Draco Compression Online — Compress Any 3D Format to GLB
Compress any 3D model format (OBJ, STL, PLY, GLB, 3DS, 3MF, X) to Draco-compressed GLB format.
Last updated Mar 2026
Usually under 3 seconds — depends on file size.
When to Use Draco Compression
WebXR Mobile Optimization
WebXR experiences on Quest, iPhone, and Android require fast asset loading. Draco-compressed GLB loads 5-10x faster than uncompressed on 4G/5G networks.
ARKit/ARCore File Size Limits
ARQuickLook (iOS) has a ~10MB limit. ARCore prefers <5MB. Draco compression reduces large models (20MB → 2MB) to fit platform constraints.
E-commerce 3D Product Viewers
Shopify 3D and WooCommerce AR viewers benefit from fast-loading Draco GLB. 1-second load time increases conversion rates by 7% (Google study).
Three.js Bundle Size Optimization
Embedding GLB files in your Three.js app? Draco compression reduces bundle size by 90%, improving initial page load.
Multi-Format Workflow
You have OBJ/STL/PLY models but need Draco GLB for WebXR? This tool handles conversion + compression in one step (vs. using separate converter + compressor).
How Draco Compression Works
Draco is an open-source compression library developed by Google for 3D meshes. It is designed for web transmission, compressing geometry data by 90-95% while preserving visual quality.
For GLB input, the tool parses the binary container to extract meshes, applies Draco encoding to compress vertex positions, normals, and UVs, then repacks the Draco-compressed meshes back into GLB with the KHR_draco_mesh_compression extension.
For non-GLB input (OBJ, STL, PLY, etc.), the tool first converts to geometry data by parsing the format and extracting vertex, normal, and UV data. It then applies Draco encoding to compress the geometry, and packages the result as a new GLB file with Draco-compressed meshes and a basic PBR material.
The compression combines several techniques for maximum reduction. Mesh quantization reduces vertex precision from 32-bit to 12-16 bit for 40-50% size reduction. Vertex deduplication merges identical vertices for 10-20% reduction. Entropy encoding applies advanced compression to the quantized data for 70-80% reduction. The total result is 90-95% file size reduction for geometry-heavy models.
You can tune four compression parameters: compression level (0-10, where 0 is fastest/largest and 10 is slowest/smallest, default 7), position quantization (8-16 bits, default 14), normal quantization (6-12 bits, default 10), and UV quantization (8-14 bits, default 12).
Typical files (1-10MB) compress in 5-15 seconds.