Scanning Photo Prints: Max Rez?

Discussion in 'Photography' started by (PeteCresswell), Feb 15, 2013.

  1. Wrong on both counts: TIFF can use JPEG lossy encoding and
    JPEGs can be encoded losslessly. Uncommon, yes, but possible.

    -Wolfgang
     
    Wolfgang Weisselberg, Feb 19, 2013
    #21
    1. Advertisements

  2. The scanner won't produce results better than his optical
    resolution, no matter what you do, even if whatever you scan
    has way higher detail.
    Unfortunately, increasing the resolution of the scan *does*
    increase the information in the result. That's why a D800
    does better than less MPix-high cameras even with less than
    stellar lenses. Unfortunately again, the amount of extra
    pixels needed to see one unit of improvement does rise very
    sharply once it's no longer the main limitation ...
    OP says 1200 does better than 300 *on his scanner and photos*.
    Assuming he evaluated them after scaling up the 300 to the
    1200 size, he should scan better than at 300 dpi and should
    test at 400, 500 and 600 dpi, IF he's gonna use the scan as
    an archive copy (which would be a good idea).

    IF he's merely trying to put the images on the web or on the
    TV or make copies, and scan time and/or storage space are
    relevant limitations, lower resolutions are of course OK.

    http://www.clarkvision.com/articles/sampling1/index.html
    (scanning film, not paper) shows that you want at least 2
    pixels per cycle at the Rayleigh Limit but can get a bit of
    additional detail up to 3 pixels/cycle at the Dawes Limit if
    you want to get out everything.

    http://www.clarkvision.com/articles/scandetail/index.html
    explains more about the limits (and a lot more) and shows that
    you need /at least/ 400 dpi to get all the detail from a /sharp/
    print. Which means that the OP does not necessarily imagine
    things when he sees 1200 as better than 300, though 1200 dpi
    may be over the top. Only testing with values in between 300
    and 1200 dpi will tell for the OPs prints and scanner.

    -Wolfgang
     
    Wolfgang Weisselberg, Feb 19, 2013
    #22
    1. Advertisements

  3. (PeteCresswell)

    Robert Coe Guest

    : On 2/15/2013 6:33 PM, Tony Cooper wrote:
    : > On Fri, 15 Feb 2013 16:06:34 -0500, "(PeteCresswell)" <>
    : > wrote:
    : >
    : >> Per Anon:
    : >>> And please don't archive them in the jpg format.
    : >>
    : >>
    : >> .TIFF?
    : >>
    : >> Photoshop's format?
    : >
    : > Depends. Normally, Photoshop files are saved as a .PSD file. But
    : > both .TIFF and .PSD retain layers. .TIFF is also an Adobe format,
    : > .TIFF is a Lossless format, but .JPG is a Lossy format.
    : >
    : > For a detailed explanation, see
    : > http://www.cambridgeincolour.com/tutorials/imagetypes.htm
    : >
    : > --
    : > Tony Cooper - Orlando, Florida
    :
    :
    : To maximize compatibility, I think it would be best to also save the
    : scans as JPEG files, even if an archival format is used.
    :
    : By the way, your "sig delimiter" doesn't work, here in Mozilla
    : "Thunderbird" 17.0.2...hence, maybe, "nospam" was right?

    I don't know what "nospam" said, but Tony's sig delimiter is non-standard
    because it lacks a space after the two hyphens. (Which hardly matters when the
    sig is only one line.) So if Thunderbird fails to notice it, that behavior is
    correct.

    Bob
     
    Robert Coe, Feb 23, 2013
    #23
  4. (PeteCresswell)

    Robert Coe Guest

    : > .TIFF is a Lossless format, but .JPG is a Lossy format.
    :
    : Wrong on both counts: TIFF can use JPEG lossy encoding and
    : JPEGs can be encoded losslessly. Uncommon, yes, but possible.

    The problem with JPEG as an archiving format isn't data loss. It's the
    annoying tendency of newer photo editors not to accept older JPEGS. It may be
    that those older JPEGs (the ones that can no longer be read) were slightly
    non-standard. But when an older editor would read them and a newer version of
    the same editor won't, you have to regard that format as unstable.

    Bob
     
    Robert Coe, Feb 23, 2013
    #24
  5. (PeteCresswell)

    Robert Coe Guest

    : I,too, have delved into archiving old photographes and tried to find out
    : the best way to scan for the best possible results. The advice I
    : received was to scan at the optical resolution of the scanner, you can't
    : get any better result. Most, if not all, scanners can output a range of
    : resolutions, both higher and lower, but only work in the one, the
    : optical resolution. This would be the fastest scan as the time taken by
    : the software to produce a different resolution is avoided.
    :
    : Also the advice to save in a lossless format is very important for
    : archiving as you or someone else will need to reprocess the image later.
    : Convert to jpeg or whatever to distribute them to others is a good idea
    : as it reduces the storage size for those who don't need to.
    :
    : If your scanner will only produce images in jpeg, I don't know if and
    : do, convert them to a lossless format before doing anything else to
    : presurve the quality even saving into a converter before saving the file
    : if you can.

    Maybe I'm missing your point, but you seem to be saying to convert the photo
    to a non-JPEG format quickly, before the JPEG image starts losing data. That's
    not how it works. A JPEG file doesn't lose data while sitting around on a disk
    drive; the data loss is a consequence of the conversion to JPEG or from one
    JPEG encoding to another. So if your scanner produces only JPEG, and you save
    the JPEG image that the scanner produces, you've done the best you can do.
    Converting the image to a lossless format doesn't preserve any data that isn't
    in the JPEG image.

    There may be reasons for converting a JPEG image to another format (such as
    the fact that evolving photo editors sometimes lose their ability to read old
    JPEGs), but preventing data loss isn't one of them.

    Bob
     
    Robert Coe, Feb 23, 2013
    #25
  6. (PeteCresswell)

    Robert Coe Guest

    : On 17/02/2013 15:30, Charles E. Hardwidge wrote:
    : > I know people say scan to the maximum resolution (of the Lens->Film->Photo
    : > *system* which is limited by the lowest resolution component in theory)
    : > but... Film grain is often treated as waste yet it is an integral part of
    : > the image and does play a role in perceived sharpness. So I'm left
    : > wondering
    : > if it's better to scan to the maximum theoretical resolution or capture the
    : > grain as well. The issue is grain comes in all shapes and sizes and may add
    : > a quality to the image. Has this been tested? If so what larger resolution
    : > should be used to capture the photo? Two times? Four times? Larger?
    : >
    : This why I was advised to scan at the optical resolution of the scanner.
    : You can change the pixal count both up and down but there will be no
    : more information available. A lot of scanners can produce images with a
    : higher resolution number than the optical resolution but does that
    : improve the image?

    I think the answer is that it might if the interpolation methodology
    (algorithm, data resolution, control of roundoff error, etc.) used by the
    scanner is better than that used by your photo editor. Otherwise, probably
    not.

    Bob
     
    Robert Coe, Feb 23, 2013
    #26
  7. (PeteCresswell)

    Whiskers Guest

    The danger is that someone will edit the jpg file and save it, perhaps
    unintentionally, thus introducing further degredation in the image; like
    the 'photo-copy of a photo-copy of a photo-copy' problem. Using a
    'lossless' file format reduces that risk.

    No matter what the file format, it is of course good practice to make the
    original 'read-only' and create a duplicate before using any software that
    can edit the image.
     
    Whiskers, Feb 23, 2013
    #27
  8. (PeteCresswell)

    Alan Browne Guest

    No. There is no "extra" information to be had. Whether you scan it at
    300 or 600; or "re-sample" from 300 to 600 in the editor, all of the
    "extra information" is interpolation.

    You simply cannot get information that does not exist.
     
    Alan Browne, Feb 23, 2013
    #28
  9. (PeteCresswell)

    Alan Browne Guest

    There is so much information on the print and no more. The scanner may
    or may not interpolate but its ability to conjure up information that
    does not exist in the first place can be no better than a typical editor.
     
    Alan Browne, Feb 23, 2013
    #29
  10. (PeteCresswell)

    Robert Coe Guest

    On 2013.02.23 09:06 , Robert Coe wrote:
    : > : On 17/02/2013 15:30, Charles E. Hardwidge wrote:
    : > : > I know people say scan to the maximum resolution (of the Lens->Film->Photo
    : > : > *system* which is limited by the lowest resolution component in theory)
    : > : > but... Film grain is often treated as waste yet it is an integral part of
    : > : > the image and does play a role in perceived sharpness. So I'm left
    : > : > wondering
    : > : > if it's better to scan to the maximum theoretical resolution or capture the
    : > : > grain as well. The issue is grain comes in all shapes and sizes and may add
    : > : > a quality to the image. Has this been tested? If so what larger resolution
    : > : > should be used to capture the photo? Two times? Four times? Larger?
    : > : >
    : > : This why I was advised to scan at the optical resolution of the scanner.
    : > : You can change the pixal count both up and down but there will be no
    : > : more information available. A lot of scanners can produce images with a
    : > : higher resolution number than the optical resolution but does that
    : > : improve the image?
    : >
    : > I think the answer is that it might if the interpolation methodology
    : > (algorithm, data resolution, control of roundoff error, etc.) used by the
    : > scanner is better than that used by your photo editor. Otherwise, probably
    : > not.
    :
    : There is so much information on the print and no more. The scanner may
    : or may not interpolate but its ability to conjure up information that
    : does not exist in the first place can be no better than a typical editor.

    Well, of course it *can* be. (I'm sure you wouldn't claim that every editor
    does the best possible job of interpolation.) But I'll readily concede that it
    probably isn't.

    Bob
     
    Robert Coe, Feb 23, 2013
    #30
  11. (PeteCresswell)

    Robert Coe Guest

    : > : I,too, have delved into archiving old photographes and tried to find out
    : > : the best way to scan for the best possible results. The advice I
    : > : received was to scan at the optical resolution of the scanner, you can't
    : > : get any better result. Most, if not all, scanners can output a range of
    : > : resolutions, both higher and lower, but only work in the one, the
    : > : optical resolution. This would be the fastest scan as the time taken by
    : > : the software to produce a different resolution is avoided.
    : > :
    : > : Also the advice to save in a lossless format is very important for
    : > : archiving as you or someone else will need to reprocess the image later.
    : > : Convert to jpeg or whatever to distribute them to others is a good idea
    : > : as it reduces the storage size for those who don't need to.
    : > :
    : > : If your scanner will only produce images in jpeg, I don't know if and
    : > : do, convert them to a lossless format before doing anything else to
    : > : presurve the quality even saving into a converter before saving the file
    : > : if you can.
    : >
    : > Maybe I'm missing your point, but you seem to be saying to convert the photo
    : > to a non-JPEG format quickly, before the JPEG image starts losing data. That's
    : > not how it works. A JPEG file doesn't lose data while sitting around on a disk
    : > drive; the data loss is a consequence of the conversion to JPEG or from one
    : > JPEG encoding to another. So if your scanner produces only JPEG, and you save
    : > the JPEG image that the scanner produces, you've done the best you can do.
    : > Converting the image to a lossless format doesn't preserve any data that isn't
    : > in the JPEG image.
    : >
    : > There may be reasons for converting a JPEG image to another format (such as
    : > the fact that evolving photo editors sometimes lose their ability to read old
    : > JPEGs), but preventing data loss isn't one of them.
    : >
    : > Bob
    :
    : The danger is that someone will edit the jpg file and save it, perhaps
    : unintentionally, thus introducing further degredation in the image; like
    : the 'photo-copy of a photo-copy of a photo-copy' problem. Using a
    : 'lossless' file format reduces that risk.

    What reduces that risk is the use of a non-destructive editor. If your editor
    saves only the changed form of your file, the inadvertent damage you can do
    greatly transcends the effect of using a lossy versus a lossless format.

    : No matter what the file format, it is of course good practice to make the
    : original 'read-only' and create a duplicate before using any software that
    : can edit the image.

    And my point was simply that if the original is a JPEG, it does no good to
    convert it to another format before preserving it, unless you're concerned
    that an editor you may be using in the future will have trouble reading
    today's JPEGs.

    Which raises a related question: How do the commonly used editors differ in
    their ability to handle old or unconventional JPEGs?

    Bob
     
    Robert Coe, Feb 23, 2013
    #31
  12. True, but the idea that a typical print has no more information
    than can be gathered at 300 dpi is false. Second, whatever the
    maximum dpi of the informatio on the print, the scanner has to
    scan at twice that rate to record that information.

    Archival scanning at anything less than the maximum optical rate
    the scanner can achieve is not smart.
     
    Floyd L. Davidson, Feb 23, 2013
    #32
  13. This problem of converting to a lossless format versus the original lossy
    format strikes me as being one of those "what problem are you trying to
    solve" questions. People wishing to create an archive quality chain may
    prefer converting to a lossless format. Other people with more general
    requirements may not find it worth the bother.

    Asking this question first should do away with most hair splitting and turf
    defending arguments.

    There's more than one way to skin a cat, as they say.
     
    Charles E. Hardwidge, Feb 23, 2013
    #33
  14. ^
    What I said.
     
    Charles E. Hardwidge, Feb 23, 2013
    #34
  15. (PeteCresswell)

    DaveL Guest

    No I'm not that daft/stupid!
    My advice is to convert your jpeg to a non compressed format and then
    work on it, make any changes such as retouching or colour/color
    correcting in that format and then chage it back to jpeg if you want to.
    I do know that in the jpeg format you can degrade the image when you
    resave it in that format.
    Dave L
     
    DaveL, Feb 23, 2013
    #35
  16. One more reason to use open source software --- there you
    won't find such stupidity.[1]
    In that case you should never use TIFF. Almost no TIFF reading
    software can read every completely correct form of TIFF.
    I wouldn't even bet on every TIFF baseline variant to be
    readable by every TIFF reader.

    I wouldn't call the format as "unstable", but the editors ...

    Anyway, if you *need* to use such a limited editor, you can
    simply use some free and open source software like imagemagick
    or graphicsmagick to simply convert the old JPEG to some file
    format that can be read by your limited editor.

    -Wolfgang

    [1] Though the newest version of the Linux kernel threw out
    the 80386 CPU (though not it's descendants).

    Considering that the 386 is a CPU from 1985 (28 years ago)
    and hasn't even been produced for the last 6 years (and had
    been produced only as embedded CPU for a considerable time)
    and could at best do 40 MHz ...
    and you can *still* use maintained 2.6, 3.0, 3.2 and 3.4
    kernels (current is 3.8) ...
    you'll be set for about the life time of your 386, even
    though probably noone is using any anymore where Linux
    support would matter.
    486SX --- the ones without an FPU --- are apparently
    still in use somewhere and therefore the FPU emulation
    code stays in the current kernel.

    That should give you an idea how 'old JPEG support' will
    be handled.
     
    Wolfgang Weisselberg, Feb 23, 2013
    #36
  17. The scanner can record every 'bit' of information at exactly
    the same rate as the print (assuming the information is
    stored in a grid pattern).
    You only have to make 100% sure that you sample every sample
    *exactly* at the place where the print stored the information.

    (If you're just a tiny bit off, you've got lots of problems.)

    -Wolfgang
     
    Wolfgang Weisselberg, Feb 23, 2013
    #37
  18. (PeteCresswell)

    John Turco Guest

    <heavily edited for brevity>

    Is the "scanner" in question a person, or a device? (Your use
    of "his" is confusing.)

    John
     
    John Turco, Feb 24, 2013
    #38
  19. It doesn't matter if it's a device, a person, an AI,
    something where a poltergeist lives in's a device, a person,
    an AI, some thing wherein a poltergeist lives, ...

    That's my story, and I'll stick to it!

    -Wolfgang
     
    Wolfgang Weisselberg, Feb 25, 2013
    #39
  20. Not in practice. It can only do that if it can
    precisely time the samples to the grid pattern, and of
    course it cannot reasonably do that.
    You cannot do that in any practical way, short of the
    simple and easy way... sample at twice the spatial rate
    of the data.
    So why are you arguing the point that was accurately stated to
    begin with?
     
    Floyd L. Davidson, Feb 25, 2013
    #40
    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.