LibTiff Mailing List
TIFF and LibTiff Mailing List Archive
April 2012
Previous Thread
Next Thread
Previous by Thread
Next by Thread
Previous by Date
Next by Date
Contact
The TIFF Mailing List Homepage
This list is run by Frank Warmerdam
Archive maintained by AWare Systems
Thread
2012.04.28 15:10 "Re: YCbCr/JPEG issue in 4.0.1", by Charles Auer
Joris: Well, back to the bug reading JPEG (compression = 6) compressed images using Libtiff 4.0, the problem is only evident if you read the subsampled YCbCr data directly, doing resampling and color conversion yourself. For example, subsampling = [ 2, 2 ] data should look like this: Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr but instead you get this: Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Y Y Y Y Cb Cr Y Y Y Y Cb Cr Y Y Y Y Cb Cr 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 You can still recover the image by ignoring the zeroes and the error message. As a matter of interest, Joris, OJPEG (compression = 7) data are read correctly. Rewriting tif_ojpeg.c was a tremendous contribution that you made to Libtiff, by the way. Also, I made a set of YCbCr test images with all possible subsampling values ( [ 1, 1 ], [ 1 , 2 ], [ 1, 4 ], [ 2, 1 ], [ 2, 2 ], [2, 4 ], [ 4, 1 ], [ 4, 2 ], and [ 4, 4 ] ), both uncompressed and LZW compressed, and all of these images are read correctly. The problem appears isolated to JPEG (compression = 6) compressed images. The easiest work-around right now to handle JPEG compressed images with Libtiff 4.0 is either: 1. Use TIFFReadRGBAImage 2. Use TIFFReadEncodedStrip or TIFFReadEncodedTile with JPEGCOLORMODE = 1 Of course, both of these methods output RGB colors. Charles > Date: Sat, 28 Apr 2012 11:46:15 +0200 > Subject: Re: [Tiff] YCbCr/JPEG issue in 4.0.1 > From: info@awaresystems.be > To: bumble731@msn.com > CC: john.evans@mathworks.com; tiff@lists.maptools.org > > Charles, > > > All required support for > > subsampling and desubsampling as well as any color conversions, is > > merely part of this. > > I'd like to add this was not meant to be library bashing or trolling, > though thinking about it, it may have seemed as such... > > In fact, it's meant to be constructive. Since subsampling and color > conversions are on the outer rim (i.e. they are some of the first > independent features in the encoding pipeline, and some of the last > independent features in the decoding pipeline), it should actually be > very feasible to add proper and full and consistent support for this > to the existing LibTiff library, with minimum code change. > > > Best regards, > > Joris