<http://www.xat.com/jpegopt/index.html> ("Selective optimization -
compress some areas more than others") too.
If I am not mistaken, JPEG compresses images in blocks of 8x8 pixels?
The best idea I could come up with so far with is to create the
different 8x8 blocks individually with different target compressions and
then use the -drop option of jpegtran to merge them into a single image
without loss/recompression. I will be testing the idea in the coming hours.
I think the problem is that JPEG compression is mostly based on dropping
not so important frequencies after the discrete cosine transform. And
what frequencies are "not that important" is jugded on the complete
content of the picture and stored only for the complete picture.
There is a really nice visualisation of the single steps in JPEG Com
pression (deplorably only in German, but you'll get through it, I guess)
at http://www.spemaus.de/studium/visjpeg/applet.html
Greetings from Germany
Wolfgang Hugemann
_______________________________________________
Magick-users mailing list
[email protected]
http://studio.imagemagick.org/mailman/listinfo/magick-users
