While I don't work in GIS myself I'm trying to support one of our users here who looks after all our mapping and GIS data. We've received a 3.2GB Tiff image containing satellite imagery of our site and I'm wanting to find a way to re-size it down a bit to make it easier to work with.

Is there any open source/free software out there that can handle this sort of file size?

I've tried Irfanview, Imagemagick, GIMP without success.

The servers I'm trying to work on have plenty of memory and disk available but perhaps not enough to fit the whole file.

Is there an easy way?


  • When you say "re-size it down...", is creating a series of smaller images a viable option, or do you need to keep as a single file?
    – user890
    Oct 11, 2010 at 16:46

4 Answers 4


If you can, use GIS software, which is designed with this problem in mind: instead of reading the entire dataset into memory, it will only sample the image to create a display and no more. Something like QGIS should allow you to visualize the data, and provides ways of exporting the view, as one approach to creating a downscaled version.

Another option is to use something like gdalwarp, part of GDAL. In example, if the raw image was 10000x6000, you could make a half size in each dimension version by doing:

gdalwarp -ts 5000 3000 big-input.tif downscaled-output.tif
  • I agree when you say that GIS software should be used if possible. Anyway, I think that there may be another issue. If mcd's computer is 32-bit, given the theoretical 4GB RAM memory limit these kind computers have and the fact that OS and other applications also consume memory, processing a 3.2GB image can be beyond machine's capabilities (I do know that virtual memory is there, but in this case the system might become irresponsive). Likely GIS aware software can internally process the image by parts, so memory limit is not an issue.
    – dariapra
    Oct 11, 2010 at 13:03
  • Thanks scw, gdalwarp did the trick. Took quite a while to process it but it got the job done. Cheers.
    – mcd
    Oct 11, 2010 at 19:58
  • 1
    If you use tiling and compression the conversion will be more efficient and the result smaller, if your other software can support that. Use -co TILED=YES -co COMPRESS=LZW for simple options with gdalwarp. You should be careful to use BigTIFF variant as well for very large files.
    – mdsumner
    Nov 30, 2010 at 22:32

I agree with the gdal suggestion; gdal_translate will let you convert from GeoTIFF to a compressed format (e.g. MrSID, JPEG2000, etc. - I use .ecw but it has licensing issues...) Compression of 20x or greater is quite common. Be aware that if you don't use GIS specific software, you will likely lose the georeferencing information associated with the file.

If you must stick with GeoTIFF, you can use QGIS (or gdal again) to build 'pyramids' which will allow you to view less detail at smaller scales, thus not load the entire image at all times. In QGIS, look under the layer properties for 'Pyramids.'

Good luck!

  • It is important to remember that the formats listed are lossy. This can be good for end users but ultimately create a degraded version. It has its place (aerial photography), but is often the wrong solution compared with tiff/lzw. Mar 24, 2011 at 11:21

As @scw and @Darren suggested, using the GDAL utilities is a a good option. I've used them for elevation and imagery data whenever a file is too large to manage in a GUI editor. gdalinfo can give you the resolution and extents of the file and perhaps help you avoid loading the file in a GUI editor. I have used QGIS only a few times, so I don't know if building the pyramids will take a long time.


Same answer from another question -

OpenEV comes with FWTools, seems OK to me but I load 300 MB images into QGIS easily and deal with them there. So I can recommend it only from casual use.


Includes NITF

and is built with GDAL - but provides a convenient GUI

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.