GitHub. Read the source before you install it.

Installation

Install the Optimize Text Mesh Pro 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/Unity-Technologies/skills/tree/main/skills/optimize-text-mesh-pro
  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/Unity-Technologies/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/skills/optimize-text-mesh-pro .claude/skills/optimize-text-mesh-pro
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Optimize Text Mesh Pro 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 Optimize Text Mesh Pro 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 Optimize Text Mesh Pro 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.

Optimize TextMeshPro

Triage — identify the symptom first

Before providing tips, identify which category the user's issue falls into. If the user has not described a specific symptom, ask: "Are you seeing a memory/atlas bloat, visual quality, CPU/performance, build size, or localization/alignment issue with TextMeshPro?"

SymptomGo To
Memory Profiler shows large or multiple TMP atlasesFont Stack & Dynamic Fallbacks, Memory Profiler: Include Font Data
Inconsistent glyph weight, fuzzy edges, visual qualityPadding & Sampling Ratios, Font Asset Scale, Atlas Render Mode: SDF16
CPU spikes during text updates or Canvas rebuildsAutoSize, Worldspace vs Canvas Text
Build size too large from shipped font filesDynamic OS Atlas Population
Mixed Latin + CJK alignment looks offFont Normalization
Need multiple font styles (italic, outline, glow)Material Presets

Core Rules

  • Main font = static asset with all glyphs baked in. Add dynamic fallbacks via the Fallback list (or TMP Settings) for everything else. Keep dynamic atlas size at 512-1024 to bound peak memory.
  • Dynamic fallback fonts -> enable Clear Dynamic Data On Build. Otherwise editor-baked glyphs ship in the player.
  • Keep Padding-to-Sampling-Point-Size ratio consistent across primary + fallback fonts. Mismatch produces inconsistent glyph weight on the same line.
  • Latin sampling point size 70-90; CJK 36-50. Different scripts need different sampling sizes for clean SDF.
  • Font asset Scale = 1. Anything else (e.g., 0.9) breaks standard point-size math.
  • Disable AutoSize at runtime once layout is locked. AutoSize is for design, not for live counters.
  • Worldspace text -> use TextMeshPro, not TextMeshProUGUI. Canvas overhead in worldspace is not free.
  • Parent often-changing TMP UI to its own Canvas to bound rebuild cost.
  • TMP material presets > duplicating font assets for italic / bold / outline / glow variants of the same font.
  • For shipping multilingual builds on iOS/Android, evaluate Atlas Population Mode = Dynamic OS (TMP 3.2.0-pre.3+) to leverage system fonts and shrink the build.

Font Stack & Dynamic Fallbacks

If the user reports memory bloat from TMP atlases, advise this font stack pattern:

Main font asset (static, all required Latin glyphs baked)
  -> Fallback 1: Dynamic font (atlas 512 or 1024) for CJK
  -> Fallback 2: Dynamic font for symbols / emoji

NEVER ship a dynamic fallback font asset without enabling Clear Dynamic Data On Build. Every glyph baked while testing in the editor is included in the player build if this toggle is off.


Padding & Sampling Ratios

If the user reports inconsistent stroke widths or glyph weight differences between primary and fallback fonts, check the padding-to-sampling-point-size ratio.

The ratio is Padding / SamplingPointSize. With Padding = 9 and Sampling Point Size = 90, ratio = 10%.

  • A primary font with one ratio and a fallback with a different ratio produces inconsistent stroke widths on the same line.
  • Pick a ratio (10% is a safe default), apply it to all font assets in the chain.

Recommended sampling point sizes:

  • Latin scripts: 70-90.
  • CJK scripts: 36-50 (CJK glyphs are visually denser; smaller sampling sizes still produce clean SDF and save atlas memory).

Font Asset Scale

If the user reports point sizes not matching design specs, check the font asset Scale value. Some imported TMP font assets ship with Scale = 0.9 instead of 1.0. The Scale value participates in the point-size-to-pixels math, so a non-1 scale produces non-standard point sizes. Advise the user to set Scale = 1 on all font assets before adjusting padding ratios.


Sprite Assets

If the user reports slow loading times for TMP Sprite Assets on mobile, check the source texture's Texture Type. It must be set to Default (not Sprite). Sprite type creates child sub-objects that TMP doesn't use; Default avoids them.


AutoSize

If the user reports CPU spikes on text fields that change frequently (timers, counters, chat, dynamic player names), check whether enableAutoSizing is on. AutoSize resizes the text whenever the string changes, causing constant CPU spikes.

Advise: disable AutoSize and hard-code the chosen point size once layout is locked. Keep AutoSize on only for genuinely static labels that auto-fit on locale change.


Atlas Render Mode: SDF16

If a static font with point size 72 or larger looks unclear or has fuzzy edges, advise switching the Atlas Render Mode to SDF16. Higher precision SDF for big glyphs, at slightly more atlas memory.


Font Normalization

If the user reports misaligned Latin + CJK text on the same line, walk them through this procedure:

  1. Window -> TextMeshPro -> Settings -> Import TMP Example & Extras (one-time per project).
  2. Add the TMP_TextInfoDebugTool component to the TextMeshPro object displaying misaligned text.
  3. Enable ShowLines toggle - the ascender, descender, and baseline render as overlays.
  4. Mix Latin + CJK strings; if the lines diverge, adjust ascender/descender on the TMP Font Asset until they align.

Caveat: importing TMP Examples & Extras has been observed to cause an infinite import loop on some project layouts. If it happens, close Unity and re-open - the import resolves on the second attempt.


Material Presets

If the user needs multiple styles (italic, bold, outline, glow) of the same font, advise material presets instead of duplicating font assets. Presets share the same font texture but override shader parameters.

How to create:

  1. Select a TMP Text GameObject.
  2. In Inspector, find the Material section.
  3. Right-click the Material header -> Create Material Preset.
  4. Rename the new material and tweak settings.
  5. On the TMP Text component, pick the preset from the Material Preset dropdown.

Dynamic OS Atlas Population (TMP 3.2.0-pre.3)

If the user is shipping multilingual builds and concerned about build size, advise evaluating Atlas Population Mode = Dynamic OS (TMP 3.2.0-pre.3+):

  • In Editor: still uses the source font from the project.
  • In a player build: the source font is not included. At runtime, Unity searches the device for a font with the matching Family + Style name.

Recommended system fonts for CJK:

PlatformRecommended system font
AndroidNotoSans (covers Chinese, Japanese, Korean glyphs broadly).
iOSPingFang for Simplified/Traditional Chinese. iOS uses unique fonts per language for CJK (different families for Chinese, Japanese, Korean) - check the fallback chain when shipping a single TMP setup across all three.

Wins: build size shrinks (no shipped CJK font files) and memory drops (system font is shared with the OS).


Memory Profiler: Include Font Data

If Memory Profiler shows unexpectedly large font asset sizes in the Editor, check whether Include Font Data is enabled on the .ttf / .ttc import settings. The Editor includes the source font file in the asset by default, but on device (especially with Dynamic OS), this cost is not paid.

To make Editor captures match device: on the font file -> deselect Include Font Data in the import settings. Memory Profiler will then show overhead without the underlying font file.


Worldspace vs Canvas Text

If the user has worldspace text (damage numbers, signs, holograms) using TextMeshProUGUI, advise switching to TextMeshPro. Worldspace Canvas is a known inefficiency.

If a TextMeshProUGUI element's text changes often (timers, counters, chat), advise parenting it under a child GameObject with its own Canvas component. Canvas rebuilds are scoped per-Canvas, so isolating the volatile field cuts rebuild cost on the rest of the UI.


Common Pitfalls

If the user's setup matches any of these, flag it:

  • One giant dynamic font asset for all languages instead of static main + dynamic fallback - the dynamic atlas balloons.
  • Inconsistent padding ratio across primary + fallback - same line of text looks like two fonts.
  • Font asset Scale = 0.9 inherited from import - point sizes won't match design specs.
  • Leaving AutoSize on for live counters - hidden CPU spikes.
  • World-space TextMeshProUGUI inside a worldspace Canvas - extra rebuilds for no benefit; use TextMeshPro.
  • Forgetting Clear Dynamic Data On Build on dynamic fallback fonts - editor-test glyphs ship in the player.
  • Capturing Memory Profiler in Editor with Include Font Data on, then being surprised the on-device build is smaller.
  • Sprite asset source texture set to Sprite type - mobile loading slows from extra child sub-objects.

References

Frequently asked questions

What does the Optimize Text Mesh Pro AI skill do?

Optimizes TextMeshPro rendering, memory, and font setup, including font asset stacks, fallback atlases, SDF quality, and worldspace text. Use when the user asks about TMP fonts, CJK or multilingual text, or text performance.

Why use Optimize Text Mesh Pro on TypingMind?

Because you install it once and use it with any model. Optimize Text Mesh Pro 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 Optimize Text Mesh Pro in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Unity-Technologies/skills/tree/main/skills/optimize-text-mesh-pro. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Optimize Text Mesh Pro?

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 Optimize Text Mesh Pro?

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

Is the Optimize Text Mesh Pro AI skill free?

It is published on GitHub by Unity-Technologies. 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 👇