RUST · 3D ASSET PIPELINE

generate-glb-rs — Zero-Dependency Binary glTF 2.0 (.glb) Compiler in Pure Rust

generate-glb-rs is a pure Rust library and CLI that compiles procedural 3D meshes, vertex colors, normals, and PBR materials directly into spec-compliant Khronos glTF 2.0 binary (.glb) files. It eliminates heavy Python, Trimesh, and Blender dependencies with deterministic 4-byte chunk alignment and sub-millisecond serialization.

$ cargo install generate-glb-rs

GitHub ↗

Capabilities

Architected for Speed & Reliability

Engineered from the ground up in memory-safe Rust with zero unnecessary dependencies.

🚀

Zero Heavy Python Stack

Eliminates multi-gigabyte virtual environments, CUDA wheel conflicts, and NumPy / Trimesh installation nightmares with a single compiled Rust binary.

🌐

Multi-LLM Backends

Seamlessly generates meshes through local Ollama instances streaming LLaMA-Mesh GGUF, OpenAI-compatible APIs (vLLM / LM Studio), or offline procedural generators.

📺

Live Terminal Mesh Streaming

Streams vertices (v) and faces (f) live to your terminal as they are inferred, with real-time progress counters and vertex velocity.

🎨

Y-Axis Gradient Shading

Calculates normalized height gradient vertex colors automatically, rendering gorgeous multi-colored preview models in engines without requiring texture UVs.

📦

Direct Binary GLB Export

Encodes spec-compliant binary glTF 2.0 files with compact binary buffers, indices, and material headers ready to drag-and-drop into Godot, Blender, or Three.js.

🌐

Embedded 3D Web Viewer

Launch an embedded local HTTP server with interactive Three.js 3D viewport and Ollama-compatible `/api/generate` streaming endpoint.

Quickstart

Get Started in Seconds

Install via Cargo and run immediately on macOS, Linux, or Windows.

Terminal

# Install generate-glb-rs
cargo install --git https://github.com/bhubbard/generate-glb-rs

# Generate a 3D model using local Ollama LLaMA-Mesh
generate-glb -p "a low-poly medieval chest" -o chest.glb

# Generate using an OpenAI-compatible API
generate-glb --backend openai --api-base http://localhost:8000/v1 -p "racing helmet" -o helmet.glb

# Start the interactive 3D Web Viewer and API server
generate-glb serve --port 8080

Systems Design

Engine Architecture

Streamlined execution pipeline with deterministic memory isolation and zero runtime overhead.

Pipeline Flow

┌───────────────────────────────────────────────────────────┐
│                 Text Prompt: "a low-poly chest"           │
└─────────────────────────────┬─────────────────────────────┘
                              │ HTTP Stream
┌─────────────────────────────▼─────────────────────────────┐
│          Backend: Ollama / vLLM / LM Studio / Mock         │
└─────────────────────────────┬─────────────────────────────┘
                              │ Streaming OBJ Tokens: v & f
┌─────────────────────────────▼─────────────────────────────┐
│                 generate-glb-rs Pipeline                  │
│  ┌──────────────────┐  ┌──────────────────┐  ┌──────────┐ │
│  │ Stream OBJ Parser│  │ Y-Gradient Shader│  │ GLB 2.0  │ │
│  │ (Zero Alloc Buf) │  │ (Norm Heights)   │  │ Exporter │ │
│  └──────────────────┘  └──────────────────┘  └──────────┘ │
└─────────────────────────────┬─────────────────────────────┘
                              │ Spec-Compliant Binary
┌─────────────────────────────▼─────────────────────────────┐
│             Output: chest.glb (Blender / Godot / Web)     │
└───────────────────────────────────────────────────────────┘