Re: Circle of confusion

Discussion in 'Digital Photography' started by Chris Malcolm, Aug 12, 2011.

  1. Alfred Molon <> wrote:
    > In article <>, Paul Furman
    > says...


    >> True. So the pixel-count/sensor-size is really the determining factor if
    >> you're going to make big prints.


    > So the minimum detail a sensor can capture is two pixel long? And that
    > would be the CoC to use (in case you make large prints)?


    I have some images of twigs against a winter sky where the twigs
    register in the image as only one pixel thick -- because the twig
    optical image was less than one pixel in width, and was smeared out a
    bit to be one pixel. When smeared out to be two pixels wide because it
    was on the boundary between them, they're much lighter grey. So as they staircase around in the image they're mostly just one pixel wide.

    Would you count that as one pixel detail or two? Seems to me that
    in that case the important CoC is one pixel or arguably a bit less
    than one in size.

    --
    Chris Malcolm
     
    Chris Malcolm, Aug 12, 2011
    #1
    1. Advertising

  2. Chris Malcolm

    Martin Brown Guest

    On 16/08/2011 11:34, Alfred Molon wrote:
    > In article<>, Chris Malcolm says...
    >
    >> Would you count that as one pixel detail or two? Seems to me that
    >> in that case the important CoC is one pixel or arguably a bit less
    >> than one in size.

    >
    > To see a one pixel thing, the adjacent pixel must have a different
    > colour otherwise you cannot see the one pixel thing. So the minimum
    > detail is two pixels wide.


    Incorrect. At least for high contrast features on a Bayer mask the
    inferred luminance alone *is* sufficient to see thin tree branches.

    There is a special case that iff you have pure monochromatic red or blue
    light then the image is compromised in resolution to an effective 2x2
    grid - although a quirk in how the JPEG image is encoded makes it look
    like 2x1.

    I have an image of the transit of Mercury taken in H-alpha and the round
    planet is decidedly oval because it is red channel only data.

    Regards,
    Martin Brown
     
    Martin Brown, Aug 16, 2011
    #2
    1. Advertising

  3. Paul Furman <> wrote:
    > Chris Malcolm wrote:
    >> Alfred Molon<> wrote:


    >>> In article<>, Paul Furman
    >>> says...

    >>
    >>>> True. So the pixel-count/sensor-size is really the determining factor if
    >>>> you're going to make big prints.

    >>
    >>> So the minimum detail a sensor can capture is two pixel long? And that
    >>> would be the CoC to use (in case you make large prints)?

    >>
    >> I have some images of twigs against a winter sky where the twigs
    >> register in the image as only one pixel thick --


    > I'd be curious to see that; it strikes me as pretty rare. Especially
    > without sharpening.


    It is very rare in the sense that I had to deliberately set up such a
    photograph to see it, with the right distances and dimensions of the
    samllest twigs. I also sharpened the image because I processed it from
    RAW and always employ sharpening when doing that, the amount depending
    on the characteristics of lens/aperture and image.

    --
    Chris Malcolm
     
    Chris Malcolm, Aug 18, 2011
    #3
  4. Alfred Molon <> wrote:
    > In article <GJq2q.3311$>, Martin Brown says...
    >> Incorrect.


    > No. If the adjacent pixels have the same RGB values, you can't see the
    > one pixel thing. You will see a multipixel blob, but not a one pixel
    > thing.


    But the adjacent pixels don't have the same RGB values, as
    demosaicing them will use different pixels. Even if the
    light gets spread to the right/bottom/right-bottom pixel.

    -Wolfgang
     
    Wolfgang Weisselberg, Sep 3, 2011
    #4
  5. Alfred Molon <> wrote:
    > In article <>, Wolfgang
    > Weisselberg says...


    >> But the adjacent pixels don't have the same RGB values, as
    >> demosaicing them will use different pixels. Even if the
    >> light gets spread to the right/bottom/right-bottom pixel.


    > You are basically saying that if you point the camera on a homogenous
    > monochrome object, bayer demosaicing produces errors. That's plausible
    > and is a reason why a full colour sensor is way better.


    Any sensor will have photon noise, even a full colour sensor.
    Especially a Foveon.

    But I was thinking of non-homogenous objects, since the topic
    was CoC and details and printing large, e.g. your question "So
    the minimum detail a sensor can capture is two pixel long?"


    BTW, yes, a sensor can capture a single detail standing out
    that's less than a single pixel wide, but it cannot capture and
    distinguish two adjacent details that do not have enough space
    between them. What is enough space also depends on the lens,
    which in turn is influenced how far the detail's from the center
    and whether it's radial or sagittal.

    A perfect lens would allow a sensor to capture repeated details
    with 1 pixel space between them. With AA-filters, Bayer and
    imperfect lenses you sometimes get moire with such close details,
    but you'll resolve not much worse than that distance. Just look
    up the resolution (line pairs per image height), multiply that
    number by 2 (because it's a dark and a bright line, i.e a pair)
    and divide it through the vertical pixel count ...

    You'll find some numbers at e.g. photozone.de

    -Wolfgang
     
    Wolfgang Weisselberg, Sep 16, 2011
    #5
    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. jgerard

    Re: 10D Circle of Confusion

    jgerard, Jul 17, 2003, in forum: Digital Photography
    Replies:
    9
    Views:
    437
    Bart van der Wolf
    Jul 19, 2003
  2. Donald Gray

    Circle of Confusion

    Donald Gray, Jul 3, 2004, in forum: Digital Photography
    Replies:
    8
    Views:
    434
    Dan Wojciechowski
    Jul 7, 2004
  3. Marvin Margoshes

    Re: Circle of Confusion

    Marvin Margoshes, Jul 4, 2004, in forum: Digital Photography
    Replies:
    2
    Views:
    538
    Gordo Laqua
    Jul 5, 2004
  4. Col's Cavern

    Jisatsu circle (Suicide Circle)

    Col's Cavern, Jun 7, 2004, in forum: DVD Video
    Replies:
    1
    Views:
    593
    Justin
    Jun 7, 2004
  5. M

    Circle of Confusion - applied values

    M, Jun 17, 2008, in forum: Digital Photography
    Replies:
    2
    Views:
    708
    Don Stauffer in Minnesota
    Jun 18, 2008
Loading...

Share This Page