TurboLERC is one of eight drop-in codecs in the Bitruvius Imagery & 3D Suite: keep your wire format, swap the backend.
Explore the SuiteSame LERC files. 2.8× faster to write. 1.7× to read.
LERC is how the geospatial industry stores elevation, radar and scientific imagery. TurboLERC is a faster library for the same format: swap it in where you use Esri's, and every file you write still opens in ArcGIS, GDAL and every other LERC reader.
Faster on every file we measured, on Intel and Apple silicon. Native on servers, and in the browser too.
Like-for-like medians against Esri's LERC library on the same machine, lossless. Captures dated 2026-06-06 (Intel) and 2026-04-27 (Mac).
The 60-second version
The format stays. The bill for it shrinks.
LERC
Esri's open compression format for rasters that hold measurements, not just pictures: heights, radar returns, spectral bands. It sits inside ArcGIS, Cloud-Optimized GeoTIFFs and GDAL.
Esri's library
The open-source reference that every LERC pipeline links today. It defines the format, so it is the benchmark any replacement has to beat, on the same files.
Encoding is compute
Every new scene, every reprocessed archive, every format migration is written through the encoder. Faster writes mean smaller batch clusters and fresher catalogues.
Decoding is servers
A tile server decodes on every map request, and analytics read every pixel. Faster reads mean fewer servers for the same traffic, and quicker maps in the browser.
Watch it
Race it yourself, in this tab.
TurboLERC and Esri's LERC library decode the same bytes in your browser, our WebAssembly against theirs, clocked live on your machine. Pick a raster type and press start.
No cherry-picking
784 head-to-heads. Esri's library won none.
392 production files across 12 kinds of data, from lidar elevation to complex radar, each written and read by both libraries on the same machine. Every data type is below, on either machine, in either direction. The slowest one is still faster.
Solid bar = median across the matched corpus.Peak column shows max across samples.
Slowest data type on this view: 1.40×. Peak is the single fastest file of that type. Hyperspectral samples were captured on the Mac only.
What it means for the bill
Same data, same format, less hardware.
Codec time is machine time. Here is the median result, translated.
Illustrative, from the measured Intel medians, where reading and writing LERC is the main cost. A pilot measures your number.
By data type
What it means for the files you have.
Elevation (DEM)
“Rebuild an elevation archive on a fraction of the compute, and serve every terrain tile faster.”
Hyperspectral
“Spectral science data, where reads gain far more than writes.”
Complex SAR
“On complex radar the reads gain most, and reads are where an archive spends its life.”
SAR video
“The read-heavy workload, and the biggest read speed-up we measured on either machine.”
SAR quicklook
“Previews generated and served faster, on the same files.”
Aerial 4-band (NAIP)
“Four-band aerial imagery, written and read faster on both machines.”
Chips: median speed against Esri's library for that data type, on each machine.
Already at work
The LERC inside Spatial Engine.
Spatial Engine, our imagery engine, converts and reprojects imagery, and TurboLERC is the LERC inside it. Writing the same LERC v4 files GDAL writes, it finishes several times sooner in a fraction of the memory. Its default is LERC v6, which GDAL reads but cannot write, so the same elevation model with zero error takes 29 MB instead of 71.
See Spatial EngineSpeed and memory: Spatial Engine against GDAL 3.13.3, LERC v4 on both sides, median over the whole test set on the same machine; Windows leads. Size: USGS 3DEP 1 m DEM, lossless.
Drops into the pipeline
Every LERC version. Read and write.
Archives hold LERC from every era. TurboLERC reads all of it through one call and writes any version you ask for, including v1: some ArcGIS web elevation layers still expect it, and Esri's own library stopped writing it at version 3.
Standard files
Everything TurboLERC writes is standard LERC. ArcGIS, GDAL, Cloud-Optimized GeoTIFF readers and your own LERC tooling open it unchanged. Readers don't know the difference.
Every data shape
8, 16 and 32-bit integers, 32 and 64-bit floats. One band or many, several values per pixel, with or without a validity mask. Every surface of the LERC API is covered.
Why it's faster
The format dates from 2010 and the reference reflects that era. TurboLERC keeps the format and rebuilds the codec for modern processors. The techniques are patent-pending and ship only as compiled binaries.
Platform matrix
All backends produce bit-exact identical output. Auto-detection at runtime; portable scalar fallback for any CPU.
How this was measured
- Machines: Intel Core Ultra 9 285HX (Windows x64, AVX2) and Apple M3 Max (NEON), each compared with Esri's LERC library on the same machine. The two are never divided into one ratio.
- The newest capture on each machine: 2026-06-06 on Intel (182 files, 11 data types) and 2026-04-27 on the Mac (210 files, 12 data types; the Mac set adds hyperspectral).
- Lossless LERC, 256-pixel tiles, the same file through both libraries. Each file's figure is the median of its repeated runs; speed is TurboLERC's throughput divided by Esri's. Licensed third-party samples are excluded.
- Sizes: on Intel 142 files came out the same size, 39 smaller and 1 larger; on the Mac 130, 64 and 16. In total, 0.04% smaller than Esri's on Intel and 0.04% on the Mac.
- The live race decodes in your browser against Esri's published WebAssembly build, so its numbers are yours, not ours.
Bring five of your LERC files.
A pilot runs both libraries on your own data, on your own machine: same files in, same format out. You keep the numbers.
TurboLERC is licensed through the Bitruvius Developer Hub and ships as compiled binaries.
Esri, ArcGIS and LERC are trademarks of Environmental Systems Research Institute, Inc., used here solely to identify the products and formats being compared. No affiliation, sponsorship, or endorsement is implied. LERC is published by Esri under the Apache 2.0 license; the live demo runs Esri's unmodified reference decoder under that license. TurboLERC is an independent implementation of the openly published LERC format; compatibility claims describe interoperability with that format as verified by our test suites. GDAL is an open-source project, named solely to identify the software compared; the elevation sample is USGS 3D Elevation Program data. Apple, Intel, Intel Core and Windows are trademarks of their respective owners and identify the test machines. Performance comparisons reflect our own measurements under the stated methodology; results vary by workload and hardware. Full third-party notices: Attributions.