--- Jani Patanen <[EMAIL PROTECTED]> wrote:
> During the first save, there are some
> pixels
> that are "on the edge" of having another value. After the save, some
> nearby pixels have different values (because of the algorithm) and
> that
> causes those pixels that were "on the edge" to go "over the edge"
> during the next save. Naturally the number of pixels "on the edge"
> degrades on every save, finally coming to a stable situation.
This is actually a pretty good explanation. jpeg does consist of
multiple stages. Some of these stages have an exact reverse (coding
followed by decoding will give exactly the same result), others have
not and are also influenced by rounding errors. So while quantize and
'de-quantize' values multiple times will not change values anymore (as
explained earlier) the DCT has some rounding errors. This can influence
the quantization and change a value. But while it might not necessary
converge it does not diverge. But this error which result from loading
and storing multiple times without making any changes will not have a
visible negative effect. But loading, changing file, saving, loading,
changing,... will negatively influence the image.
Personally, if image quality is important I would chose the raw format.
This should save almost 50% of a tiff image but does not lose any
information. On a D1, and I assume it similar on a D30, one can see a
(small) difference even between tiff and the best (biggest) jpeg. With
one level loser (the middle level) the difference becomes quite
obvious.
Robert
__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/
*
****
*******
***********************************************************
* For list instructions, including unsubscribe, see:
* http://www.a1.nl/phomepag/markerink/eos_list.htm
***********************************************************
