On Wed, 13 Jan 1999, Richard Ross wrote:
> As I understand it, if you open a jpeg file and then save it again, you
> will lose more information, and you'll lose more each time you do this.
> The amount of loss depends on the compression ratio.
Provided that you have the same compression ratio, the iteration (open -
save - open - save ...) does not cause "much" information loss compared
to the first JPEG file. Due to the same quantization level of DCT
coefficients of the 8x8 blocks, they somehow converge in one or two
iterations. We had tested that in the signal processing lab.
HOWEVER(!), if you crop or resize or do any kind of "touch", things
change, of course.
> Formats such as TIFF, GIF etc use non-lossy compression, i.e. all the
> information in the original is retained throughout the compression /
> decompression process.
GIF sucks. It is indeed a very lossy coder. It codes only 256 colors with
a stupid color quantizer, therefore your image gets dithered and has the
un-appealing look. JPEG preserves the color depth, therefore should be
preferred for photo scans. You can notice the color loss effect by
switching between 24bit(or 16) and 256 color modes of the computer while
displaying a 24 bit jpeg. Quite annoying.
> JPEG gives you nice small files for use on the web etc, but you will lose
> information when you use it. I didn't know until recently that the process
> is not reversible - if you save, open, save, the two saved files aren't the
> same.... AFAIK only the JPEG and MPEG formats suffer from this.
There was (and somehow, is) a huge research on lossy and lossless image
compression methods. JPEG is an old standard, therefore adopted by the
browsers. I strongly recommend the SPIHT coder (a wavelet based very
smart and fast coder) for coding your images for your own backup. It is
really visually transparent at 1:4 to 1:8 compression ratios. Furthermore
they provide a truly lossless coder on their web site, as well.
For SPIHT, check (their executables are free):
http://ipl.rpi.edu/SPIHT/spiht3.html
For a comprehensive list of image compression standards, reasearch,
software links, check:
http://image.kongju.ac.kr/~jwkim/splguy/@CompPoint.html
Best,
OMer
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|