The effect of pixel increases on images

Discussion in 'Digital Photography' started by Rich, Mar 5, 2006.

  1. Rich

    Rich Guest

    I constructed a small gallery to illustrate the relative differences in
    image detail at
    different megapixel counts. Comparing the actual cameras is the best
    way to do it,
    since you will likely buy based on that alone, but the problem with it
    is that the cameras, their internal software and the sensors are all
    different. There are no two (unfortunately) cameras
    that use the same of everything, except for pixel counts on sensors.
    That would be an ideal
    situation because it would mean you could have 4 meg P&S cameras with
    the noise levels of
    8 meg DSLRs, simply by cutting down the sensor to a 4 meg pixel count.
    But this has
    not be done.

    What I did was to take images at distances that corresponded to varying
    pixel counts.
    So (for eg) the image area under the lens for the 3.7Meg equivalent
    occupied approx. 1/4
    of the size of the 16Meg shot. The lens was set at its highest
    resolution
    point, f5.6. Only the centre of the image was used for the 100% crops.
    Lastly, I used "Smart Interpolation" to upsize the lesser megapixel
    count shots until they matched the 16 meg shot size. Very little image
    artifacting was noticed, even with relatively
    large (3.7meg to 16 meg size) upsizings so the specific details of each
    shot are preserved.
    The images were taken in RAW, sharpened equally, converted to 16 bit
    TIFF cropped and "Smart interpolated" and finally saved as 100% quality
    JPEGs.
    I used a halogen light source to illuminate the object, tripod mounted
    the camera
    and used a remote shutter release.

    The reason I used close-up shots (2 feet) was to avoid the uncontrolled
    aspects of long
    distance (landscape) shots, such as exposure variation, etc.

    Anyway, the result seems to be that the detail in the lesser megapixel
    images is reduced
    noticeably and this has resulted in a sharpness degradation as well.
    Sharpness and resolution (detail) do not always go hand in hand, I've
    seen unsharp shots that had more
    detail than some sharp shots, but at a maximum resolution onscreen of
    72dpi or so,
    the differences are (IMO) very noticeable. At print resolutions of
    300dpi, it would be even
    more drastic. Now its clear why someone would be willing to pay $7000
    or so for that
    16 meg Canon!
    Here is the gallery:

    http://www.pbase.com/andersonrm/uprezing
     
    Rich, Mar 5, 2006
    #1
    1. Advertising

  2. Rich

    Broake Guest

    "Rich" <> wrote in message
    news:...
    Now its clear why someone would be willing to pay $7000
    > or so for that
    > 16 meg Canon!
    > Here is the gallery:
    >
    > http://www.pbase.com/andersonrm/uprezing
    >


    Interesting - to my inexpert eyes the best trade-off in terms of
    price/performance comes at 10mp. The quality increase after that seems
    entirely disproportionate to the costs involved.

    Having said that, I once had a 4mp Panasonic Lumix p&s that produced shots
    greatly in excess of the quality you show for the 3.7mp image - so perhaps
    this test is somewhat flawed?
     
    Broake, Mar 5, 2006
    #2
    1. Advertising

  3. Rich

    JohnR66 Guest

    "Broake" <> wrote in message
    news:...
    >
    > "Rich" <> wrote in message
    > news:...
    > Now its clear why someone would be willing to pay $7000
    >> or so for that
    >> 16 meg Canon!
    >> Here is the gallery:
    >>
    >> http://www.pbase.com/andersonrm/uprezing
    >>

    >
    > Interesting - to my inexpert eyes the best trade-off in terms of
    > price/performance comes at 10mp. The quality increase after that seems
    > entirely disproportionate to the costs involved.
    >
    > Having said that, I once had a 4mp Panasonic Lumix p&s that produced shots
    > greatly in excess of the quality you show for the 3.7mp image - so perhaps
    > this test is somewhat flawed?
    >

    You missed his point. The 3.7MP was upsampled to match the size of the 16mp.
    Of course it is going to be soft. There would be no need for 16mp if 4mp
    looked as good.

    It would be interesting if he included a 1mp version to show how it was in
    the early days. It would show that 1mp to 4mp was a huge step. To match step
    up in resolution we'd have to to go from 4 to 16mp. For point and shoot
    digital, this has not been achieved and may not. It has been said that the
    megapixel race is over in the compacts as cramming more pixels on a tiny
    chip results in more noise. Never say never : )

    John
     
    JohnR66, Mar 5, 2006
    #3
  4. Rich

    Broake Guest

    "JohnR66" <> wrote in message
    news:ErsOf.14765$...
    >
    > "Broake" <> wrote in message
    > news:...
    >>
    >> "Rich" <> wrote in message
    >> news:...
    >> Now its clear why someone would be willing to pay $7000
    >>> or so for that
    >>> 16 meg Canon!
    >>> Here is the gallery:
    >>>
    >>> http://www.pbase.com/andersonrm/uprezing
    >>>

    >>
    >> Interesting - to my inexpert eyes the best trade-off in terms of
    >> price/performance comes at 10mp. The quality increase after that seems
    >> entirely disproportionate to the costs involved.
    >>
    >> Having said that, I once had a 4mp Panasonic Lumix p&s that produced
    >> shots greatly in excess of the quality you show for the 3.7mp image - so
    >> perhaps this test is somewhat flawed?
    >>

    > You missed his point. The 3.7MP was upsampled to match the size of the
    > 16mp. Of course it is going to be soft. There would be no need for 16mp if
    > 4mp looked as good.
    >
    > It would be interesting if he included a 1mp version to show how it was in
    > the early days. It would show that 1mp to 4mp was a huge step. To match
    > step up in resolution we'd have to to go from 4 to 16mp. For point and
    > shoot digital, this has not been achieved and may not. It has been said
    > that the megapixel race is over in the compacts as cramming more pixels on
    > a tiny chip results in more noise. Never say never : )
    >
    > John


    Ah, yes, 1mp! - I remember those.

    About 9 years ago I used to pass a shop every day, and look at a 1mp Fuji
    (MX???) that was in the window at a cool £999. I so much wanted that
    camera - but could never quite bring myself to spend a grand. Of course,
    now, I'm very glad I didn't.
     
    Broake, Mar 5, 2006
    #4
  5. Rich

    Guest

    JohnR66 wrote:
    <big snip>
    > It would be interesting if he included a 1mp version to show how it was in
    > the early days. It would show that 1mp to 4mp was a huge step. To match step
    > up in resolution we'd have to to go from 4 to 16mp. For point and shoot
    > digital, this has not been achieved and may not. It has been said that the
    > megapixel race is over in the compacts as cramming more pixels on a tiny
    > chip results in more noise. Never say never : )
    >
    > John


    Here is my simple version of a different resolutions demo:
    <http://members.iinet.net.au/~therealm//dj_nme/res-test/>
    The lower res versions were done by resampling the whole image and then
    cropping to the same part of the picture for each one.
     
    , Mar 5, 2006
    #5
  6. Rich

    Bob Williams Guest

    wrote:
    > JohnR66 wrote:
    > <big snip>
    >
    >>It would be interesting if he included a 1mp version to show how it was in
    >>the early days. It would show that 1mp to 4mp was a huge step. To match step
    >>up in resolution we'd have to to go from 4 to 16mp. For point and shoot
    >>digital, this has not been achieved and may not. It has been said that the
    >>megapixel race is over in the compacts as cramming more pixels on a tiny
    >>chip results in more noise. Never say never : )
    >>
    >>John

    >
    >
    > Here is my simple version of a different resolutions demo:
    > <http://members.iinet.net.au/~therealm//dj_nme/res-test/>
    > The lower res versions were done by resampling the whole image and then
    > cropping to the same part of the picture for each one.
    >

    Your original 8 MP image is really poor.
    The writing on the street sign is in poor focus even though its distance
    from the camera is about the same as that of the fire breather.
    There is really no detail in the image that one could use to compare
    sharpness and resolution.
    Bob Williams
     
    Bob Williams, Mar 5, 2006
    #6
  7. Rich

    Guest

    Rich wrote:

    > What I did was to take images at distances that corresponded to varying
    > pixel counts.
    > So (for eg) the image area under the lens for the 3.7Meg equivalent
    > occupied approx. 1/4
    > of the size of the 16Meg shot. The lens was set at its highest
    > resolution
    > point, f5.6. Only the centre of the image was used for the 100% crops. ...


    > Anyway, the result seems to be that the detail in the lesser megapixel
    > images is reduced
    > noticeably and this has resulted in a sharpness degradation as well.


    This test effectively always keeps the physical pixel size
    constant. That is, if your camera has say 8 Mp total and
    3 micron pixel size, when you crop to the inner 3 Mp you
    effectively have a 3 Mp sensor with 3 micron pixels, so
    a physically smaller sensor.

    Thus the test ignores that if _sensor size_ is constant,
    higher Mp count means smaller pixels, which means
    each pixel is noisier (and if the pixels get small enough,
    the lens resolution becomes a limiting factor). For example,
    in many popular series of digital P+S, the sensor size
    has stayed constant while the Mp count has increased
    (e.g. Canon S20, S30, S45, S50, S60, S70 all have a 5.3x7.2mm
    sensor while the pixel count went from 3Mp to 7Mp).
    As you go to higher pixel count there is a point of diminishing
    returns.

    So this test emphasizes pixel count and ignores sensor size,
    which I don't think is an accurate way to represent image
    quality.
     
    , Mar 5, 2006
    #7
  8. Rich

    Rich Guest

    That was my point, not to vary anything except the pixel count. You
    start changing
    pixel sizes and the test conditions change. But your point is
    understood, that smaller
    pixels produce other problems. Which is why a sensor of 4-5 meg based
    on the
    exact pixel configuration of a 6-8 meg DSLR sensor would be very good
    for a P&S
    camera.
    -Rich
     
    Rich, Mar 5, 2006
    #8
  9. Rich

    Guest

    Bob Williams wrote:
    > wrote:

    <snip>
    > > Here is my simple version of a different resolutions demo:
    > > <http://members.iinet.net.au/~therealm//dj_nme/res-test/>
    > > The lower res versions were done by resampling the whole image and then
    > > cropping to the same part of the picture for each one.
    > >

    > Your original 8 MP image is really poor.
    > The writing on the street sign is in poor focus even though its distance
    > from the camera is about the same as that of the fire breather.


    You obviously didn't bother looking at the change in detail in the
    cropped images.
    Whatever else you believe that you see in the rest of the full
    thumbnail of the image is not relevent to the obvious change in detail
    in the different resolution cropped images.

    > There is really no detail in the image that one could use to compare
    > sharpness and resolution.


    It is plainly obvious that the amount of visible detail changes with
    the change in resolution.
    What else is relevent in this case?
     
    , Mar 5, 2006
    #9
    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. Abrasha

    Hot pixel vs. stuck pixel

    Abrasha, Aug 2, 2003, in forum: Digital Photography
    Replies:
    5
    Views:
    8,512
    Steven Buglass
    Sep 2, 2003
  2. Tom Thackrey

    Re: Pixel size of individual Pixel

    Tom Thackrey, Sep 14, 2003, in forum: Digital Photography
    Replies:
    2
    Views:
    829
  3. Robert E. Williams

    Re: Pixel size of individual Pixel

    Robert E. Williams, Sep 14, 2003, in forum: Digital Photography
    Replies:
    2
    Views:
    550
    Don Stauffer
    Sep 16, 2003
  4. Peter H

    Fuji S3000 - 3.2 m/pixel or 6 m/pixel?

    Peter H, Nov 18, 2003, in forum: Digital Photography
    Replies:
    3
    Views:
    707
  5. Peter H
    Replies:
    43
    Views:
    1,224
    Bill M
    Dec 4, 2003
Loading...

Share This Page