JPEG degradation

Discussion in 'Digital Photography' started by TSKO, Jan 26, 2007.

  1. TSKO

    Ron Hunter Guest

    If one keeps the quality setting high (minimal compression, if any),
    then the loss of quality would be small, and the number of iterations
    before it became significant would probably exceed the number used for
    normal editing. I would hope that anyone who was sufficiently concerned
    about such degradation would be smart enough to use another, lossless,
    format for any editing iterations, and only save the final product to
    ..jpg format.
    Ron Hunter, Jan 27, 2007
    1. Advertisements

  2. TSKO

    HEMI-Powered Guest

    Today, Ron Hunter made these interesting comments ...
    Your assetion was my point exactly. I have found that to preserve
    fine detail in car pictures, my most common subject, that I must
    keep the general JPEG compression BELOW 12 on the standard 1-100
    scale (it is NOT a percentage as some would claim), often I must
    drop down to 10, 8, or even 6. And, I have 3 commonly used chroma
    subsampling settings which I vary according to the compromize I
    am trying to achieve between small size and least damage.

    If one is interested in absolute quality with no damage at all
    (beyond what they themselves might do <grin>) again, I agree with
    you, stay with a lossless format, preferably one which is
    reasonably universal such as TIFF or PNG. But, in order to hold
    proprietary elements, it can happen that the same image must be
    saved to at least 2 formats, sometimes 3 or 4. But, for us
    "normal" folk who want "good" pictures from "good" cameras", JPEG
    judiciously used can yield more than acceptable results except
    when examined under lab conditions or when look for specific
    kinds of damage. e.g., if one has to put text onto the image,
    JPEG is especially poor and will generally mangle the crispness
    of the text.
    HEMI-Powered, Jan 27, 2007
    1. Advertisements

  3. TSKO

    TSKO Guest

    Now how simple was that. Simple easy post to understand (and hopefully

    Thanx Ron for clearing up my question
    TSKO, Jan 27, 2007
  4. TSKO

    ASAAR Guest

    No, I vaguely recall the model number and it was nothing like
    3380. From what little I recall, 2314 seems more likely than 2319.
    It's just that the front of the "box" that's pulled out to remove
    the disk packs resembled what was described on the web page as a
    pizza oven (flatter) than the picture shown for the earliest 2314,
    and it also had those round selecter plugs that I didn't see in the
    ASAAR, Jan 27, 2007
  5. Why not give it a try, with the programs you actually use and the
    settings you actually use? It's really quick and easy, couple of
    minutes (per case). And then you'll actually know, and won't have to

    Note that I got the degradation on the very first testable resave; it
    didn't take many (I ran 5, but found I got clear enough results on #1).
    David Dyer-Bennet, Jan 27, 2007
  6. TSKO

    -hh Guest

    Believe it or not, some people still dispute that.

    Of course, they're probably also the type to also say things like they
    "cropped the image with JPEG compression" too.

    Sure. The challenge there is that the answer not only depends on the
    compression rates and number of repetitions, but also probably the
    nature of the scene.

    -hh, Jan 27, 2007
  7. That's because - by my observations - the biggest losses happens at the
    first iterations.

    I've put together an interactive page with some images, where it is
    possible to observe the degradation at different quality level.

    Give it a go:
    Toke Eskildsen, Jan 27, 2007
  8. The exact opposite happens when I try, except for a special case with a
    manifactured image, which should never have been saved as a JPEG in the
    first place.

    I've places my test images at .
    Toke Eskildsen, Jan 27, 2007
  9. Does that require scripting or Flash or something? Mine won't
    work - images just say LOADING.
    Paul D. Sullivan, Jan 28, 2007
  10. TSKO

    Ron Hunter Guest

    I believe the 3380 had the same arrangement, but larger disks.
    Ron Hunter, Jan 28, 2007
  11. TSKO

    Ron Hunter Guest

    Humm. What are you testing? Conversion to .bmp, or recompression. And
    is this valid for all programs (I doubt it), or only the program you used?
    Ron Hunter, Jan 28, 2007
  12. TSKO

    Ron Hunter Guest

    Certainly the nature of the scene has much to do with compression in
    that the size of a picture with grass and trees will be much larger, and
    compression artifacts more visible, and a picture of clear blue sky.
    However, the grass may look better from a distance than the sky does.
    Introduce some text into the image, and watch the fireworks.
    Ron Hunter, Jan 28, 2007
  13. Sorry to hear that. I used plain JavaScript, but I have only tested in
    IE7, Firefox 1.5 and Opera 9 under Windows. What are you using?
    Toke Eskildsen, Jan 28, 2007
  14. I'm forcing recompression by converting to JPEG => BMP => JPEG.

    That means I'm testing recompression, not how clever the program is to
    reuse information about the source format. Skipping the BMP step would
    mean zero change for some programs, as they would just copy the
    original file.
    Yes it is. It forces recompression for all programs.
    Toke Eskildsen, Jan 28, 2007
  15. TSKO

    Alan LeHun Guest

    Is this true? AFAIK(BICBW) the jpeg file format does not record the
    quality level that was used to compress it and so the program has no
    idea wether you are saving at the original compression ratio or have
    chosen a different level in which to save.

    If this is the case, then a well written program has little choice but
    to re-compress the image and save it.
    Alan LeHun, Jan 28, 2007
  16. Alan LeHun wrote:

    [Reuse JPEG information to skip compression]
    It has some idea - Thumbs+ gives a qualified guess as to the quality.
    If it the program itself that has produced the source, such a guess
    could very well be exact all the time.
    That's true.

    I mostly did the BMP-step to be sure that it wouldn't be a source of
    error. Call me paranoid.
    Toke Eskildsen, Jan 28, 2007
  17. Alan LeHun wrote:
    There is no single "quality level" measurement in JPEG, because there are
    different options as to how the compression is achieved.

    David J Taylor, Jan 28, 2007
  18. TSKO

    JC Dill Guest

    Opening JPEGs does not degrade them. SAVING JPEGs is what degrades
    them. When you open and read the original data is left unchanged.

    Web servers send the same jpeg file off thousands of times a day, or
    hour, or minute. The file doesn't degrade even though it is
    opened/copied/sent at a furious rate. :)

    JC Dill, Jan 28, 2007
  19. TSKO

    ASAAR Guest

    Sure it does. How else would you explain the problem Bret's
    having with his most recent blankety blank Pbase jpeg file?
    ASAAR, Jan 28, 2007
  20. I'm still loading it here after a couple of minutes, nothing visible in
    the page yet. Firefox, Windows XP SP1.
    David Dyer-Bennet, Jan 28, 2007
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.