difference between dynamic range and image contrast

Discussion in 'Photography' started by camera_dude, Feb 26, 2008.

Thread Status:
Not open for further replies.
  1. camera_dude

    camera_dude Member

    Messages:
    15
    Likes Received:
    0
    Trophy Points:
    0
    I am trying to understand the difference between dynamic range and image contrast

    Does the definition of image contrast apply to shades of grey while dynamic range applies to the spectrum.

    As the pixels are affected by the same light (photons) hence will the image contrast and the dynamic range be the same.

    I was looking at the ISO 14524 test chart, if the value of the range is say 128, does that equate to 7 f-stops?
     
  2. David Rasnake

    David Rasnake News/Review Writer News/Review Writer

    Messages:
    2,170
    Likes Received:
    7
    Trophy Points:
    0
    Dynamic range refers to how wide a range of values a camera sensor can capture. At its most technical, dynamic range has little to do with contrast (which is primarily a tone curve issue): it is simply a measure of the highest and lowest possible light amounts a camera can capture (technically expressed, it is a ratio of these two values).

    Contrast (that is, how much differentiation exists between various tones in an image) is defined by an image's tone curve. My understanding is that by default, an image sensor captures light in a linear fashion - this is to say, no particular tonal value is emphasized over any other, and thus if we were to plot the capture on a graph, it would appear as a straight line. In-camera processing applies a particular tone curve to make images look more pleasing and have more contrast. Thus our straight line takes on an curved shape: in most cases, camera processing defaults to an s-curve, or contrast curve, in which the lower (shadow) values are bumped up slightly, and the higher (highlight) values are scooped downward. The result is better differentiation of mid-tone values - that is, more contrast.

    This tone curving can alter apparent dynamic range by pushing highlight values and shadow values at the very ends of the spectrum off the ends of the camera's reproducible dynamic range.

    On the ISO chart you reference, a value of 128 is indeed 7 stops of dynamic range (128=2^7).

    Does this help?
    dr
     
  3. camera_dude

    camera_dude Member

    Messages:
    15
    Likes Received:
    0
    Trophy Points:
    0
    thanks for the reply, it definitely does help

    does contrast just gives the range, and not the absolute maximum and minimum values.

    so if i understand correctly the camera image sensor defaults to a s curve, then it does shed the highest and lowest tones.

    and the contrast has a lower or at most equal range compared to the dynamic range.

    i guess the tonal curve is something which is unique for each camera image sensor, can the user vary this?
     
  4. David Rasnake

    David Rasnake News/Review Writer News/Review Writer

    Messages:
    2,170
    Likes Received:
    7
    Trophy Points:
    0
    Contrast curving is actually a function of the camera's image processor, and not a function of the sensor. As I said, raw sensor capture is linear; you can get linear (uncurved) output if you have a camera that shoots raw files.

    I'm on a tester laptop at the moment, so I don't have a great file to show this off with, but I've imported a couple of JPEGs of the same shot that I created from a raw file (click the thumbs for a larger shot).

    contrast.jpg

    linear.jpg

    In the first example, a pretty steep contrast curve was applied. The values used pretty closely mimic the kind of contrast curving you get from the processors in most point-and-shoot digital cameras, as consumers tend to find this look more pleasing and immediately printable.

    Compare that to the second image, which is the camera's raw linear capture. Note that there's much more detail in the mid-shadows, but that the values approaching black also look much more washed out. While there's less detail loss, this isn't a very appealing look, as we tend to see the world in a way that more closely resembles the contrast curve.

    To your question about if the contrast curve is variable, you can (obviously) vary it if your camera shoots raw (if you bring the image into Adobe Camera RAW or some similar you have a curves option, in which the contrast curve can be manipulated by the user). If your camera has a contrast adjustment (usually buried deep in the menus) there's also some variability, and some cameras with less true manual control provide different shooting modes (i.e. normal, vivid, natural, etc.) that alter both color/saturation and apply different curves.

    dr
     
  5. camera_dude

    camera_dude Member

    Messages:
    15
    Likes Received:
    0
    Trophy Points:
    0
    thanks for the detailed reply and the pics, David

    The RAW files hold linear data, without any contrast curving applied, correct?

    with a steep contrast curve, one would accentuate the illumination (affecting all colors) in the midrange at the expense of the tones at the ends.
    I guess the sensor's contrast curve is the least steepest you can go for, since if you reduce the slops, you just lose the ability to accentuate with no gain

    Do all cameras, even the cheapos have contrast curving processor capability?

    so the different shooting modes (i.e. normal, vivid, natural, etc.) result in different curve being applied?

    What are mid shadows :D
     
  6. David Rasnake

    David Rasnake News/Review Writer News/Review Writer

    Messages:
    2,170
    Likes Received:
    7
    Trophy Points:
    0
    Correct.

    Yep, this is standard practice for image processors (image processors themselves only come from a few sources, much like sensors, though it's hard to know what processor is in what camera - only a few companies, like Canon with the DIGIC processors, actually make/spec their own). Though it wasn't easy to tell in the example I posted, linear contrast curving tends to look strange and unnatural, so no consumer camera would process an image this way by default.

    In many cases, yes, though it's often hard to know/tell.

    I was trying to describe the values that are between the mid-point and true shadow range. I guess you could call them lower mid-tones. "Mid shadows" is certainly not a technical term; just me fumbling for a word. :)

    dr
     
  7. camera_dude

    camera_dude Member

    Messages:
    15
    Likes Received:
    0
    Trophy Points:
    0
    thanks for the reply and all the explanations. i feel like i can go into jessops and have a proper conversation with them now :)

    true shadow is the saturated blank correct, and the midpoint would be the tone at the centre of the spectrum?

    with a steep contrast curve, one would accentuate the illumination (affecting all colors) in the midrange at the expense of the tones at the ends, correct?

    I guess the sensor's contrast curve is the least steepest you can go for, since if you reduce the slops, you just lose the ability to accentuate with no gain?
     
  8. David Rasnake

    David Rasnake News/Review Writer News/Review Writer

    Messages:
    2,170
    Likes Received:
    7
    Trophy Points:
    0
    Right. Mid-point is the 50% value (50% gray in monochrome).

    It's easier to demonstrate what happens than it is to discuss it (for me at least), and with that in mind, here's the same photo with three different curves:

    NoCurve.jpg
    Linear

    ShallowCurve.jpg
    Shallow Curve

    SteepCurve.jpg
    Steep Curve

    Trying to "shallow out" the curve beyond a linear progression results in varying degrees of color inversion:

    Inverted.jpg

    Simply stepping a linear curve down the scale (reducing the "slope" as you suggest) makes the image uniformly darker. In either case, there is no dynamic range gain.

    dr
     
  9. camera_dude

    camera_dude Member

    Messages:
    15
    Likes Received:
    0
    Trophy Points:
    0
    thanks for the reply,

    is the linear curve a true picture (what you see is what you get)

    would the use of a contrast curve be to prevent under exposure and over exposure?

    i came across this contrast tone tutorial
    http://www.cambridgeincolour.com/tutorials/photoshop-curves.htm

    is dynamic range the input to the contrast curve, and contrast the output?
     
    Last edited: Mar 1, 2008
  10. David Rasnake

    David Rasnake News/Review Writer News/Review Writer

    Messages:
    2,170
    Likes Received:
    7
    Trophy Points:
    0
    Not really. While it records the light without boosting any particular area, this isn't the way our eyes see things. In the sample images from the previous post, the shallow contrast curve comes much closer to actual lighting and contrast as I saw it.

    Not usually. Contrast curving largely changes the relationships between the lighting values within an image, though contrast curving could be used to "compress" highlight and shadow levels toward the center.

    Yes, the image's total dynamic range is the input axis, and the curved image is the output. If the line has a 1/1 (45-degree) straight-line slope, this is linear contrast.

    dr
     
Thread Status:
Not open for further replies.

Share This Page