resampling a digital image to make it bigger?

Discussion in 'Digital Photography' started by Mr.Will, Oct 22, 2003.

  1. Mr.Will

    Mr.Will Guest

    Recently I was asked to print several of my shots at a 10 x 8 size, and to
    crop quite heavily (about a quarter of the original frame). I took this
    image with a Canon d60, and was concerned about noise and pixellation etc.
    The day after that I saw the same photograph on the cover of a UK paper
    blown up even bigger - and without pixellation or any noise! I was surprised
    to say the least.

    I had always thought that all images should be printed at 300dpi - so if one
    wanted a bigger image than the frame, a loss in DPI would be neccessary. Not
    so according to the people at the press, they merely resampled it in
    photoshop.

    I tried this instantly with the picture I had been asked for, and indeed it
    worked out very nicely at 10 x 8. I simply resampled and kept the resolution
    at 300dpi. What I dont understand is, what has the computer and photoshop
    done to actually create more "pixels" for want of a better term?

    --
    Mr.Will
     
    Mr.Will, Oct 22, 2003
    #1
    1. Advertising

  2. Mr.Will

    Bob Hatch Guest

    "Mr.Will" <> wrote in message
    news:fkmlb.1354$...

    >
    > I tried this instantly with the picture I had been asked for, and indeed

    it
    > worked out very nicely at 10 x 8. I simply resampled and kept the

    resolution
    > at 300dpi. What I dont understand is, what has the computer and photoshop
    > done to actually create more "pixels" for want of a better term?
    >
    > --

    Yes, and you can print at 250 PPI and you'll never see the difference
    between that and 300. Our pro lab wants all images at size at 250 PPI. I
    print at 250 PPI on my Epson 7600.

    You know by now that your computer did not blow up and by using these extra
    pixels you have not created a warp in the fabric of the universe. :)

    --
    "Your money does not cause my poverty. Refusal to believe
    this is at the bottom of most bad economic thinking." --P. J. O'Rourke
    http://www.bobhatch.com
     
    Bob Hatch, Oct 22, 2003
    #2
    1. Advertising

  3. Mr.Will

    MikeWhy Guest

    "Mr.Will" <> wrote in message
    news:fkmlb.1354$...
    > I tried this instantly with the picture I had been asked for, and indeed

    it
    > worked out very nicely at 10 x 8. I simply resampled and kept the

    resolution
    > at 300dpi. What I dont understand is, what has the computer and photoshop
    > done to actually create more "pixels" for want of a better term?


    I have heard, but have not actually confirmed, that enlarging by repeated
    steps of 110% each time is better than a single large blow-up. FWIW. Maybe
    it's something to do with Photoshop's bicubic interpolation, but I don't
    know any details.
     
    MikeWhy, Oct 22, 2003
    #3
  4. "Mr.Will" wrote:

    > Recently I was asked to print several of my shots at a 10 x 8 size, and to
    > crop quite heavily (about a quarter of the original frame). I took this
    > image with a Canon d60, and was concerned about noise and pixellation etc.
    > The day after that I saw the same photograph on the cover of a UK paper
    > blown up even bigger - and without pixellation or any noise! I was surprised
    > to say the least.
    >
    > I had always thought that all images should be printed at 300dpi - so if one
    > wanted a bigger image than the frame, a loss in DPI would be neccessary. Not
    > so according to the people at the press, they merely resampled it in
    > photoshop.
    >
    > I tried this instantly with the picture I had been asked for, and indeed it
    > worked out very nicely at 10 x 8. I simply resampled and kept the resolution
    > at 300dpi. What I dont understand is, what has the computer and photoshop
    > done to actually create more "pixels" for want of a better term?
    >
    > --
    > Mr.Will


    To create new pixels, PS invents them. But it does so in an intelligent manner.
    Its bicubic resampling algorithm looks at every pixel as well its nearest
    neighbor and their nearest neighbors......etc.
    It then creates new pixels that that blend best with all the other pixels in
    the neighborhood.
    This helps to make the images look smooth rather than pixelated at high
    enlargements.

    The 300 pixels/inch (ppi) criterion is for making prints indistinguishable from
    traditional wet chemical prints.
    High quality Art Magazines may also request 300 dpi ( really ppi).
    I usually try to print at 240 uninterpolated ppi. If I don't have enough native
    pixels in my image to print at 240 ppi, I'll have PS invent them. At other
    times I'll just print at 200 ppi and get very good results
    Newspapers are such low resolution media, that they can get by with 100 ppi, or
    less.
    Bob Williams
     
    Robert E. Williams, Oct 22, 2003
    #4
  5. On Wed, 22 Oct 2003, MikeWhy wrote:

    > "Mr.Will" <> wrote in message
    > news:fkmlb.1354$...
    > > I tried this instantly with the picture I had been asked for, and indeed

    > it
    > > worked out very nicely at 10 x 8. I simply resampled and kept the

    > resolution
    > > at 300dpi. What I dont understand is, what has the computer and photoshop
    > > done to actually create more "pixels" for want of a better term?

    >
    > I have heard, but have not actually confirmed, that enlarging by repeated
    > steps of 110% each time is better than a single large blow-up. FWIW. Maybe
    > it's something to do with Photoshop's bicubic interpolation, but I don't
    > know any details.
    >
    >


    To me it would seem that doing it many small step would be bad for exactly
    the same reason photocopying photocopies of photcopies is bad.

    Things don't get better as you interpolate on top of interpolation.
     
    Fraser Wright, Oct 22, 2003
    #5
  6. Mr.Will

    Xiaoding Guest

    >
    > The 300 pixels/inch (ppi) criterion is for making prints indistinguishable from
    > traditional wet chemical prints."



    GASP! Surley you jest! INDISTINGUISHABLE? I think not! 300 crappy
    pixels per inch does not make a Cibachrome print, sir. Please!!
     
    Xiaoding, Oct 22, 2003
    #6
  7. Mr.Will

    HRosita Guest

    Hi,

    There are additional algorithms for resampling. I use Genuine Fractals )a
    Photoshop plugin).
    Irfan view uses Lanczos.

    Whatever you can always experiment and do your own evaluation at which one is
    best.
    Rosita
     
    HRosita, Oct 22, 2003
    #7
  8. "Fraser Wright" <> wrote in message
    news:p...
    SNIP
    > To me it would seem that doing it many small step would be bad
    > for exactly the same reason photocopying photocopies of
    > photcopies is bad.


    That would intuitively seem what to expect, degradation with each next
    generation. However, it's different with Bi-Cubic interpolation.

    > Things don't get better as you interpolate on top of interpolation.


    In a way they don't. However, when you interpolate/resize to a larger
    dimension, Photoshop can add a little edge contrast at sharp brightness
    transitions, depending on the surrounding pixels. Typically, edges (which
    are most important for our impression of sharpness) will thus gain some
    contrast. When repeated, this results in (visually) better (but slightly
    exaggerated) edge contrast and less pixellation with extreme enlargements.
    It offers a benefit over one step resampling which averages the edge
    transitions but exaggerates individual pixel edges.

    Bart
     
    Bart van der Wolf, Oct 22, 2003
    #8
  9. On Wed, 22 Oct 2003 05:48:04 +0000, MikeWhy wrote:

    > "Mr.Will" <> wrote in message
    > news:fkmlb.1354$...
    >> I tried this instantly with the picture I had been asked for, and indeed

    > it
    >> worked out very nicely at 10 x 8. I simply resampled and kept the

    > resolution
    >> at 300dpi. What I dont understand is, what has the computer and
    >> photoshop done to actually create more "pixels" for want of a better
    >> term?

    >
    > I have heard, but have not actually confirmed, that enlarging by repeated
    > steps of 110% each time is better than a single large blow-up. FWIW. Maybe
    > it's something to do with Photoshop's bicubic interpolation, but I don't
    > know any details.


    AFAIK, Lanczos interpolation is considered to be the best available
    interpolation method by those who are supposed to know about such things.
    Photoshop doesn't do Lanczos, but ImageMagick (which is free) does. I'd be
    interested to see a comparison of the two sometime, and to know whether
    the difference in quality between the algorithms is visible.

    Mike.
     
    Mike Brodbelt, Oct 22, 2003
    #9
  10. "Mr.Will" <> wrote in message
    news:fkmlb.1354$...
    > Recently I was asked to print several of my shots at a 10 x 8 size, and to
    > crop quite heavily (about a quarter of the original frame). I took this
    > image with a Canon d60, and was concerned about noise and pixellation etc.
    > The day after that I saw the same photograph on the cover of a UK paper
    > blown up even bigger - and without pixellation or any noise! I was

    surprised
    > to say the least.
    >
    > I had always thought that all images should be printed at 300dpi - so if

    one
    > wanted a bigger image than the frame, a loss in DPI would be neccessary.

    Not
    > so according to the people at the press, they merely resampled it in
    > photoshop.
    >
    > I tried this instantly with the picture I had been asked for, and indeed

    it
    > worked out very nicely at 10 x 8. I simply resampled and kept the

    resolution
    > at 300dpi. What I dont understand is, what has the computer and photoshop
    > done to actually create more "pixels" for want of a better term?
    >
    > --
    > Mr.Will
    >
    >

    Make a Photoshop Action to do it for you!


    Open an image you want res up.

    Open the Actions palette and select "New Set".

    Name it "Creeping Increase" or whatever you like.

    Now select New Action and call it "1.5 Increase"

    Next go to Image/Size and put a check in "Resample" and "Constrain
    Proportions"

    with Bicubic and change the width drop down to "percent" and enter 110.

    Now go to "Image Size" do the same 3 more times.

    Then "Stop Recording"



    Next "New Action" Name it "2.0 Increase"

    Now "Edit/Purge All"

    Next select the "1.5 Increase" in the Actions Palette

    and press the Play button.

    select the "1.5 Increase" again,

    press the Play button again.

    Stop Recording.



    "New Action" Name it 3.0 Increase.

    "Edit/Purge All"

    Select and play the "2.0 Increase"

    Select and play the "1.5 Increase"

    Stop Recording.



    "New Action" Name it 4.0 Increase.

    "Edit/Purge All"

    Select and play the "3.0 Increase"

    Now "Image Size" 110 percent.

    And again "Image Size" 110 percent.

    Stop Recording.



    Save the Action.
     
    /\\BratMan/\\, Oct 22, 2003
    #10
  11. Mr.Will

    Bill Hilton Guest

    >From: Mike Brodbelt m.brodbelt@__NO_SPAM_PLEASE__coruscant.demon.co.uk

    >AFAIK, Lanczos interpolation is considered to be the best available
    >interpolation method by those who are supposed to know about such things.
    >Photoshop doesn't do Lanczos, but ImageMagick (which is free) does. I'd be
    >interested to see a comparison of the two sometime, and to know whether
    >the difference in quality between the algorithms is visible.


    Sure you can see a difference ... if you couldn't then there's no reason to
    have more than one way to do it.

    Here's a comparison of Stair Interpolation (free if you have Photoshop),
    Genuine Fractals, Lanczos, S-Spline and plain old Photoshop Bicubic. YMMV
    depending on the data structure of the image and the amount of enlargement you
    end up doing (at some point GF is supposed to shine, for example).

    http://www.fredmiranda.com/SI/index.html

    I have Fractals (bundled for free with my scanner) and have tested in several
    times but for the type images I shoot (film scans) and the amount of resizing I
    rarely have to do I got better results with Stair Interpolation.

    Photoshop CS supposedly has two new resampling options.

    Bill
     
    Bill Hilton, Oct 22, 2003
    #11
  12. Stanley Krute, Oct 22, 2003
    #12
  13. Hi Bill

    > Stair Interpolation (free if you have Photoshop)


    Do you have a link to a free download of Stair Interpolation ?

    Thanks.

    Stan
     
    Stanley Krute, Oct 22, 2003
    #13
  14. Mr.Will

    Bill Hilton Guest

    >> Bill Hilton wrote ...
    >>
    >> Stair Interpolation (free if you have Photoshop)



    >From: "Stanley Krute"
    >
    >Hi Bill
    >
    >Do you have a link to a free download of Stair Interpolation ?


    It's "free" if you write your own Photoshop action, but if you don't want to do
    this (takes about 30 seconds) then you'll have to buy one of Fred Miranda's SI
    actions.

    To make your own SI action that steps up say 3 times just get a small file (to
    make recording go faster), open up the Actions palette and set to record (name
    it "SI 3x" or whatever), then Image > Image Size > (check resample & constrain,
    change units to percent and change 100 to 110 and click OK), and repeat 3 times
    and stop recording. Then when you're ready to upsample just run this action as
    often as required, perhaps running a dose of USM (like Fred does) when/if the
    file needs it.

    If you have a specific size image that you want to resize to a specific value
    (say a 3 Mpix digital camera file that you want to resize to print to a
    specific size, like 300 ppi @ 12 x 18") you can make a SI action specifically
    for that file size to speed things up.

    Bill
     
    Bill Hilton, Oct 22, 2003
    #14
  15. Mr.Will

    Tore Lund Guest

    Resampling a digital image to make it smaller? (was: resampling adigital image to make it bigger?)

    While we're at it, does anyone have strong opinions on what is the best
    program or plug-in for downsampling? Bicubic seems fine, and some other
    algorithms are close contenders. But I wonder if there is a clear
    winner somewhere. And can step-wise sampling help if you make images
    smaller?
    --
    Tore
     
    Tore Lund, Oct 22, 2003
    #15
  16. Mr.Will

    Guest

    In message <oSolb.12342$>,
    "MikeWhy" <> wrote:

    >"Mr.Will" <> wrote in message
    >news:fkmlb.1354$...
    >> I tried this instantly with the picture I had been asked for, and indeed

    >it
    >> worked out very nicely at 10 x 8. I simply resampled and kept the

    >resolution
    >> at 300dpi. What I dont understand is, what has the computer and photoshop
    >> done to actually create more "pixels" for want of a better term?

    >
    >I have heard, but have not actually confirmed, that enlarging by repeated
    >steps of 110% each time is better than a single large blow-up. FWIW. Maybe
    >it's something to do with Photoshop's bicubic interpolation, but I don't
    >know any details.


    Maybe the math used in Photoshop loses some dynamic range with radical
    size changes. I know the math is non-optimal in other operations; 360
    one-degree rotations will not get you back where you started, for
    instance (you'd expect the image to be blurry, but not rotated at the
    end).
    --

    <>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
    John P Sheehy <>
    ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><
     
    , Oct 22, 2003
    #16
  17. "Stanley Krute" <> wrote in message
    news:2jzlb.1056$...
    > Hi Bill
    >
    > > Stair Interpolation (free if you have Photoshop)

    >
    > Do you have a link to a free download of Stair Interpolation ?


    This one works for me:
    http://www.imphotography.com/downloads/ssimagesize.htm

    Bart
     
    Bart van der Wolf, Oct 22, 2003
    #17
  18. Mr.Will

    MikeWhy Guest

    "Fraser Wright" <> wrote in message
    news:p...
    > On Wed, 22 Oct 2003, MikeWhy wrote:
    >
    > > "Mr.Will" <> wrote in message
    > > news:fkmlb.1354$...
    > > > I tried this instantly with the picture I had been asked for, and

    indeed
    > > it
    > > > worked out very nicely at 10 x 8. I simply resampled and kept the

    > > resolution
    > > > at 300dpi. What I dont understand is, what has the computer and

    photoshop
    > > > done to actually create more "pixels" for want of a better term?

    > >
    > > I have heard, but have not actually confirmed, that enlarging by

    repeated
    > > steps of 110% each time is better than a single large blow-up. FWIW.

    Maybe
    > > it's something to do with Photoshop's bicubic interpolation, but I don't
    > > know any details.
    > >
    > >

    >
    > To me it would seem that doing it many small step would be bad for exactly
    > the same reason photocopying photocopies of photcopies is bad.
    >
    > Things don't get better as you interpolate on top of interpolation.


    And the only reason I bring it up is *because* it seems counter-intuitive. I
    haven't actually tried it. The source is one of the recent glossy paper Ps
    books.
     
    MikeWhy, Oct 23, 2003
    #18
  19. Xiaoding wrote:

    > >
    > > The 300 pixels/inch (ppi) criterion is for making prints indistinguishable from
    > > traditional wet chemical prints."

    >
    > GASP! Surley you jest! INDISTINGUISHABLE? I think not! 300 crappy
    > pixels per inch does not make a Cibachrome print, sir. Please!!


    I don't really consider Cibachrome (now called Ilfochrome) prints "traditional".
    You have to really look around for someone who will process Cibachrome, nowadays.
    I don't even see many Cibachromes showing up in competition anymore.
    I used to be impressed by Cibachrome 25 years ago, but now its super shiny,
    in-your-face, plastic look is a bit tiring.
    Of course, that's only my opinion. However, if you like that look, a 300ppi dye
    sublimation print on Hi-Gloss paper will fool any but the most critical viewer.
    See what this Ex-Cibachromer has to say about Cibachrome vs Fuji Crystal Archive
    (traditional silver halide prints).
    http://www.erwinvoogt.com/order.html
    Also, see what this Cibachrome Guru says about Cibachrome vs Inkjet. See paragraph
    entitled, "Inkjet vs Light Jet."
    http://luminous-landscape.com/tutorials/handmade.shtml
    Bob Williams
     
    Robert E. Williams, Oct 23, 2003
    #19
  20. Mr.Will

    Bill Hilton Guest

    > To me it would seem that doing it many small step would be bad for exactly
    > the same reason photocopying photocopies of photcopies is bad.
    >
    > Things don't get better as you interpolate on top of interpolation.


    It's simple to just interpolate both ways and look at the results and *see*
    what looks better.

    When you do this (if you do this) you'll see that resampling up in small steps
    DOES give better results. See the link I gave earlier in this thread for Stair
    Interpolation to see the difference.

    Bill
     
    Bill Hilton, Oct 23, 2003
    #20
    1. Advertising

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

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Marty

    Effects of Resampling and Brightness Adjust

    Marty, Sep 7, 2003, in forum: Digital Photography
    Replies:
    1
    Views:
    321
  2. Desmond
    Replies:
    5
    Views:
    918
    Bob D.
    Sep 27, 2003
  3. Replies:
    10
    Views:
    1,613
  4. Lee
    Replies:
    35
    Views:
    1,096
    John McWilliams
    Jun 12, 2006
  5. 'Magic' kernel for image zoom resampling?

    , Jun 6, 2006, in forum: Digital Photography
    Replies:
    5
    Views:
    1,593
    Jani Huhtanen
    Jun 7, 2006
Loading...

Share This Page