[gdal-dev] Question about strategy on building big mosaic

[David Fawcett]

I am working on a project where am creating (and re-creating) a large image
from ~125 image tiles.  Each tile is a 5000x5000px jp2000 compressed image,
with no internal tiles.

I have a vrt set up and I have merged them all together using
gdal_translate, but it took ~20 hours to run on an admittedly old Macbook
Pro.  Even explained that the large input tiles make it really inefficient
for the OpenJPEG2000 driver to read the input images as part of the
translate operation.

I am trying to figure out if it is a reasonable strategy to introduce an
intermediate step where I create tiled copies of the input tiles.  At the
same time, I don't want to decrease image quality with the intermediate
step.

If I used -co REVERSABLE=YES, would I be able to end up with the same
quality in my final output?

I realize that I will have to do the performance test to determine if it
makes sense to add this intermediate step from a processing time
standpoint, but does this sound like a reasonable approach?

Thanks,

David.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20140529/7cbfb647/attachment.html>