3D GFX/3D GFX.iso/pcutils/os2/jview/jvwclrs.hlp

OS/2 Help File | 1995-12-31 | 24.9 KB | 517 lines

ΓòÉΓòÉΓòÉ 1. Color Utility Add on ΓòÉΓòÉΓòÉ

The functions contained in this DLL are designed to work with the JView imaging
system.  You will need to reference JView's main help file or the written
manual that comes with JView for help on controls that are common to all JView
dialogs, such as the colorwell.  This particular module is provided by:

                                 Crunch Products
                                  P.O. Box 392
                             Berkeley, CA 94701-0392
                                       USA


ΓòÉΓòÉΓòÉ 1.1. Common Controls ΓòÉΓòÉΓòÉ

All functions provided in the colors utility will have some of the following
controls.  For our purposes a control is considered to be a push button, a
slider, a checkbox, etc.  Their use is fairly straightforward but a detailed
explanation is provided here.

Low Priority  This checkbox is a signal to run the current operations using a
low priority thread.  A basic understanding of how OS/2 works is needed here.
OS/2 is multi-threaded, that is it allows multiple operations to occur
simultaneously.  So for instance, you may have your word processor doing a
spell check at the same time as you are converting a color image to black and
white.  Well, OS/2  doesn't really run both applications at exactly the same
time, rather it spends a few thousandths of a second running one program and
then switches to the other for a few thousandths.  Because we are talking about
such small amounts of time, it appears as if both programs are running
simultaneously.  What the low priority option does is allow you to say  "hey,
run the other program more often than you run this one."  It will make the
other program (and that other program could be you doing a file listing) much
snappier.  You can change the priority even while the process is running, if
you change your mind about  which program you are most interested in.
Something to beware is that sometimes DOS (or Windows) programs are such hogs
that if you specify low priority, this program will run so slowly as to be
unbearable.  In those instances simply turn low priority off.  Feel free to
play, you can do not harm.

Done  This button indicates that you are satisfied with the current state of
the image and are done using the current function.

Cancel.  This button will have two different effects depending upon whether you
are currently applying the function.  If you are testing the function, hitting
cancel will abort that current test and return the image back to the state it
was in before the test was started.  If no test is being run, Cancel will
return the image back to the state it was in before the current function was
run at all.  The function will also close.

Undo  This button will bring up the Undo Dialog.  This will allow you to
reverse changes you have made to the image.

Test  This button will apply the current function to the image.  It can be
undone by either using the Undo button or the Reset button.

Reset  The button will cause the image to revert to its most recent prior
state.  Beware that the current image state will be lost forever. This control
can be applied multiple times until only the last level of undo is reached.
This control is very similar to the Undo control, but it always goes back one
level.  It is most often used when you have mangled the image and know for sure
that you do not want to keep the changes.

Image  This will switch  focus to the image window and away from the current
dialog.

Real Time Preview  This control is only provided on some functions.  Basically,
it will popup a small copy of the current image in the upper left corner of the
screen.  As adjust other controls, the effects of those changes will be shown,
real time, on the sample image.

In addition to these controls, the image window will still have access to a
popup menu that is accessed via the right mouse button.  The commands presented
in that popup menu will vary, but are straightforward.

It is very important to realize that virtually all the function provided in
this utility pack can be applied on parts of the image as well as the entire
image.  Simply select that portion of the image you are interested in by using
the left mouse button (with the appropriate selection method - box, ellipse,
freehand).  You can also invert the selection region, so that everything but
the parts inside the bounding area will be affected.


ΓòÉΓòÉΓòÉ 1.2. Adjust (Colors) ΓòÉΓòÉΓòÉ

This function allows the users to adjust the image's colors in a variety of
different colorspaces.

RGB colorspace represents the image as being made up of pixels that each share
a red component (R), green component (G), and a blue component (B).  Each
component can have a value ranging from 0-255.  By moving the appropriate
slider, the component for all pixels selected will be adjusted by the indicated
amount.  For instance, if the blue slider is set to -25, then the blue
component (or channel) will be reduced by 25 to a minimum of 0.  A value of 0
would mean that the pixel will have no blue.  The RGB option also allows the
user to choose to manipulate only a portion of the image's tones.  The low
tones are the darker colors, while high is lighter.  So if you only want to add
a red tint to the clouds of an image that is otherwise dark, choose the High
radio button before moving the sliders.  Note The real time display option will
always show changes as if the entire image is being changed and as if all tones
are being changed.

HSV stands for Hue, Saturation, and Value.  It is often called the HSB
colorspace, where B stands for Brightness instead of Value.  This colorspace
can be thought of as a deformed version of the RGB colorspace.  Now, the RGB
colorspace is represented by a cube, with each channel represented by a
different orthogonal axis.  The HSV colorspace can be visualized as a cone,
except the base is hexagonal instead of circular.  The center of the base
corresponds to white, while the top of the cone is black.

HLS This colorspace is very similar to the HSV colorspace. It stands for Hue,
Lightness and Saturation.  Whereas HSV is a hexagonal cone, HLS is two
hexagonal cones with their bases back to back

Luv  This is the brother of the more commonly known Lab colorspace.  Since Lab
is not designed for use with electronic displays, Luv is used instead.  Luv
(Lightness, a red-green content, and a yellow-blue content parameter) is
considered by some to be a more accurate method of representing colors.  The
main advantage of Luv over HSV type colorspaces is that the three axis of
interest are truly independent which makes it possible to specify a luminance
that is held constant while hue or saturation are changing.

CMY The colorspace is also referred to as CMYK.  It is the complement of the
RGB colorspace, except now the axes are Cyan (C), Magenta (M), and Yellow (Y).
Whereas RGB is typically used to specify colors that are displayed on an
electronic display, CMY is more often used to specify colors for printing.  The
colors are additive such that when all three components are at  their maximum
(255), the resulting color is black, while the 0, 0, 0 point represents white.


ΓòÉΓòÉΓòÉ 1.3. Black and White ΓòÉΓòÉΓòÉ

The first thing you notice with the black and white dialog is the curve on the
top of the dialog.  When an image is converted to black and white (B&W), it is
first converted to grayscale.  The curve is used to modify how the colors get
mapped to grayscale. There are 255 possible levels of gray and any given value
of color (RGB) can be mapped to a given level of gray.  This curve shifts how
that mapping occurs.  By pulling the curve down, colors that would normally map
to light levels of gray, get shifted to darker levels.  The reason for doing
this is that you will often achieve more aesthetic images by having a slightly
S shaped curve.  You may use the right mouse button to change the type of curve
being used while the left mouse button can be used to move the curve's control
points.

There are several methods presented for doing the actual conversion to B&W.

Floyd Steinberg Dither  This is one of the most popular dithers and was
developed by R.W. Floyd and L. Steinberg.  It uses an error diffusion approach.
If you are familiar with error diffusion notation, it is represented as

  X  7
3  5  1
Of the error diffusion methods available, it is the fastest.

Stucki Dither  This is another error diffusion method, in this case one assumes
it was developed by someone named Stucki. It is represented as

      X 8 4
2 4 8 4 2
1 2 4 2 1
As you can see, it is larger than the Floyd Steinberg dither.  It takes more
time to run, but in many cases will provide superior results.  The definition
of superior is troublesome when applied here because what we are really
interested in is aesthetics.  So, it technically dithers better than FS, but
may not be aesthetically more pleasing.

Stevens and Arce Dither  This one is the king coconut of error diffusion.  See
the note about aesthetics above, but if you subscribe to the bigger is better
attitude, this one's for you. It is represented as

                   X        32
12        26       30       16
      12        26       12
5          12       12        5

Threshold  Quite simply, this one makes the pixel black if it's grayscale value
(see the first paragraph) is below the indicated threshold value, and white if
it is above it.  No one is quite sure what happens if the grayscale value is
exactly equal to the threshold value.  Some say that such a situation can't
occur, that it is an aberration against the natural order of the universe.
Others claim to have observed just such a situation under laboratory
conditions, but the experiments have been impossible to verify because they
depend on the use of a Tokamak and three dozen Cornish game hens.


ΓòÉΓòÉΓòÉ 1.4. Borders ΓòÉΓòÉΓòÉ

This is one of the few functions where the entire image must be used.  Use the
colorwell (double click) to choose an appropriate background color.  Enter the
amount of padding for each side by entering the amount in the appropriate spin
field.  Alternatively, check the Equal Sides checkbox, and all sides will be
based on the value indicated in the Left spin field.

The four buttons (640 X 480, etc.) represent screen dimensions and by pushing
on one, the padding needed to fill such a screen is automatically calculated.


ΓòÉΓòÉΓòÉ 1.5. Contrast ΓòÉΓòÉΓòÉ

Contrast refers to the difference in extremes of colors a given image uses. By
increasing the contrast, an image can appear more vibrant and details can
become more visible.  Imagine you image is composed of three colors, all gray
that are very similar in intensity.  By increasing the contrast, the lowest
intensity gray would be made darker, while the lightest gray would be made even
lighter.  The result is that each color is more distinguishable.  The opposite
effect is also possible.

The downside of contrast enhancement is that once colors reach a limit in terms
of brightness, they can begin to do a color shift.  For example, image the
color being adjusted starts out with an RGB point of 128, 255, 255.  This means
that it already has maximum green and blue.  To increase this color's contrast
the only possible option is to increase red, but this serves to make the color
whiter.

The contrast slider has a range of -128 to 128.  The scale is not too
important, you will do better to play with values to achieve the desired
effect.

Instead of the setting a contrast value, you may choose to let the application
attempt to Auto Enhance the image.  This algorithm works by computing a
histogram (a record of how many times each color is used) of the image and then
forcing the percentage of the image indicated towards black and also the other
amount indicated to white.  The higher the percentages indicated, the greater
the contrast that will be acheived, but the cost will be the potential color
shift mentioned above.


ΓòÉΓòÉΓòÉ 1.6. Edit Palette ΓòÉΓòÉΓòÉ

This function is only applicable for images that have 256 or fewer colors.  If
your image is 24 bit, you must first reduce it to 8 bit.

Basically, just double click on the color you are interested in changing.  This
will bring up the colorwheel so that you may indicate the desired changes.  The
changes will take place over the entire image, not just a selected area.  The
palette does support dragging and dropping of colors from the colorwheel or
other palettes.

 Swap and Copy are used to indicate what will occur when colors within the
displayed palette are dragged and dropped.  If Swap is selected, then the color
that is picked up will be swapped with the color the is dropped onto.  Copy
will cause the color that is dropped onto to become the color that was dropped.

Find Color will bring up a magnification window in the upper left corner of the
screen.  Simply position the cursor over the color you are interested in
finding and hit the right mouse button.  The color can be anywhere on the
screen. If no exact match can be found, the closest color is used.

There is a slim button at the top of the palette control.  Push it to bring up
different ways of sorting the displayed colors.  If it wasn't obvious, use the
right mouse button to drag and drop colors.  Note the colors display may be
only approximations if your system is only running at 8 bits.


ΓòÉΓòÉΓòÉ 1.7. Gamma ΓòÉΓòÉΓòÉ

Ideally an electronic display will linear for luminance with respect to
voltage.  For many Apple Macs, this is the case, but for most PCs, it is not.
The result of this is that an image displayed on one screen, may look murky on
another.  So, gamma correction simply alters how an image's luminances are
actually displayed.  In theory, the correction factor can vary depending on age
of the monitor, weather, etc.

Gamma correction is really an aesthetic question, it is best to just try
different values.  Greater than one will brighten the image, less than one will
darken it.


ΓòÉΓòÉΓòÉ 1.8. Grayscale ΓòÉΓòÉΓòÉ

There are 255 possible levels of gray.  The mapping of a given color into the
corresponding gray level is simple.  However, aesthetics are often improved by
shifting that mapping.  The S shaped curve presented at the top of this dialog
is used to adjust the grayscale mapping.  A straight line from lower left to
upper right would mean no mapping.  The S curve means that some of the dark but
not black colors get mapped to black, and some of the light but not white
colors get mapped to white.  Use the right mouse button to choose the type of
curve, and the left mouse button to adjust the curves control points.

You may choose how many levels of gray to provide by selecting the appropriate
radio button.


ΓòÉΓòÉΓòÉ 1.9. Histogram ΓòÉΓòÉΓòÉ

This function shows the user the histogram of the current image.  A histogram
is a calculation of how many pixels of the image are a given intensity.  Or
more simply, how many times is a given color used.  The figure at the top of
the dialog is a qualitative display of how often different colors are used.  A
single line would mean that only one level of gray is used.  It is in terms of
gray because we are interested in intensities, not true colors.  Histogram
equalization can be applied to your image, but the image will be reduced to
grayscale.  This equalization (which occurs when AHE is selected) attempts to
spread out the use of colors.  To a certain degree, this is similar to contrast
enhancement.

The AHE checkbox represents the use of tradition histogram equalization
techniques.  The user can elect to preserve the maximum and minimum values the
image possesses or allow them to be mapped to 255 and 0 respectively for
greater contrast.

Alternatively, the user can use a clip limit in conjunction with regions to
perform Contrast Limited Adaptive Histogram Equalization, as described in
"Graphics Gems IV", Academic Press, 1994.  This will be used when AHE is not
checked.  Basically the image is broken into regions and histogram equalization
is performed on each individual region.  The effects can be quite profound in
bringing out lost details. Try a large number of regions and a high clip limit
(20,000).


ΓòÉΓòÉΓòÉ 1.10. Invert ΓòÉΓòÉΓòÉ

Not too many options here!  Inversion is quite simple.  A given colors is
represented by its three color components, Red, Green, and Blue.  These
components have levels that range 0-255.  Inversion takes the component level
and subtracts it from 255 (mathematically red = 255 - red).  The result is the
inverse of the image or for you photographers out there you get the
photographic negative.


ΓòÉΓòÉΓòÉ 1.11. Reduce ΓòÉΓòÉΓòÉ

This dialog is used to reduce the number of colors that the current image is
using.

Count will count the number of colors currently in use.  On a 24 bit image the
count will give up when the number of unique colors exceeds 65536.  On 8 bit
images it will calculate the number of unique colors, not just the number of
colors in separate channels.  What does this mean.  Well, your image might
think it has 234 different colors (an example number).  But in reality, it has
one color that is white and 233 colors which are all the same - say black.
Counting will actually compare the colors used and detect when the image has
duplicates.

If the current image is 24 bit, it can be reduced to either 16, 15, or 8 bit.
As far as OS/2 goes, there are only 24 bit and 8 bit images so there is no
memory saving for going to 15/16 bit.  It is useful to go to those color depths
if the image is not displaying well at 24 bit.  This may occur if your system
is set up to display at 15 or 16 bit. When going to 15/16 bit either an Error
Diffusion method or Jitter method can be used.  With the jitter method a noise
level is also specified.  The higher levels of noise will break up banding more
but will also begin to degrade the image.

When going to 8 bit, color mapping can be achieved by going to the most used
colors (which is the most conventional way) or by collapsing the colors used by
combining those which are closest together (nearest color).  When most used
colors is selected, you can optionally apply error diffusion (Floyd Steinberg).
With 8 bit mapping, you can specify the actual number of colors used, not just
the color depth.


ΓòÉΓòÉΓòÉ 1.12. Remap ΓòÉΓòÉΓòÉ

Remapping is one of the most versatile operations that can be performed.  It
allows the ability to do very controlled contrast manipulations as well as
color inversions.

All changes are input through the curve control.  You may operate on all three
color channels at once or allow them to be adjusted independently (it uses RGB
color space).  When you have selected the R (red) channel, the other channels
will still be applied as displayed, however you will not be able to change
them.  This is so that you can easily manipulate a particular channel and still
have the others displayed.

The curve control should be thought of as representing two axis.  The vertical
axis is where the value indicated on the horizontal axis will be mapped to.
Each axis ranges from 0 to 255, which corresponds to the range of the RGB
colorspace.  For example, a straight line going from the lower left corner (the
origin at 0, 0) through the upper right corner (255, 255) can be mathematically
represented by the curve x = y.  What this means is that a value of 45 for red,
would be converted to a value of 45 - in other words there would be no change.
If instead of the line x = y, we had a horizontal line starting at the origin,
its equation would be y = 0.  Then our value of 45 for red, would now be
converted to 0.  Such a curve would map the whole image to black.  Sound
confusing?  You bet!  Well, just play with it for a few minutes and you'll see
what's happening.  Don't forget to use the right mouse button to change the
types of curves available.

The dialog allows the use of real time changes display.


ΓòÉΓòÉΓòÉ 1.13. Swap Palette ΓòÉΓòÉΓòÉ

Swap palettes will display two palettes.  The current palette (on the left) and
the palette that will be used instead.  This dialog can only be used on 8 bit
images.

The current palette cannot be edited, but the new palette can be edited just as
described in the Edit Palette section.  Although you cannot edit he current
palette, you can reorder the display of colors.

The colors of the current palette and new palette are swapped position for
position.  Thus, the color displayed in the upper left hand corner of the
current palette will become the color displayed in the upper left hand corner
of the new palette.

The list box in the lower right corner of the dialog contains several built in
palettes that the user may find interesting.  Two of them may seem confusing.
7R 8G 4B is the standard palette used on the 8514 video card.  7R means it has
7 levels of Red, 8G means 8 level so green, etc.  6R 6G 6B follows a similar
convention.  Fire is just a cool palette, try it with your original palette
expanded and sorted by intensity.

The Load and Save buttons allow you to store and retrieve palettes.

Add allows you to add a new color to the new palette, assuming it contains less
than 256 colors.

Find will show you where a particular color is on your current palette.  Use
the right mouse button, with the help of the magnified display that will appear
in the upper left hand corner of the screen, to indicate the color of interest.

Expand is used when your current palette is less than 256 colors.  Expand will
attempt to fill in spaces in the color table so that it appears as if your
current palette is 256 colors.  This is used so as to minimize the amount of
positioning needed to align palettes of different sizes.

Note about 8 bit systems.  An 8 bit system means you can only display 256
unique colors at a given time.  Yet, this very function means that your are
likely trying to display upwards of 500 different colors at the same time.
Something has to give.  Either the colors will not be represented in their true
state, or a certain amount of mapping will occur depending on which control has
the focus.  There is no cure for this except for upgrading to a 15 bit or
higher video system


ΓòÉΓòÉΓòÉ 1.14. System Colors ΓòÉΓòÉΓòÉ

Convert the image to display using the 8 bit system palette. If error diffusion
is selected, Floyd Steinberg error diffusion is used, otherwise colors are
mapped to the closest system color.

Having images in the system colors is useful when multiple images are to be
displayed on an 8 bit system.  Further, when creating background or lockup
images, system colors give the most pleasing results.

In order to use the error diffusion, you will have had to have created the
Jview.dth file.  To create this file, close the System Colors dialog.  Bring up
the popup menu with the right mouse button and look under Misc. for the create
JView.dth command.  Follow its instructions.


ΓòÉΓòÉΓòÉ 1.15. Tones ΓòÉΓòÉΓòÉ

This is really one of the cooler functions once you get the hang of it. A
little background about printing.  You start with white paper and then add a
drop of color to it - lets say that drop has equal parts of red, green and
blue. That first drop will look really pale, because there isn't too much ink
and the light can penetrate the ink and gets reflected off the white paper
without too much loss of energy.  Now add a second drop, right on top of the
first.  The point has gotten darker, because the light can't penetrate as
easily.  Keep adding drops and the point turns black.  Now, once that point is
black, it doesn't matter if you add more red, or green, or blue, it just stays
black.  Now, many printing operations only use a very limited number of ink
colors, for now lets say one.  That ink color is usually black.  By varying the
amount of ink used, (remember above) a given point will appear white (no ink),
gray (some ink), or black (lots of ink).  In this manner, you get a grayscale
image.

Now for the fun part.  Why use black ink?  You can do the same with a nice warm
magenta.  No ink still means white, but were gonna cheat and say lots of ink is
just solid magenta.  This is a monotone image.  In our case you use the
colorwell to indicate the type of ink you want, and the curve window indicates
how the ink gets added.  The x, or horizontal axis, represents what level of
intensity on the image we are talking about (range is 0 - 255), while the
vertical axis indicates how much ink (what level of intensity) of the color
will appear there.  So the zero point on the x axis refers to the dark parts of
the image - 0 intensity, while the end of the x axis refers to the lightest
parts of the image.  Now the y axis may seem a little confusing.  We want high
values y to correspond to a high brightness level of the color.  And high
brightness means little ink.  This contrasts to low brightness which means lots
of ink (remember how printing works).  So, if we have a really low y axis value
(say 0) that means use all the ink you can, or make the corresponding x axis
get solid color.  A y value of 255  (the top of the axis) would mean no ink, or
solid white.

But why only use one color, lets try blending two colors.  Things are a little
trickier, but the results are worth it.  Lets do a dualtone (select the
dualtone radio button). Choose black for the first color, this will give us a
grayscale image to start with.  For now, just drag the black color in the first
colorwell onto the second colorwell.  Test the result, you will get a very dark
image.  What happened, the color is the same for both tones?  Remember how
printing works, you have effectively printed one image and then printed and
identical image right on top of the first - you are using lots of ink.  Think
of it as printing the image multiple times, one on top of the other.  You need
to reduce the amount of ink used on each pass.  Take the color control curves
for both colors and move them up and test the image again (also remember to
reset the image first!).  Things probably look better now, if not, move the
curves up even higher (my God, not that high!!!).  Ok, pretty dull.  But now
change the color in the second colorwell to a new color, try a magenta-like
color.  You can add warmth to a grayscale image this way.

Just as the method outlined above applies to dualtone, so it applies to the tri
and quad tones.  Just remember that you are effectively adding more ink, the
more tones you use and so you will have to reduce the amount of ink each tone
adds.


ΓòÉΓòÉΓòÉ 1.16. Transpose ΓòÉΓòÉΓòÉ

This tool is used to replace a color, or range of colors with one new color.

Three different colors can be entered.  The colorwell in the upper left region
is the base color, i.e. the main color to be changed. Beneath the base color is
the optional range color.  If a range color is used, then any color contained
in the RGB colorspace existing between the base color and the range color will
be mapped to the new color.  The new color is located in the colorwell to the
right of the base color.

If no range color is indicated, the tolerance value will be used with the base
color.  A tolerance of 0 means that only colors that exactly match the base
color will be changed.  If a tolerance of 2 were to be used, then all colors
which were with 2 units of the base color in the red, green, and blue channels
would be replaced by the new color.

Colors are input into a color well by drag/drop or by double clicking with the
left mouse button.