Hello!
This is something that isn't really clear and conserns many of us
- specially those with EOS D30 - and that's why I don't even think it's
off-topic anymore on this list.... so here goes:
On Sat, 28 Jul 2001, Keith Green wrote:
> Jani Patanen wrote:
>
> > FYI. When you open file, save as jpeg, open, save, open, save etc.
> > the quality doesn't get worse after the first save.
>
> My apologies. I stand corrected. I have just tried it and the file size
> remained the same.
>
> I cannot remember where I first read this, but I accepted it as a fact
> without checking it.
Facts are as I know them to be:
- Jpeg is a lossy compression. That means you will lose some information
when saving with Jpeg compression
- When you open a jpeg file to Photoshop (for instance), it becomes just a
raw pixels image in PS. If you then save it from that, it will apply the
compression to the file again - after all, it assumes this is another
image you are saving. (It is possible that the program would notice that
you did absolutely nothing to the image, and would then not save the
image, just close it or copy it to another name.) So, when you save it,
you will loose some of the information again. But read on...
- So if you open and save, open and save, the image will degrade, but the
question is how and how much? I did do a test with Matlab where I did a
script that opened a JPEG image and then saved it, and I looped over that
many times. Every time I opened the new image, I substracted the pixels
from the image before that to see how many pixels had changed. Quite
surprisingly the change converged very fast (I don't remember how
fast, and it might depend on the image as well as the compression rate)
to an image that did not change anymore. It was in my case less that 10
saves and openings. And after just a 3-4 rounds, the change was already
quite small. This was surprising as I've never seen a mathematical proof
of this convergence (nor have I tried one myself) and with algorithm like
JPEG it is not that clear that it would converge to a point where the
compression algorithm doesn't change the image anymore.
- If the compression ratio was changed during the cycle even a little bit,
then the pixels would change again. But as long as the compression was not
heavy, visually the image did not degrade much more.
- I assume other programs handle Jpeg compression the same way as Matlab.
I.e. they image would converge in those too.
- I wish Photoshop would have an option to save Jpeg images with the same
compression they were originally made. I.e. the compression camera used.
Or with less compression of course.
One could probably do this kind of check with the current Photoshop
actions too. To me it was easier to do on Matlab (besides then I knew
exactly what I was doing.) if somebody has a new PS, I invite somebody to
try this. If I ever have time, I could also put my findings on some
web-page, but I doubd that I have time for that.
Anyways, that's my findings... you must all decide if the degration of the
Jpeg image during the first few saves is more than you can tolerate. I
think that the larger problem is the initial compression ratio. Me? Well
I'm still saving those large TIFF's from my scans... even though I'm not
sure if that is such important anymore.
Maybe the reasons for using RAW file format on D30 are different also...
for instace then you can supposedly alter the white balance, sharpness,
contrast and curves/levels/gamma/white point when you "develop" the RAW
image. That might make more of a difference than losing some information
with Jpeg compression.
Best regards,
Hugo.
************************************************************
** Hugo Gävert **
** [EMAIL PROTECTED] http://www.hut.fi/~hugo **
************************************************************
*
****
*******
***********************************************************
* For list instructions, including unsubscribe, see:
* http://www.a1.nl/phomepag/markerink/eos_list.htm
***********************************************************
