Simplify an STL Mesh
Reduce STL triangle count locally with an estimated error budget and locked borders. Compare the achieved reduction, original preview and mesh health.
Supported input and output
ASCII and binary STL → binary STL. Keep 75%, 50% or 25% of the original triangle count.
A target is a request, not a promised triangle count
The simplifier collapses mesh edges while honoring locked open borders and a 1% estimated relative error budget. A small or tightly constrained mesh may retain more triangles than requested. Compare the achieved count, bounds and health report, and toggle the original preview before deciding whether the reduction is useful.
How to use this tool
- Load the source STL and review its triangle count and topology.
- Choose to keep 75%, 50% or 25% of the triangles, then run simplification.
- Compare original and result previews and download only after reviewing the achieved reduction.
Limitations to check
- meshoptimizer 1.3.0 uses edge-collapse simplification with a 1% relative error budget and locked open borders. Topology or the error budget may stop reduction above your target.
- The estimate is not an exact Hausdorff bound. Small features can change, and watertightness is not guaranteed. Inspect the before/after mesh and topology report.
- Non-manifold edges, duplicate triangles, inconsistent winding and zero-area triangles require source repair first.
File and processing limits
- STL input up to 8 MiB; 150,000 stored vertices and 50,000 triangles.
- Topology analysis: at most 150,000 edge records and 5 million weld comparisons.
- 20 seconds per job; generated STL output up to 3 MiB.
Processing happens in this browser; your file and filename are not uploaded. Cancel, reset or replace the file to stop its worker. Downloads are generated locally.
Questions about this workflow
Is the error estimate a dimensional guarantee?
No. It is an algorithm estimate relative to mesh extent, not an exact maximum surface-distance or manufacturing tolerance. Small features may change even when the estimate is low.