Enlarging digital images - how big is big?

Discussion in 'Digital SLR' started by Doug MacDonald, Jan 28, 2007.

  1. With resolution and clarity equal to the original: 6.3 Megapixel image will
    usually go to 24" x 32" (36" APS size) and way, way larger if you care to
    introduce a "viewing distance equal to the distance needed to view all the
    image without moving your head.

    For a couple of years now I've endured taunts, ridicule and personal
    vilification from a few narrow minded people who never bothered to
    investigate my claims... How could most of them? The basis for me providing
    samples and confirmation for investigation was QUALIFICATIONS TO PASS
    JUDGMENT. All but one, never had any. Some really nice people in this lot, I
    can tell you.

    So... I have in the past made statements like:
    "Enlarge from a postcard size image to a poster"
    "Up to 1000% enlargement"
    "Add data to an image to enlarge it"

    All perfectly true and ridiculed by zealots looking for recognition they
    have no right seeking in the first place.

    I have been able to, and founded a business on, enlarging digital images
    using Interpolation. Until now I have chosen to protect my industrial
    secrets and formulas by refusing to provide enlarged digital images or pixel
    level crops of enlarged images to anyone for examination... Until now.

    Recently I began to post to the 'groups' again after being away from them
    due to the pressures of work only to discover the idiots and morons who last
    year were so vocal in their condemnation of me, are still out there with the
    worst of them. Offers of a reward for the identity of the worst, turned up
    zilch and those I did discover had nothing worth suing for anyway.

    So maybe today the disbelievers and pundits will be satisfied, maybe not. I
    have provided an example image which I enlarged and a full pixel crop of
    part of the ENLARGED image to demonstrate that not only is it possible but
    possible at virtually zero loss of sharpness and detail in the final image.

    http://canvas.photosbydouglas.com/interpolation.htm

    Use the link at the bottom of the text on that page to take you to the
    actual examples and explanation. Anyone who wants to have the original
    (un-enlarged) image only needs to ask me without aggression or taunts and
    I'll supply it as an attachment to an email only to a valid domain address
    (no hotmail, Yahoo or Gmail).
     
    Doug MacDonald, Jan 28, 2007
    #1
    1. Advertisements

  2. A 'few' points.. (Long, but strictly ontopic and with relatively few
    "insults, taunts or aggression" in what follows from me..)

    The original image will show 'real' detail down to about the two pixel
    level. But when interpolating, any enlargement method can at best
    make a 'sensible guess' as to the extra pixels required to fill in the
    gaps. Many existing algorithms do this quite well, using a
    combination of bitmap and vector theorems.

    The interpolated pixels are *not* real detail, therefore, at actual
    pixel size the resulting image *cannot* have "resolution and clarity
    equal to the original".

    Finely detailed images enlarged to these sizes will show the results
    of interpolation on close inspection, especially on things like grass,
    fine foliage, fine details in a cityscape, etc. Indeed, Douglas'
    'actual pixel crop' seems to have problems, even though this is not a
    particularly challenging image.

    It is perfectly true that on some types of images you *will* get away
    with enlargements of this size and greater - but it largely depends on
    image content, and skill at interpolating. Judging from the Douglas
    'actual detail' crop, I would suggest you look closely at QImage
    instead - it introduces less jaggies and artefacts. Some of this
    could be 'real detail', but we don't know without the original.
    If you have to be moved back some unspecified distance, then you can
    enlarge anything to any size.. So let's be scientific here - without
    moving my head but allowing my eyes to wander to their full extent, I
    have a sharp field of view that spans well over 90 degrees. A quick
    check shows that I can easily view all of a 32" x 24" image without
    moving my head, from about 14".
    Indeed, how could we, other than to point out flaws in what was
    displayed? (all these pages have now been removed by Douglas). He
    never posted an original image and when he posted crops, they were
    *always* reductions or at best actual-pixel size - they were not
    enlargements. This is the *first time* he has actually shown what
    looks to be a real enlargement... But we are still missing the
    original file - if we had that, the artefacts (or to be fair, the lack
    of them) that his method has introduced would be clear.
    I suspect the 'one' who Douglas thought was qualified may have been
    Gordon Moat. Gordon had to clarify the claims made by Douglas, here:

    http://groups.google.com/group/rec.photo.equipment.35mm/browse_frm/
    thread/f0bc2af22e84217b/

    Make your own mind up.
    Of course they are all true. But data DOES NOT equal detail. It's
    the *quality* claims that were in dispute. Eg:
    "Enlarged digital images with more detail than the original"
    http://groups.google.com/group/comp.graphics.apps.photoshop/browse_frm/
    thread/90dff1d6c6425889
    ...using an algorithm Douglas claimed was patented (it wasn't) and had
    been bought by Samsung.. Having now seen the results, it looks to be
    somewhat inferior to the many good algorithms already in existence,
    for example QImage's excellent Vector, Pyramid and Hybrid methods.
    Can Douglas offer his comments on how those compare to his, or show us
    the final result comparison?
    Bold move! But we still await the original file. Douglas is still
    short of the proof and genuine examples offered by all of his
    'competitors'.
    I wonder who he 'discovered'? No-one turned up at my door.
    Personally, I would have thought Douglas could better spend his time
    simply *proving* detractors wrong (or ignoring them), than undertaking
    all these investigations..
    Nope. But Happy Birthday anyway!
    Note the "virtually zero" - that's a drop from "equal" or "more detail
    than the original".

    By the way, if you visit Douglas' page, use the position of the spider
    web - you can not only see it at top right of the crop but also just a
    trace of it at bottom left - to get an idea of how accurate his
    'Approximation' is. These errors have always been in his favour -
    must just be coincidence. But having said that and to give credit
    where due, it's not actually important this time as the crop appears
    to be a true enlargement! In fact he has gone way too far.. But
    judge for yourself the quality - note the jaggies, haloing, and other
    artefacts. At least they *look* like artefacts. Without the original
    actual-pixels crop we can't tell, but it looks like 'mush' (â„¢David)to
    me.

    If Doug were to post the *original* actual pixels crop, it would tell
    us much more, and of course we could run it through GF, QIMage, even
    PS BiCubics, SSI, maybe even some of the more exotic methods shown
    here:

    http://www.general-cathexis.com/interpolation.html

    ... and then make a *useful* comparison. The argument would be over,
    the truth of the final quality would be clear.
    If anyone bothers, let us know how that crop looks when enlarged using
    *other* methods, especially QImage - I think there's a free 30 day
    trial?

    Anyway, I'm sure he does *lovely* prints - if I could only find one of
    his 'TechnoAussie Digital Print Centres', I'd have one done myself!
     
    mark.thomas.7, Jan 28, 2007
    #2
    1. Advertisements

  3. If this is an advertisement (that is what it looks like), or a hint for
    other to try,IMHO, it should be denoted in the subject and should not be
    disguised as a question.

    -Michael
     
    Michael Schnell, Jan 28, 2007
    #3
  4. Doug MacDonald

    Rudy Benner Guest

    It would be nice if the links to your Gallery actually worked.
     
    Rudy Benner, Jan 28, 2007
    #4
  5. Doug,
    where are the pictures...on my browser I see just the background...??!!
    rgds Georg Nyman
     
    Dr. Georg N.Nyman, Jan 28, 2007
    #5
  6. : Doug,
    : where are the pictures...on my browser I see just the background...??!!
    : rgds Georg Nyman
    :
    :
    I guess they must be too big for your monitor?
    Sick joke, I know.
    I have no idea actually. That site is hosted by godaddy.com and is (usually)
    totally reliable. Where some of these things occur is when your ISP's
    transparent proxy server is holding a half loaded page and feeds that up to
    you as the full thing. Sometimes too, your browser can be set for no refresh
    to save time when viewing regular pages.

    Heck Georg... You don't even say what brand your browser is. Try again and
    hit the refresh if it happens again.
     
    Doug MacDonald, Jan 29, 2007
    #6
  7. : Doug,
    : where are the pictures...on my browser I see just the background...??!!
    : rgds Georg Nyman
    :
    :
    Georg... The issue is still unknown to me but I put the article on a
    different server in another part of the USA which works with everything I
    tried. Have a look:
    http://www.brisbaneweddingphotographers.com/canvas/html/interpolation.htm

    I would appreciate you letting me know if you have a problem viewing this
    link.
    Douglas
     
    Doug MacDonald, Jan 29, 2007
    #7
  8. message SNIP
    Hi Doug,

    First things first, happy birthday. Hope your back is holding up after
    your earlier surgery (and being a kind of tallish guy myself, I know
    what I'm talking about).

    Secondly, too bad you are just offering a JPEG version of the original
    capture (unless your interpolation was also based on that). But to be
    frank, assumptions about fundamental image quality aren't very
    convincing in an objective test, they will only (re)enforce the
    position of critics ...

    As you suggest in your web-page, and I support that mentality, well
    founded peer review *will* further progress. After all, that's what
    regular scientific publications are all about.

    I wouldn't mind providing my interpolated version of the crop area you
    showed, again assuming you started from the same JPEG (and again,
    assumptions are not the most credible starting points).

    Print files also have different sharpening requirements depending on
    the print process used, so a comparison to a print you supply/offer
    may (or not) be very relevant. How about just settling for a web
    version comparison to the one you published on your "example-1.htm"
    page?

    If you are confident enough, just respond to my e-mail. If not, well
    we'll never know ...

    Bart
     
    Bart van der Wolf, Jan 30, 2007
    #8
  9. Unfortunately, my email bounced from your supplied e-mail address :-(
    Without a means to supply you with my (e-mail) address, we'll not get
    very far, I'm afraid.

    Bart
     
    Bart van der Wolf, Jan 30, 2007
    #9
  10. Doug MacDonald

    Pete D Guest

    OMG, OMG, OMG his always changing email address did not work. Makes it
    bloody hard to killfile the bastard.
     
    Pete D, Jan 30, 2007
    #10
  11. :
    : message : SNIP
    : > So maybe today the disbelievers and pundits will be satisfied, maybe
    : > not.
    : > I have provided an example image which I enlarged and a full pixel
    : > crop
    : > of part of the ENLARGED image to demonstrate that not only is it
    : > possible
    : > but possible at virtually zero loss of sharpness and detail in the
    : > final image.
    : >
    : > http://canvas.photosbydouglas.com/interpolation.htm
    : >
    : > Use the link at the bottom of the text on that page to take you to
    : > the
    : > actual examples and explanation. Anyone who wants to have the
    : > original
    : > (un-enlarged) image only needs to ask me without aggression or
    : > taunts
    : > and I'll supply it as an attachment to an email only to a valid
    : > domain
    : > address (no hotmail, Yahoo or Gmail).
    :
    : Hi Doug,
    :
    : First things first, happy birthday. Hope your back is holding up after
    : your earlier surgery (and being a kind of tallish guy myself, I know
    : what I'm talking about).
    :
    : Secondly, too bad you are just offering a JPEG version of the original
    : capture (unless your interpolation was also based on that). But to be
    : frank, assumptions about fundamental image quality aren't very
    : convincing in an objective test, they will only (re)enforce the
    : position of critics ...
    :
    : As you suggest in your web-page, and I support that mentality, well
    : founded peer review *will* further progress. After all, that's what
    : regular scientific publications are all about.
    :
    : I wouldn't mind providing my interpolated version of the crop area you
    : showed, again assuming you started from the same JPEG (and again,
    : assumptions are not the most credible starting points).
    :
    : Print files also have different sharpening requirements depending on
    : the print process used, so a comparison to a print you supply/offer
    : may (or not) be very relevant. How about just settling for a web
    : version comparison to the one you published on your "example-1.htm"
    : page?
    :
    : If you are confident enough, just respond to my e-mail. If not, well
    : we'll never know ...
    :
    : Bart
    :
    Your comments are all valid Bart. The original file was actually a JPG which
    when enlarged, I saved as another JPG. I use the so called "lossless" JPEG
    format of no compression for these things. I know some people are concerned
    with artifacts growing from them but I seriously haven't found any in
    extremely large prints although there are some in the files.

    I concern myself with the final product. Others (maybe you too) worry at a
    pixel level about artifacts that don't reproduce in printing. Keeping in
    mind the available printers are inkjet based and although considered "photo
    quality" I notice recently a new (higher res?) description of Photo LAB
    quality. In any event, the Epson printer drivers all feed 720 dpi stuff to
    their 4800dpi printers! Presumably this is because the real specs are 4800
    dpi x 720 or 1440dpi. It's not hard to print at 4800 dpi in a straight line
    across the sheet.

    You may already know that Olympus cameras make very good quality hi-res jpeg
    files and RAW files of huge size. Offering a JPEG file is actually offering
    a copy of the original (unedited) camera file in this instance. Reply to the
    address attached to this message (hit the reply to sender button) and I'll
    get the message. I wasn't aware of the problem with the email on the web
    site. I'll fix it in the morning.

    Thanks for the well wishing... Yes, the surgery eventually cost me the price
    of a new BMW (two sessions) but it was worth every penny. I suppose I'll get
    used to the Toyota someday!!

    Douglas
     
    Doug MacDonald, Jan 30, 2007
    #11
  12. :
    : : >
    : > : >>
    : >> : >> SNIP
    : >>> So maybe today the disbelievers and pundits will be satisfied, maybe
    : >>> not. I have provided an example image which I enlarged and a full
    pixel
    : >>> crop of part of the ENLARGED image to demonstrate that not only is it
    : >>> possible but possible at virtually zero loss of sharpness and detail
    in
    : >>> the final image.
    : >
    : > Unfortunately, my email bounced from your supplied e-mail address :-(
    :
    : OMG, OMG, OMG his always changing email address did not work. Makes it
    : bloody hard to killfile the bastard.
    :
    : >
    : >> If you are confident enough, just respond to my e-mail. If not, well
    : >> we'll never know ...
    : >
    : > Without a means to supply you with my (e-mail) address, we'll not get
    very
    : > far, I'm afraid.
    : >
    : > Bart
    :
    :
    Hey Pete... School's out. Time to go home for your nappy change.
     
    Doug MacDonald, Jan 30, 2007
    #12
  13. :
    : : >
    : > message : > SNIP
    : >> So maybe today the disbelievers and pundits will be satisfied,
    : >> maybe not. I have provided an example image which I enlarged and a
    : >> full pixel crop of part of the ENLARGED image to demonstrate that
    : >> not only is it possible but possible at virtually zero loss of
    : >> sharpness and detail in the final image.
    :
    : Unfortunately, my email bounced from your supplied e-mail address :-(
    :
    : > If you are confident enough, just respond to my e-mail. If not, well
    : > we'll never know ...
    :
    : Without a means to supply you with my (e-mail) address, we'll not get
    : very far, I'm afraid.
    :
    : Bart
    :
    Hi Bart, as promised yesterday...
    http://www.photosbydouglas.com/canvas/canvas-contact.htm
    and http://www.canvasphotos.com.au/canvas-contact.htm
    Provides a contact form for requests. Stay anonymous and I won't know who to
    send what to!

    Douglas
     
    Doug MacDonald, Jan 30, 2007
    #13
  14. Doug MacDonald

    Pete D Guest

    Thanks I will take that as a compliment. :)

    But seriously Doug how could anyone that changes their email address and
    name so often be taken seriously, if that is your name of course.

    And why do you give a shit what anyone on here actually thinks anyway?
     
    Pete D, Jan 31, 2007
    #14
    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.