AI generates 3D models.
GLBForge ships them.

Raw AI meshes are 90MB, two million triangles, and wrong for the web. GLBForge analyzes them against real performance budgets, optimizes them 90%+ smaller, forges logos into 3D for free, and hands you a viewer — one command, on your machine.

$ npx glbforge ui

Node 20+ · MIT · runs locally, nothing uploaded

a PNG logo, forged to layered 3D in one second — glbforge extrude --layers 4 --pillow · drag to orbit

−94%

89MB Meshy export → 5.5MB, in 7s

4×

less GPU memory with KTX2 textures

credits to forge a logo into 3D

100%

local — your assets stay on your machine

The pipeline, not the hype

Generation is a commodity. The gap between "generated" and "shipped" is where projects die — GLBForge closes it with deterministic tooling: same input, same output, every time.

▦Analyze

Lighthouse-style report card: triangle/file/texture/GPU-memory budgets, topology defects, named lint rules with fixes. Exits non-zero over budget — it drops into CI like a linter.

⚒Optimize

Weld → simplify to budget → smooth normals → WebP or KTX2 textures → meshopt compression → LOD chains. Typically 90%+ smaller with no visible loss.

◆Forge

Logos, wordmarks, SVGs → watertight 3D. Layered colors, puffy-sticker relief, enamel/chrome/neon/acrylic presets. Instant, free, exact silhouette.

⎔Generate

Meshy image/text-to-3D built in — every generation lands already analyzed and one click from web-ready. Bring your own key.

⬇Export

Web-ready GLB, print-ready STL (mm-scaled, watertightness verdict), React Three Fiber viewer scaffold with LOD switching.

✓Budgets as contracts

mobile-hero, desktop-hero, product-configurator — opinionated caps that describe what ships comfortably, not what merely loads.

Forge: 2D artwork → real 3D, free

Image-to-3D models produce mush from flat graphics. GLBForge traces the silhouette exactly, stacks the colors in layered depth, pillows the surface, and hands you a watertight mesh you can ship — or print.

neon preset — layers glow in their own colors

Meshy generation: 73MB raw → this 3.2MB web asset, one flag

GLBForge Studio

Try it right now in your browser — drag in a GLB or a logo, read the report card, optimize with one click, compare original-vs-optimized with a split slider. Everything runs client-side: your assets never leave your device. The same studio runs locally with Meshy generation and KTX2 via npx glbforge ui.

# the whole studio, no install:
npx glbforge ui

# or the CLI, piece by piece:
npx glbforge analyze model.glb --profile mobile-hero
npx glbforge optimize model.glb --ktx2 --lods 40000,10000
npx glbforge extrude logo.png --layers 4 --pillow 0.03 --preset enamel
npx glbforge stl model.glb --size 80
npx glbforge scaffold model.web.glb -o viewer

Built for agents and CI, not just humans

The same pipeline is an MCP server — Claude Code, Cursor, and friends can take an asset from generation to web-ready inside one conversation — and a GitHub Action that posts report cards on pull requests.

# MCP server (Claude Code):
claude mcp add glbforge -- npx -y @glbforge/mcp

# then just ask:
"make logo.png into a web-ready
 3D asset under the mobile budget"

# .github/workflows/assets.yml
- uses: actions/checkout@v4
  with: { fetch-depth: 0, lfs: true }
- uses: glbforge/glbforge@main
  with:
    profile: mobile-hero
# → report card comment on every PR

AI generates 3D models.GLBForge ships them.