Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 




      :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[filmscanners] Re: JPEG Lossless mirror?

---  <david_bookbinder@sprynet.com> wrote:
> I believe you are not correct, here. I have read in several accounts,
> both
> from people who have tried this experimentally and from people who
> understand the theory of JPEG compression

Well, then it's probably because these people don't know how to do an
experiment or they only know the theory. Do the following:

1) Open i.e. a *.tiff file in PS.
2) Save it as jpeg at quality 12 -> *1.jpg
3) Close the file.
4) Reopen the jpeg file (*1.jpg)
5) Save this file as jpeg at qulity 12 ->*2.jpg (If does actually
compress it again although it could just copy the original file and
rename it since no alteration took place. But that would be streching
it a bit)
6) Close the jpeg file (*2.jpg)
7) Open both jpeg file (*1.jpg and *2.jpg)
8) Go to Image->Calulations
9) Chose *1.jpg as source 1 (background and red, blue or red channel)
10) Chose *2.jpg as source 2 (background and same channel as for
11) Chose Difference for Blending, Opacity 100%
12) Chose New Document for Result
13) Since the error is very small you have to adjust the levels. The
easiest to just use 'Image->Adjust->AutoLevels' = 'Shift-Ctrl-L'

If you wish you can also do the following additional experiment:
14) Open the original *.tiff file
15) Calculate the difference but know between *.tiff and *1.jpg (step
16) Do Shift-Ctrl-L to adjust the levels.

You will see that in the second case the error is bigger. You start
seeing contours of your object and that although you stored the image
at the highest jpeg compression. You actually can look at the jpeg file
in a hex monitor and figure out the quantization coefficients PS uses.
Now if you had chosen a lower jpeg compression you would see an even
bigger error (because the quantization coefficients are bigger).
Now in the first experiment you see only a small error. You don't see
any contours which is why I said it's noise like. Now, I am not sure if
for each consecutive saving of the image the same amount of loss would
occur. Maybe not. It's possible that the loss will gradually decrease
until there is no loss anymore at all. Not sure about that.

> (Though
> why you would want to open a JPEG and then resave it unchanged, at
> the same
> compression ratio, rather than simply closing it, is an open
> question.)

Well, you might want to change only a small part of the jpeg image,
i.e. to remove a person etc. After the change obviously you have the
actually save the image. But now not only the area where you have made
the change but the whole image will be affected. Again there would be a
way to keep the rest of the image as it was and just recompress the
changed part (with a max of an additional 7 neighbour pixels in each
direction) but that would be asking too much.


Do You Yahoo!?
Send FREE Valentine eCards with Yahoo! Greetings!

Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 


Copyright © Lexa Software, 1996-2009.