Frank Warmerdam wrote: > Gilles Bassière wrote: >>> with COMPRESS=JPEG: >>> >>> Creating output file that is 10000P x 10000L. >>> Processing input file ortho-dept04-2004-la2e.vrt. >>> 0...10...20...30..gdalwarp: tif_jpeg.c:691: JPEGPreDecode: Assertion >>> `sp->cinfo.comm.is_decompressor' failed. >>> Aborted > > Gilles, > > The likely solution to this problem is to build GDAL with > the internal copy of libtiff. I have fixed a variety of > bugs related to random read/write support in libtiff on > compressed images and some of those fixes have not yet made > it out into distributions copy of libtiff. > >> The largest image has a black gap in the middle instead of the actual >> data. The smaller image doesn't show this gap and prove that there is >> actually data in the input dataset. > > I suspect black gaps would be due to a problem introduced in > GDAL warp for 1.7.x and only recently fixed. I would suggest you > build GDAL from either a stable or trunk snapshot so you will > have the fix for this problem with will also appear in 1.7.2 when > it is released. > >> Could this be due to wrong values read in the ECW input files? > > While I've had lots of problems with ecw over the years, I don't > see any obvious sign it relates to the ecw side of things. > > Best regards,
Frank, I've compiled the svn-stable daily snapshot and used --with-libtiff=internal. It seems to solve my problem. I don't see anymore black area and the Tiff/JPEG driver doesn't throw any error. Thanks you very much -- Gilles Bassière - Web/GIS software engineer http://gbassiere.free.fr/ _______________________________________________ gdal-dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/gdal-dev
