2010.06.23 17:14 "Re: [Tiff] LZMA2 codec for libtiff", by Bob Friesenhahn

On Jun 23, 2010, at 10:16 AM, Andrey Kiselev wrote:

Compression is better than the one with Defalte codec for almost all data files I've tried. Sometimes it is MUCH better.

You might play with bzip. Same advantages re open implementation and patent-free. In my (very limited) tests, compression was only slightly worse than lzma, and performance was much better.

While I have not tried Andrey's patch, I have done testing in the past of bzip2 and lzma on whole image files and found that (for the images I tested) lzma did not seem to offer much compared with bzip2. Results are quite different with ASCII text where lzma is clearly a win. There may be particular images where lzma is a clear win. It is useful to do a study of file size, compression times, and TIFF predictor influence to understand the issues.

My interest tends in a different direction. It would be nice to have a very fast compression algorithm in TIFF which compresses faster than LZW or gzip yet is similar to 'gzip -3' compression levels. This compressor could be the compression used in lzop. The value of this is that uncompressed TIFF files can be excessively large and burdensome for system I/O whereas compressors are often quite CPU bound. Using the right compression algorithm may improve read/write performance for large files.

Bob
--
Bob Friesenhahn
bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/