Pops_uk said:
A long but interesting article against the storing of images in the JPG format can be found here
That article's demonstration is a great example of what happens to images of that type. Meaning images with hard graphic lines and saturated and solid colors. That image is about as unlike any typical photograph as I can imagine and I don't see it as particularly relevant. Additionally, magnified views must be shown in order to make the damage easily perceived. I'm betting the differences - even with this poor example - would be difficult to pick up if the image were printed at 200 DPI.
It is absolutely true that JPEG compression introduces artifacts and destroys data that cannot be recovered. The real question though, is how much data is lost and is it significant?
The following is a quote from a "challenge" I posted during a discussion of this topic on a camera forum.
............................................................
Based on my experience in the world of professional video and the
used of Motion JPEG for video production, I entered the world of
digital photography with the firm conviction that multiple
applications of JPEG compression is a "bad" thing. But discussions
in another groups and actually doing some comparison convinced me
that the difference between saving in a lossless format and in JPEG
at the Maximum quality setting in Photoshop was so small as to have
no practical significance. I have just done a new comparison and the
sample images I created can be found in this directory:
http://www.jayandwanda.com/JPEG/
(But I really urge people to do critical tests like these for
themselves.)
What I did was open up an image that was taken with my CP995 at Full
size but with Normal JPEG compression (not FINE, even though I
usually shoot at FINE). I made a simple adjustment to the levels to
get the backlit flowers to "pop" a bit. I then immediately saved the
file as an LZW compressed TIFF and as a JPEG at the Maximum quality
setting. (The LZW TIFF is about 5MB and the Maximum quality JPEG is
about 2MB.) The images were re-opened and cropped sections were set
side-by-side at 300% magnification. I saved that view and it is
located at:
http://www.jayandwanda.com/JPEG/compare.jpg
If people think they can tell me which side is the TIFF and which is
the JPEG, I'd be happy to compile the results and report back how
well they jibe with the truth. It is my opinion that there is no
significant difference even though [most pixels are] slightly different.
The fact is that the various RGB values for each pixel seem to vary
by a value of 0 to 3. If you copy one image over the other in
Photoshop and set the top layer to "Difference", you can then move
your cursor over the apparently black image and read the difference
between pixels in the Info window. It is typically something like
0,1,1 or 0,0,2. So the additional JPEG save cycle does degrade the
image - but it is a very, very small amount that I maintain is not
visually perceivable.
I suggest people check for themselves. For myself, I no longer save
TIFFs. I save Maximum quality JPEGs instead. I might save the
occasional complex layering job as a PSD, but this is rare.
.........................................................
I have also found that many images can be save ten or more times at Photoshop's highest JPEG quality setting and the error stacking isn't much more significant. But test for yourself.
Any digital file can be corrupted. JPEGs are no different. I have seen no objective data that they are any worse and my experience suggests no such fragility. But why take chances with any digital file? I make duplicate archives of my origninal files (which are Fine JPEGs for the most part) on good quality CD-R. I can always go back to them if I need all of my original data. If you are really concerned, store the second archive set off-site somewhere.
If I modify the image, I save it as a JPEG with Photoshop's best quality setting. I can't see a practical difference at 300% magnification which leads me to believe that nobody can see the difference when printed.