If you use this software for academic research, please quote:
Deroulers et al., Analyzing huge pathology images with open source software, Diagnostic Pathology 8:92 (2013).
And/or if you find this software useful, please send us an email! This will help us to get support from our funding agencies to keep working on it. You can use following address (remove the anti-spam):

Pre-compiled binaries

64-bit Windows: tiffmakemosaic.exe tiffmakemosaic-j.exe tifffastcrop.exe tiffsplittiles.exe
32-bit Windows: tiffmakemosaic.exe tiffmakemosaic-j.exe tifffastcrop.exe tiffsplittiles.exe

Source code:



Version française


LargeTIFFTools is a collection of software that can help managing (very) large TIFF files, especially files that are too large to fit entirely into your computer's memory. It is composed of the following programs:

Getting the software

The software is open source, distributed under the GNU General Public License v. 3.0. It uses noticeably the libtiff and libjpeg or libjpeg-turbo software, made free and open by its authors, which we acknowledge.

You can get the software through download (see the links in the frame). We provide some pre-compiled versions. The 32-bit Windows version should run under all versions of Windows (XP, Vista, 7; either 32- or 64-bits) but will produce images which require at most ca. 1 GiB memory. The 64-bit version goes beyond that limit.

If you install the software from source, you need the following installed libraries: libtiff (we recommend a version >= 4.0.0) and libjpeg (and possibly other libraries that your version of libtiff might use, e.g. zlib). Basic instructions for installing from source are: extract with tar xfj <file>, compile and install with ./configure ; make ; make install.

Using the software

Under Windows, you can simply drag and drop a TIFF (.tif) file over the .exe file or icon of the relevant program. tiffmakemosaic.exe and tiffsplittiles.exe will open the file and, if appropriate, split it into multiple TIFF image files (while preserving the original). tiffmakemosaic-j.exe does the same as tiffmakemosaic-j.exe, but stores the resulting images into JPEG files (option -j is hardcoded). For options, please refer to the documentation of each program for details.

On all platforms (including Windows, Linux, Mac OS X), you can also use it with a command line. Open a shell (e.g. command interpreter,, xterm...) and launch the program by typing its name (preceded by its path if needed) followed by a space then the path to the TIFF file. The resulting TIFF or JPEG file(s) will be produced in the directory where the original TIFF file resides.

A typical use is for anatomopathology or microscopy slide images, like the one that the NDPITools produce.


In principle, generating pieces from a large TIFF file can also be achieved with several tools, as tiffcrop from the libtiff, ImageMagick, and GraphicsMagick (one has to first compute and specify explicitly the dimensions and positions of the pieces, though). However, most of the software start with opening and deciphering the whole image either in memory or in a huge temporary file on the disk, which makes them quite slow or often unable to complete the task by lack of memory.

In contrast, tiffmakemosaic, tifffastcrop and tiffsplittiles avoid opening the whole image, which yields speedup and guarantees successful termination of the process even on computers with modest memory. Eg. to make a mosaic of 64 JPEG files requesting less than 512 Mib of memory to open from a RGB image of 103168x63232 pixels, on a computer with 16 Gib of RAM and an i5 CPU, tiffmakemosaic needs 2.5 minutes while GraphicsMagick needs 70 minutes.

Non-exhaustive list of enhancements

Credits and acknowledgments

This software was developed by the modelling team of IJCLab (formerly IMNC) near Paris, France, during a research project funded by the IN2P3 and INSB Institutes of the CNRS and by Université de Paris (formerly the Paris Diderot-Paris 7 university) and the university Paris-Saclay (formerly the university Paris South-11).

Contact: Christophe Deroulers

Linux, Mac OS X and Windows are registered trademarks of their respective owners.

Valid HTML 4.01 Transitional