|
Effects Of Filters
on Detail |

|
Lets take a look at the effects
of some common filters on the histogram. Before I do, I want to
mention something about detail. Where do you get detail from? The
detail in any image is contained in the subtle difference between
adjacent pixels. If you have a row of white pixels with a row of
black pixels beside them, you will see this as an edge. Most edges
are not this clear cut, though. Most are more like a gradual change
across a few pixels, with each being a little different from its
neighbor. You have to be very careful with some of the filters that
shift color, since they can remove the differences between these
pixels, thereby destroying detail.
|

|

|
Let's go back to the
image I used in the color shift. Here is the original
image with a histogram of luminosity. |

|
I chose
this image because it fills almost the whole range of
the histogram (OK, and because I'd rather look at a
woman than a tree). It is fairly even, as far as the
color distribution is concerned. The first filter I
am going to use is the contrast filter. The contrast
filter changes the range of luminosity values of the
pixels in your image. |
|
 |
Contrast Filters |
Decreased
Contrast |

|

|
In this image, I have
reduced the contrast by 50. The image is washed out,
faded. This is because reducing contrast tends to narrow
the total range of luminosity in an image. Notice how
all the information is now squeezed into the center
of the graph. |
This is
typical of a low contrast image. If you would do this
to your image and then save it, you would be losing
some detail. Try this. Load any image into your paint
program and reduce the contrast to 0. The whole image
will go gray. That is because when you are reducing
contrast, you are reducing the difference in luminosity
value between pixels. Since detail is contained in those
differences, you are thereby discarding detail. |
|
|
Increased Contrast |
 |

|
In this image I have
increased contrast by 50. This will increase the range
of luminosity values in your image. Notice how the mid-range
has been suppressed, and the spikes at the high and
low end. If you save this image, you are also discarding
detail. Any subtle differences in the highlights and
shadows have now been squeezed together. The effect
of this is to lose detail in these areas. |
There
are two other filters that work about the same as this
one, the equalize and stretch filters. These filters
are essentially automatic contrast filters, and I have
never found much use for them, nor do I recommend using
them. If you are going to adjust your image, don't let
the computer make your choices for you. It can not see
the image, and can't control the effect. |
|
 |
|