Large GLB and GLTF files load slowly in web viewers, AR applications, and mobile experiences - and large file size directly impacts page performance, bandwidth costs, and user experience. Draco compression is the standard solution: a Google-developed open-source compression algorithm specifically designed for 3D mesh geometry that reduces GLB file sizes by 70–80% in typical cases with no visible quality loss in rendered output.
Autoconverter applies Draco compression when saving GLTF and GLB files, with three compression modes to balance speed against file size reduction. This guide explains how Draco works, when to use it, and which compression mode to choose.
What Is Draco Compression?
Draco is an open-source library developed by Google for compressing and decompressing 3D geometric meshes and point clouds. It is included in the GLTF 2.0 specification as the KHR_draco_mesh_compression extension - an official, standardized way to embed compressed geometry inside GLTF and GLB files.
Draco compresses the geometry data buffers inside a GLB file - vertex positions, normals, UVs, and face indices - using a combination of quantization and entropy coding. The result is a GLB file that contains the same mesh, materials, and textures, but with dramatically smaller geometry data:
- Typical geometry compression: 70–80% smaller than uncompressed GLTF/GLB geometry buffers
- Overall file size reduction varies by model - geometry-heavy models see the largest gains; texture-heavy models see less reduction since textures are not affected by Draco
- No visual quality loss in rendered output - Draco is a lossy algorithm for vertex coordinates, but the precision loss is below the threshold of visual detection at standard quantization levels
GLB vs GLTF: Which to Compress?
GLTF is a JSON file with external binary buffer (.bin) and texture files. GLB is a single binary file containing the JSON, geometry buffers, and textures bundled together. For web deployment and AR/VR, GLB is preferred because it is self-contained - one file to host and load. Draco compression works the same way in both formats, but GLB is the more practical choice for delivery.
When Autoconverter saves a Draco-compressed GLB, all geometry buffers are compressed and the KHR_draco_mesh_compression extension is declared in the GLTF header. Any viewer or runtime that supports this extension decompresses the geometry automatically at load time.
Draco Compression Modes in Autoconverter
Autoconverter offers three Draco compression modes when saving GLTF or GLB files:
| Compression Mode | Speed | File Size Reduction | Best Use Case |
|---|---|---|---|
| Fast Compression | ⚡ High - seconds per file | ⬇️ Moderate (40–60%) | Quick iteration, test builds, internal previews where speed matters more than minimum file size |
| Maximum Compression | ⏳ Lower - minutes for complex models | ⏬ Significant (70–80%) | Production web deployment, e-commerce 3D viewers, mobile AR apps where download speed is the priority |
| Custom Settings | ⚖️ Balanced | ✨ Adjustable | Fine-tuning quantization levels for specific geometry precision requirements or project constraints |
For final production assets destined for web or mobile, use Maximum Compression. For iterative workflow where you need fast turnaround to verify that models convert and display correctly, use Fast Compression.
Why GLB File Size Matters
GLB file size directly affects several practical outcomes across different delivery contexts:
- Web 3D viewers and product configurators - large GLB files cause slow initial load in WebGL viewers (Three.js, Babylon.js, model-viewer). Google recommends keeping individual GLB assets under 5MB for web use; Draco compression routinely brings 20–40MB models within that range.
- AR on mobile (WebXR, Apple AR Quick Look, Google Scene Viewer) - mobile networks and processors make compressed assets essential for acceptable AR launch times
- E-commerce 3D product visualization - page load performance directly impacts conversion rates; Draco-compressed GLB reduces the 3D asset's contribution to total page weight
- CDN bandwidth costs - compressed assets cost less to serve at scale when hosting thousands of 3D product models
- Game engines and real-time applications - faster asset streaming for runtime-loaded GLB files in Unity, Unreal Engine, or Godot
How to Compress a GLB File with Autoconverter
- Download and install Autoconverter on Windows. Launch it from the Start menu.
- Click Open... and load your source file - GLB, GLTF, OBJ, FBX, SKP, or any supported format.
- Click Save As... and select GLB (*.glb) or GLTF (*.gltf) as the output format.
- In the save parameters dialog, select the desired Draco compression mode (Fast, Maximum, or Custom).
- Click Save. Autoconverter writes the Draco-compressed GLB file.
For the full GLTF/GLB save workflow including Draco settings, see the GLTF Draco compression tutorial.
Frequently Asked Questions
Does Draco compression affect visual quality?
At default quantization levels, Draco compression is visually lossless - the vertex position precision loss is below what is visible in a rendered 3D viewer. At aggressive custom settings with very low quantization bits, some geometry distortion becomes visible. The default Maximum Compression mode uses quantization levels that do not produce visible quality degradation.
Do textures get compressed by Draco?
No. Draco compresses geometry data only - vertex positions, normals, UV coordinates, and face indices. Textures are stored separately in the GLB file and are not affected by Draco. To reduce texture file size, use compressed texture formats (KTX2/Basis Universal) in addition to Draco, or reduce texture resolution before export.
Can I compress existing GLB files without re-converting from a source format?
Yes. Open the existing GLB file in Autoconverter, then Save As GLB with Draco compression enabled. The geometry is re-exported with Draco compression applied.
Is there a free trial?
Yes. The free evaluation version of Autoconverter supports up to 10 file conversions, including GLB export with Draco compression. The full licensed version provides unlimited conversions and batch processing.
Summary
Draco compression reduces GLB geometry buffer size by 70–80% with no visible quality loss, making it the standard optimization step before deploying GLTF/GLB assets to web viewers, AR applications, e-commerce platforms, and mobile experiences. Autoconverter applies Draco via three modes (Fast, Maximum, Custom) when saving to GLB or GLTF, and can compress existing GLB files directly by reopening and re-saving them.
👉 Ready to compress? Download Autoconverter and try it free for up to 10 conversions.