@threeforged/asset-analyzer

Analyze 3D assets for performance issues. Scans GLB, GLTF, and OBJ files and reports potential problems like high polygon counts, duplicate materials, oversized textures, and more.

Installation

# 1. Install the ThreeForged CLI (one-time setup)
npm install -g @threeforged/cli

# 2. Install the asset analyzer plugin in your project
npm install --save-dev @threeforged/asset-analyzer

# 3. Analyze your assets
threeforged analyze ./models/

CLI Usage

# Analyze a single file
threeforged analyze model.glb

# Analyze a directory of assets
threeforged analyze ./models/

# Save report to file
threeforged analyze ./models/ -o report.txt

# JSON output
threeforged analyze ./models/ --json

Output

The analyzer produces a report with:

• Summary — total meshes, triangles, vertices, materials, textures, draw calls, animations, GPU memory
• Files table — per-file breakdown with format, size, mesh count, vertices, triangles, materials, draw calls, and animations
• Warnings — actionable issues found in your assets

Analysis Rules

| Rule | Severity | What it checks | |---|---|---| | high-poly | warn/error | Meshes exceeding polygon thresholds (50k warn, 100k error) | | duplicate-materials | warn | Materials with identical properties that could be merged | | large-textures | warn | Textures exceeding max dimensions (default: 4096x4096) | | texture-memory | error | Total GPU texture memory exceeding budget (default: 64 MB) | | unindexed-geometry | warn | Meshes without index buffers (higher memory usage) | | duplicate-meshes | info | Meshes sharing identical geometry across files (instancing candidates) |

Programmatic API

import { analyzeAssets } from '@threeforged/asset-analyzer';

const report = await analyzeAssets('./models/');

console.log(report.metrics.totalTriangles);
console.log(report.warnings);

Configuration

Thresholds can be customized via threeforged.config.js:

export default {
  polyCountThresholds: { medium: 30000, large: 80000 },
  maxTextureSize: 2048,
  maxTextureMB: 32,
};

License

MIT