mcpField
C
Coating Lab

Coating Lab

Local
github.com/hpdkhoaMedia & Design1Updated 2w ago

About Coating Lab

Coating Lab is an MCP server for game-weapon artists who need sourced material finishes and wear baked into Blender models. It supports finish variants, validation, renders, and engine-ready material values using a shared wear formula.

Key features

  1. 01Sourced physically based finish library
  2. 02Per-slot finish variants and wear levels
  3. 03Bake evidence-based wear masks in Blender
  4. 04Validate finishes and coating specs
  5. 05Render variants and export engine values

Use cases

  1. 01Apply varied finishes to weapon model slots
  2. 02Bake wear patterns into vertex colours
  3. 03Check coating variants against wear rules

MCP support

Exposes finish lookup, coating-spec validation, wear baking, rendering, and engine-parameter tools to MCP clients.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"hpdkhoa-coatinglab": {
"command": "uvx",
"args": [
"coatinglab-mcp"
],
"env": {
"COATINGLAB_OUT": "<COATINGLAB_OUT>",
"COATINGLAB_BLENDER": "<COATINGLAB_BLENDER>",
"COATINGLAB_LIBRARY": "<COATINGLAB_LIBRARY>",
"COATINGLAB_WAIT_FOR": "<COATINGLAB_WAIT_FOR>"
}
}
}
}
Run commanduvx coatinglab-mcp
Environment variables to set
COATINGLAB_OUTThe only folder Coating Lab writes to; defaults to ~/.coatinglab
COATINGLAB_BLENDERBlender's executable, if it is not on the PATH or in the usual install folder
COATINGLAB_LIBRARYA finishes JSON of your own, merged over the built-in library
COATINGLAB_WAIT_FORProcess names that hold the machine (e.g. UnrealEditor.exe); a bake refuses to start while one runs

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
pypicoatinglab-mcp0.1.0

Tools

13

Saved tool list · updated 2w ago

This tool takes no parameters.

Details

From the project README

Coating Lab

Real surface finishes for game gun models, worn where use wears a gun.

license: MIT Python 3.10+ PyPI: coatinglab-mcp MCP Registry: coatinglab-mcp Blender: 4.2+ tools: 13 finishes: 21 sourced

An MCP server that gives game gun models real surface finishes, worn where use wears a gun.

  • A finish library with sources. Anodized aluminium, manganese and zinc phosphate, blued and nitrided steel, ceramic coatings (FDE, OD green, tungsten, graphite, coyote), stainless, polymer, rubber, brass, copper, lead, walnut. Each is a physically based value (linear colour, metallic, roughness) with what shows when it wears through and the sources of its numbers. Values reasoned from descriptions are marked as estimates.
  • Coating variants. A spec puts a finish on each material slot of a model, in as many variants as you want (an issued rifle, a two-tone FDE build, an abused one), each at a wear level: new, issued, worn or abused.
  • Wear from evidence, baked once. In Blender, into the model's vertex colours: exposed edges, where the hands work the gun (its grip, the support hand, the controls), where parts slide (a magazine in its well, a stock on its tube), and dirt where the surface is occluded and round the muzzle and the port. Every variant reads the same masks, so a new variant costs nothing.
  • Checks. Every finish against the rules physically based rendering rests on. Every variant's worn share against its level, and no wear inside corners.
  • One formula everywhere. The checks, the look-dev renders and an engine's material compute wear the same way (coat.py), so what passes is what the player sees.

Six AR-15 coatings on one Model Lab build

Model Lab (modellab-mcp) builds give their own slots (upper, lower, barrel and more), marks and parts: spec_for_modellab makes a coating spec from a build in one call.

Install

Python 3.10 or later and Blender 4.2 or later.

pip install coatinglab-mcp
{"mcpServers": {"coatinglab": {"command": "coatinglab-mcp"}}}
SettingMeaning
COATINGLAB_OUTThe only folder Coating Lab writes to. ~/.coatinglab by default.
COATINGLAB_BLENDERBlender, when it is not on the PATH or in the usual install folder.
COATINGLAB_LIBRARYA finishes JSON of your own, merged over the built-in library.
COATINGLAB_WAIT_FORProcess names that hold the machine. A bake waits for none of them to run.

Tools

ToolWhat it does
list_finishes, read_finishThe library, and one finish with its notes and sources.
check_library, check_finishThe library, or a finish of your own, against the physical rules.
list_wear_levels, list_templatesThe wear levels, and the ready variant sets.
spec_for_modellabA coating spec for a Model Lab build.
validate_specA spec's problems: a slot without a finish, a finish on the wrong material.
bakeBakes the wear masks into a copy of the model.
check_coatingEach variant's wear against its level, and wear kept out of corners.
render_variantsLook-dev renders and a sheet.
engine_paramsA variant's values by slot for an engine material.

The masks and the formula

The bake writes the vertex colour Col:

ChannelMeaning
REdge: 1 on flat ground, 0 on a sharp exposed edge
GHandling: 1 where the hands work the gun
BRub: 1 along the lines parts slide on
ADirt: 1 where the surface is occluded, and round the muzzle and the port

An engine's material follows the formula in src/coatinglab/coat.py. A wear level is a threshold: the higher the level, the weaker the mask that already wears through. A grunge texture breaks wear up but never makes wear.

What is estimated

  • Every roughness value. No source maps a finish's surface to a rendering roughness.
  • The colours of the anodized, phosphate, blued, nitrided and most ceramic finishes. The standards set processes, not colours. OD green and coyote come from FED-STD-595 colour numbers.
  • The wear amounts. Sources say where guns wear, not how fast. Each level is set so Model Lab's M4A1 lands in the middle of its band.
  • A hand's zone is a sphere: on a round handguard the support hand's wear reaches its top as well as its underside.

License

MIT

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.