Version 8c of libjpeg
Release Notes: A -block N option was added to the compression library and cjpeg for using different DCT block sizes. Values of 1 to 16 are possible, and the default is 8 (baseline format). Larger values produce higher compression, while smaller values produce higher quality. A SmartScale capable decoder (introduced with IJG JPEG 8) is required.
Other releases
Release Notes: The "-rgb" option was added to cjpeg to allow creation of RGB JPEG files. An option to disable the region adjustment in the transupp crop code was added. Various minor fixes were made.
Release Notes: A -block N option was added to the compression library and cjpeg for using different DCT block sizes. Values of 1 to 16 are possible, and the default is 8 (baseline format). Larger values produce higher compression, while smaller values produce higher quality. A SmartScale capable decoder (introduced with IJG JPEG 8) is required.
Release Notes: A problem in the new memory source manager with corrupt JPEG data has been resolved. An issue in the Makefile.am test target was fixed. MinGW installation is now supported with automatic configuration.
Release Notes: New data source and destination managers, jpeg_mem_src() and jpeg_mem_dest(), were added to the API. The accuracy of floating point IDCT calculations was improved. jpegtran now supports the same "-scale" option as djpeg for "lossless" resize. Support for 32-bit BPMs (RGB images with an alpha channel) was added. A potential crash in the BMP reader module was fixed.
Release Notes: New scaled DCTs were implemented. New API functions were added. Support for arithmetic entropy encoding and decoding was added. The support programs gained several new options, such as a new "lossless" cropping feature for jpegtran.
