Le jeudi 27 octobre 2016 18:47:38, Peter Schmitt a écrit : > On Thu, Oct 27, 2016 at 10:42 AM, Even Rouault <[email protected]> > > wrote: > > Le jeudi 27 octobre 2016 18:33:01, Peter Schmitt a écrit : > > > On Wed, Oct 26, 2016 at 1:39 PM, Even Rouault < > > > > [email protected]> > > > > > wrote: > > > > I've written recently a best practice for generating cloud optimized > > > > geotiffs: > > > > https://trac.osgeo.org/gdal/wiki/CloudOptimizedGeoTIFF > > > > (it mentions deflate compression, but jpeg-compressed can work of > > > > course) > > > > > Thanks again, Even! I did some experimentation with a 19584x19584 file > > > > with > > > > > 6 overview levels. When the image is deflate-compressed: > > > > gdal_translate in.tif deflate.tif -co TILED=YES -co COMPRESS=DEFLATE > > > > -co > > > > > COPY_SRC_OVERVIEWS=YES > > > > > > then this gdalinfo command does two range requests: > > > > env AWS_ACCESS_KEY_ID=foo > > > > > > AWS_SECRET_ACCESS_KEY=bar CPL_VSIL_CURL_ALLOWED_EXTENSIONS=.tif > > > CPL_CURL_VERBOSE=YES > > > VSI_CACHE=TRUE gdalinfo /vsis3/pschmitt-test/deflate.tif 2>&1 | grep > > > "Content-Range:" > > > > > > > < Content-Range: bytes 0-49151/1038305419 > > > > < Content-Range: bytes 49152-147455/1038305419 > > > > > > When I jpeg-compress > > > > > > > gdal_translate deflate.tif jpeg.tif -co TILED=YES -co COMPRESS=JPEG > > > > -co > > > > > > PHOTOMETRIC=YCBCR -co COPY_SRC_OVERVIEWS=YES > > > > > > then gdalinfo on the jpeg-compressed image does 13 range requests: > > > > env AWS_ACCESS_KEY_ID=foo > > > > > > AWS_SECRET_ACCESS_KEY=bar CPL_VSIL_CURL_ALLOWED_EXTENSIONS=.tif > > > CPL_CURL_VERBOSE=YES > > > VSI_CACHE=TRUE gdalinfo /vsis3/pschmitt-test/jpeg.tif 2>&1 | grep > > > "Content-Range:" > > > > > > > < Content-Range: bytes 0-49151/83731710 > > > > < Content-Range: bytes 83656704-83705855/83731710 > > > > < Content-Range: bytes 83722240-83731709/83731710 > > > > < Content-Range: bytes 83705856-83722239/83731710 > > > > < Content-Range: bytes 25231360-25280511/83731710 > > > > < Content-Range: bytes 25165824-25214975/83731710 > > > > < Content-Range: bytes 25214976-25231359/83731710 > > > > < Content-Range: bytes 7602176-7651327/83731710 > > > > < Content-Range: bytes 2228224-2310143/83731710 > > > > < Content-Range: bytes 655360-704511/83731710 > > > > < Content-Range: bytes 229376-278527/83731710 > > > > < Content-Range: bytes 131072-180223/83731710 > > > > < Content-Range: bytes 98304-131071/83731710 > > > > Could you post the outpout of "tiffdump jpeg.tif" ? There are some hacks > > in libtiff regarding JPEG compression and in particular JPEG tables that > > might perhaps cause spurious IFD rewriting. > > Here's the ouput of "tiffdump jpeg.tif": http://pastebin.com/raw/pX6XMYVT
That's what I suspected. The IFD are not placed at the beginning of the file, but rather rewritten after each "data section". Which libtiff version is used in your GDAL build: internal libtiff or system libtiff (if so, which version)? But I wouldn't surprise that the issue exists with any libtiff version because of what I mentionned with jpeg compression. That might probably be fixable (likely with some headaches) -- Spatialys - Geospatial professional services http://www.spatialys.com _______________________________________________ mapserver-users mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/mapserver-users
