[gdal-dev] Q: on gdalwarp of ecw file

Tue Dec 18 04:16:16 PST 2012

Jan,

Its in the docs (http://gdal.org/frmt_gtiff.html)

About JPEG compression of RGB images

When translating a RGB image to JPEG-In-TIFF, using PHOTOMETRIC=YCBCR can make the size of the image typically 2 to 3 times smaller than the default photometric value (RGB). When using PHOTOMETRIC=YCBCR, the INTERLEAVE option must be kept to its default value (PIXEL), otherwise libtiff will fail to compress the data.

Note also that the dimensions of the tiles or strips must be a multiple of 8 for PHOTOMETRIC=RGB or 16 for PHOTOMETRIC=YCBCR

From: Jan Hartmann <j.l.h.hartmann at uva.nl<mailto:j.l.h.hartmann at uva.nl>>
Date: Tuesday, December 18, 2012 6:12 AM
To: Michael Smith <michael.smith at erdc.dren.mil<mailto:michael.smith at erdc.dren.mil>>
Cc: Jukka Rahkonen <jukka.rahkonen at mmmtike.fi<mailto:jukka.rahkonen at mmmtike.fi>>, "gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>" <gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>>
Subject: Re: [gdal-dev] Q: on gdalwarp of ecw file

Any ideas why this should be so, Mike? Is this the case for all images? I found a few pointers with Google, like the one below, but I am still unsure about the reasons.

http://boatfloater.wordpress.com/2012/02/13/photometric-rgb-ycbcr/

On 12/18/2012 12:59 PM, Smith, Michael ERDC-RDE-CRREL-NH wrote:
I've found setting PHOTOMETRIC=YCBCR to allow much higher compression when using jpeg compression in gtifs.

Mike

--
Michael Smith
US Army Corps
Remote Sensing GIS/Center

From: Jan Hartmann <j.l.h.hartmann at uva.nl<mailto:j.l.h.hartmann at uva.nl>>
Date: Tuesday, December 18, 2012 5:47 AM
To: Jukka Rahkonen <jukka.rahkonen at mmmtike.fi<mailto:jukka.rahkonen at mmmtike.fi>>
Cc: "gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>" <gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>>
Subject: Re: [gdal-dev] Q: on gdalwarp of ecw file
Resent-From: Michael Smith <michael.smith at usace.army.mil<mailto:michael.smith at usace.army.mil>>

Thanks Jukka, very informative. I'll start with jpeg compressed tiffs and will do additional tests next year. I'm on a Cloud environment nowadays, so it's easy to set up clean VMs for testing. Any suggestions for experiments with different raster formats would be very welcome.

Jan

On 12/18/2012 12:33 PM, Jukka Rahkonen wrote:

Jan Hartmann <j.l.h.hartmann <at> uva.nl> writes:

    Hi Even, are there any
      benchmarks to compare (uncompressed) gtif with the three formats
      above? My production maps are always tiled to 2000*2000 pixels,
      all zoomlevels precomputed. Very efficient with uncompresssed
      gtif, but they take lots of space. How much slower are these
      formats?
      Jan

Hi,

Here are some 6 years old numbers from some Mapserver tests:
http://permalink.gmane.org/gmane.comp.gis.mapserver.user/23540

The speed feels now rather slow compared with the feeling I have from our recent
services but differences may be alike. Or then they are not because drivers are
not the same anymore and new hardware can suit better for one than for another.

We are using aerial images mostly as JPEG compressed tiffs now but I have not
bothered to do any timings lately. They save 90-95% of disk space, they are not
much slower and users have not noticed the difference in quality so for us the
choice was pretty easy.

It is usually too simplified to say that some file format is slower that some
other because there are so many other factors also in play. JPEG200O is an
infamous example. It is a complicated file format and many software are very
slow with big geospatial JPEG2000 images, but there are applications which can
handle them very fast. Sluggish software does not necessarily mean that the
format itself is slow. Six year old numbers 270/120/24 images per minute for
tiff, ECW, and JPG2000 with Kakadu prove mainly that there must have been
something wrong in how Kakadu and GDAL were working together at that time.

You can get the best information about the real speed in your environment by
making your own tests. Others will praise you later if you make controlled tests
and publish the arrangement and results somewhere for future references.
Unfortunately I do not remember myself how I did my own tests but I believe I
was using FWTools 2.0.4 binaries.

_______________________________________________
gdal-dev mailing list
gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>http://lists.osgeo.org/mailman/listinfo/gdal-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20121218/5beec5ed/attachment.html>