Go Back   Steve's Digicams Forums > Digicam Help > Newbie Help

Reply
 
Thread Tools Search this Thread
Old Oct 23, 2004, 1:51 AM   #1
Senior Member
 
Join Date: Oct 2004
Posts: 141
Default

I keep reading that if we open a Jpg image, do some editing (adjust color, brightness, etc.), and then re-save it in its edited form, it loses quality each time this is done.

So Iopen a Jpg in AcdSee,adjust the exposure, or make a color photo into 256 Gray, and then save the edited image. The edited image is then larger(in terms of MB or KB) than the original.

So, what is happening here? Are the edited images truly of lesser quality than the originals? They don't appear on my computer monitor to be lesser quality. Would the lower quality of the edited images be evident in prints of, say, 8x10, or 11x14? If they are lesser quality, then why do the lesser quality images appear as larger files than the originals? I thought larger file sizes wasa good thing.

And what exactly does 'lossless' mean when saving an editedJpg image? How is 'lossless' achieved?

Thanks for any explanations which follow.
EOS RT is offline   Reply With Quote
Sponsored Links
Old Oct 23, 2004, 3:20 AM   #2
Member
 
Join Date: Apr 2004
Posts: 76
Default

Yes, its confusing, isn't it?

But no fear, light is ahead. When you edit a JPG, its decompressed to allow making these changes. When you then compress the image to a JPG format you have to choose (or your app just chooses) a quality level. If you choose a very high level, the file size is rather large without actually containing more details. If you had choosen just the same quality level as the JPEG was saved with, you would get almost the same file size (except for something extreme like greyscale conversion, but you could do this lossless anyway ).

The file size itself is only an indication for a picture quality for the unmodified picture straigth out of the camera. Image you would resave a picture with a very low quality level, open this again and resave then with a very quality level. The compression does not know that most of the details are gone during the first compression with low quality and now tries to compress preserving the most detail. A large file is the result but you don't get the quality back once it was discarded.

The loss less thing is something different. Instead of decompressing the JPEG, the loss less operations just work on the compressed image itself. With JPEG, all picture data is compressed in blocks. If you lossless rotate, these blocks are just rearranged. If you lossless crop, some of these blocks are just deleted. You never ever decompress, you work in the compressed domain all the time.

And, you cannot save an edited JPEG lossless (because you had it decompressed and therefore looses quality during the save if you are not careful).
Jurgen Eidt is offline   Reply With Quote
Old Oct 23, 2004, 10:43 AM   #3
Senior Member
 
Join Date: Dec 2002
Posts: 5,803
Default

To add another piece of info....
If you edit a picture and think "this looks good so I'll save it" and do so. Then you edit is some more and then save it and edit it some more and save it.

So far, you have not lost any data to distructive compression. All you've done is save the image a few times to make sure that what you've done so far isn't lost (if you screw up while doing more edits.

Now lets say you do this:
open the picture, edit it some, save it, edit it some more, save it, edit it some more... and decide that you dislike what you did. So you reload the image and start editing again.

Now you have lost some data & detail. Because the instant that you save a jpg and load it from disk you will load the copy which had another around of distructive compression applied.

Eric
eric s is offline   Reply With Quote
Old Oct 23, 2004, 9:46 PM   #4
Senior Member
 
BillDrew's Avatar
 
Join Date: Jun 2002
Location: Hay River Township, WI
Posts: 2,512
Default

I agree with Jurgen and eric, but would like to emphisise one point.
EOS RT wrote:
Quote:
... then re-save it in its edited form, it loses quality each time this is done.
...
Do not overwrite your original photos. Ever. Save them exactly as they come from your camera with no changes of any kind whatsoever.

If you have your original images, it really doesn't much matter if you trash any of the copies while you are learning. You can always go back to the original and start over. Save your "digital negatives" and don't worry.
BillDrew is offline   Reply With Quote
Old Oct 24, 2004, 12:41 AM   #5
Guest
 
Posts: n/a
Default

I favor (& use quite a bit) Genuine Fractals. Whenever I open a RAW file, I immediatly save it as a GF-lossless ".stn" file. I then do all my work from that file, plus I can increase the file size while ususally improving the image quality. Some people don't really look at this file type as valid (due to archiving reasons), but I have experienced nothing but excellent results from it's use. You can try it yourself here: http://www.lizardtech.com/solutions/gf/

I forgot to add...GF normally reduces the file size by 20 to 40% while remaining lossless.
  Reply With Quote
Old Oct 24, 2004, 2:21 AM   #6
Senior Member
 
Join Date: Feb 2004
Posts: 175
Default

Kalypso wrote:
Quote:
I favor (& use quite a bit) Genuine Fractals.
Is GF the only software that can work with .stn files? I have an older version of GF somewhere, used it a couple times for upsizing with excellent results. I just never reinstalled after flattening my HD.

I remember it seemed kind of cumbersome to work with but it did interpolate very well, sometimes almost seeming to create detail intelligently. Could you elaborate a bit on GF's place inyour workflow?

David
Amateur is offline   Reply With Quote
Old Oct 24, 2004, 2:35 AM   #7
Senior Member
 
Join Date: Oct 2004
Posts: 141
Default

Wait a minute...

According to what Eric wrote, I can brighten or darken a Jpgimage in ACDSee,save it, and NOT lose any quality? And I can brighten/darken it several times, without it losing any quality?

Is this correct, or am I misunderstanding something?

As well, is this also the case when converting a color Jpg image to 256 Gray, or to Sepia? Can this, too, be done without any loss of quality in the image?

Bill yes, I realized this. Fortunately, Ilost only one photo in discovering that one should always make a copy, and edit that, rather than edit the original. Thanks for the info, though.
EOS RT is offline   Reply With Quote
Old Oct 24, 2004, 2:52 AM   #8
Member
 
Join Date: Apr 2004
Posts: 76
Default

What eric wrote is correct, but he meant something different.Its about the workflow. Cascaded edits will show some loss. For example, you edit an image, save it, open it again and then save it again. Every time you make a decompression/compression round trip you will loose some quality. When you have the image the first time in the editor and save it several times during your work, you only make a compression everytime but not the round trip. So there is no continous loss when you save after each step because everything is based on the first decompessed image. Umm, let me know if it makes sense to you.

You can make a lossless greyscale conversion, but not having a lossless special filter effect like Sepia. When the JPEG is compressed, the RGB data is converted tothecolor format YCbCr which separates color from brightness (Y is the brightness, Cb and Cr the color components). The lossless greyscale simply cuts off the color data Cb and Cr. JpegCrop and cPicture can do this lossless greyscale.

Check out http://jpegclub.org/losslessapps.html
Jurgen Eidt is offline   Reply With Quote
Old Oct 24, 2004, 2:58 AM   #9
Senior Member
 
Join Date: Feb 2004
Posts: 175
Default

I think Eric is referring to multiple save in a single editing session. In a single session even though you save, you are working on the original, until or unless you reload the image you have saved. Now you have stepped down one generation.

(I think)

is that right Eric?
Amateur is offline   Reply With Quote
Old Oct 24, 2004, 3:24 AM   #10
Senior Member
 
Join Date: Oct 2004
Posts: 141
Default

Ok - I get it now. That makes more sense. Thanks to you both.



And thanks for the lossless greyscale info, Jurgen.

One more question, though...

How can I tell if/how much quality has been lost in an edit? Ioriginally thought that one could tell based on how much of the file is lost in terms of size (for example, from 1.5 MB to 1.3 MB) - but I now know that's not the case. Is there any measure to tell how much quality has been lost? I'd like to know before I print, not after.

Thanks.



EOS RT is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 11:02 PM.