An Alpha value indicates the transparency of a pixel. The smaller
the alpha value of a pixel, the more visible the colors below it.
A pixel with an alpha value of 0 is completely transparent.
Within GIMP, Alpha values can be associated
with the image as a whole (the Alpha Channel) and with individual
layers (a Layer Mask). You can view these by using the Channels
dialog and the Layers dialog, respectively.
With some image
you can only specify that a pixel is completely transparent or
completely opaque. Other file formats allow a variable level of
Antialiasing is the process of reversing an alias, that is,
reducing the “jaggies”. Antialiasing
produces smoother curves by adjusting the boundary between the
background and the pixel region that is being antialiased. Generally,
pixel intensities or opacities are changed so that a smoother
transition to the background is achieved. With selections, the
opacity of the edge of the selection is appropriately reduced.
A spline is a curve which is defined mathematically and has a set
of control points. A Bezier spline is a cubic spline which has
four control points, where the first and last control points
(knots or anchors) are the endpoints of the curve and the inner
two control points (handles) determine the direction of the curve
at the endpoints.
In the non-mathematical sense, a spline is a flexible strip of
wood or metal used for drawing curves. Using this type of spline
for drawing curves dates back to shipbuilding, where weights were
hung on splines to bend them. The outer control points of a
Bezier spline are similar to the places where the splines are
fastened down and the inner control points are where weights are
attached to modify the curve.
Bezier splines are only one way of mathematically representing
curves. They were developed in the 1960s by Pierre Bezier, who
worked for Renault.
Bezier curves are used in GIMP as component
parts of Paths.
The image above shows a Bezier curve. Points P0 and P3 are points
on the Path, which are created by clicking with the mouse. Points
P1 and P2 are handles, which are automatically created by
GIMP when you click on the line between P0 and
P3 and stretch it. They change position when you stretch the
curve in different ways. You can also create and move the handles
by dragging them out of the anchor points to bend the curves in
the desired direction.
The image above shows a path which consists of two components,
having both straight and curved segments, being worked on with the
Path tool. Here, the open
circle indicates the selected anchor and the two open squares are
the two handles which are associated with this anchor from the
curves on either side of it.
BMP is an uncompressed image
designed by Microsoft and mainly used in Windows. Colors are
typically represented in 1, 4 or 8 bits, although the format also
supports more. Because it is not compressed and the files are large,
it is not very well suited for use in the internet.
The Free On-line Dictionary of Computing (13 Mar 01)
bitmap — A
data file or structure which corresponds bit for bit with an image
displayed on a screen, probably in the same format as it would be
stored in the display's video memory or maybe as a device independent
bitmap. A bitmap is characterised by the width and height of the image
in pixels and the number of bits per pixel which determines the number
of shades of grey or colours it can represent. A bitmap representing a
coloured image (a “pixmap”) will usually have pixels
with between one
and eight bits for each of the red, green, and blue components, though
other colour encodings are also used. The green component sometimes
has more bits than the other two to cater for the human eye's greater
discrimination in this component.
Bump mapping is a technique for displaying extremely detailed
objects without increasing the geometrical complexity of the
objects. It is especially used in 3-dimensional visualization
programs. The trick is to put all the necessary information into
a texture, with which shadowing is shown on the surface of the
Bump mapping is only one (very effective) way of simulating surface
irregularities which are not actually contained in the geometry of
A Channel is a single component of a pixel's color. For a colored
pixel in GIMP, these components are usually Red,
Green, Blue and sometimes transparency (Alpha). For a
Grayscale image, they are
Gray and Alpha and for an
Indexed color image,
they are Indexed and Alpha.
The entire rectangular array of any one of the color components for
all of the pixels in an image is also referred to as a Channel. You
can see these color channels with the Channels dialog.
When the image is displayed, GIMP puts these
components together to form the pixel colors for the screen, printer,
or other output device. Some output devices may use different
channels from Red, Green and Blue. If they do,
GIMP's channels are converted into the
appropriate ones for the device when the image is displayed.
Channels can be useful when you are working on an image which needs
adjustment in one particular color. For example, if you want to
remove “red eye” from a photograph, you might work on
the Red channel.
You can look at channels as masks which allow or restrict the output
of the color that the channel represents. By using Filters on the
channel information, you can create many varied and subtle effects
on an image. A simple example of using a Filter on the color
channels is the
In addition to these channels, GIMP also allows
you to create other channels (or more correctly, Channel Masks),
which are displayed in the lower part of the Channels dialog. You
can convert a selection to a channel mask by using the
Save to Channel
command. You can also create a channel by right-clicking in the
Channels dialog and using the
New channel command.
See the glossary entry on Masks
for more information about Channel Masks.
The Clipboard is a temporary area of memory which is used to
transfer data between applications or documents. It is used when you
Cut, Copy or Paste data in GIMP.
The clipboard is implemented slightly differently under different
operating systems. Under Linux/XFree, GIMP uses
the XFree clipboard for text and the GIMP
internal image clipboard for transferring images between image
documents. Under other operating systems, the clipboard may work
somewhat differently. See the GIMP documentation
for your operating system for further information.
The basic operations provided by the clipboard are
“Cut”, “Copy”, and “Paste”.
Cut means that the item is removed from the document and copied to
the clipboard. Copy leaves the item in the document and copies it to
the clipboard. Paste copies the contents of the clipboard to the
document. The GIMP makes an intelligent decision
about what to paste depending upon the target. If the target is a
canvas, the Paste operation uses the image clipboard. If the target
is a text entry box, the paste operation uses the text clipboard.
Color Depth is simply the number of bits used to represent a color.
With a color depth of 1, the image can only contain black and white
pixels. With a color depth of 4, 16 colors can be represented.
Colors in GIMP have a depth of 24 bits, with 8
bits each for Red, Green and Blue, which results in 16,777,216
A color model is a way of describing and specifying a color. The term
is often used loosely to refer to both a color space system and the
color space on which it is based.
A color space is a set of colors which can be displayed or
recognized by an input or output device (such as a scanner, monitor,
printer, etc.). The colors of a color space are specified as values
in a color space system, which is a coordinate system in which the
individual colors are described by coordinate values on various axes.
Because of the structure of the human eye, there are three axes in
color spaces which are intended for human observers. The practical
application of that is that colors are specified with three
components (with a few exceptions). There are about 30 to 40 color
space systems in use. Some important examples are:
CMYK is a color model
which has components for Cyan, Magenta, Yellow and Black. It is a
subtractive color model, and that fact is important when an image
is printed. It is complementary to the
RGB color model.
The values of the individual colors vary between 0% and 100%,
where 0% corresponds to an unprinted color, and 100% corresponds
to a completely printed area of color. Colors are formed by
mixing the three basic colors.
The last of these values, K (Black), doesn't contribute to
the color, but merely serves to darken the other colors. The
letter K is used for Black to prevent confusion, since B usually
stands for Blue.
This is the mode used in printing. These are the colors in the
ink cartridges in your printer. It is the mode used in painting and
in all the objects around us, where light is reflected, not
emmitted. Objects absorb part of the light waves and we see
only the reflected part. Note that the cones in our eyes
see this reflected light in RGB mode. An object appears Red
because Green and Blue have been absorbed. Since the
combination of Green and Blue is Cyan, Cyan is absorbed when
you add Red. Conversely, if you add Cyan, its complementary
color, Red, is absorbed. This system is
subtractive. If you add Yellow, you
decrease Blue, and if you add Magenta, you decrease Green.
It would be logical to think that by mixing Cyan, Magenta and
Yellow, you would subtract Red, Green and Blue, and the eye
would see no light at all, that is, Black. But the
question is more complex. In fact, you would see a dark brown.
That is why this mode also has a Black value, and why your
printer has a Black cartridge. It is less
expensive that way. The printer doesn't have to mix the other
three colors to create an imperfect Black, it just has to add
Dithering is a technique used in computer graphics to create the
illusion of more colors when displaying an image which has a low
color depth. In a
dithered image, the missing colors are reproduced by a certain
arrangement of pixels in the available colors. The human eye
perceives this as a mixture of the individual colors.
The Gradient tool uses
dithering. You may also choose to use dithering when you convert an
image to Indexed
format. If you are working on an image with indexed colors, some
tools (such as the pattern fill tool) may also use dithering, if the
correct color is not available in the colormap.
The Newsprint filter
uses dithering as well. You can use the
NL Filter (Non Linear filter)
to remove unwanted dithering noise from your image.
Also note that although GIMP itself uses 24-bit
colors, your system may not actually be able to display that many
colors. If it doesn't, then the software in between
GIMP and your system may also dither colors while
A file format or file type is the form in which computer data is
stored. Since a file is stored by an operating system as a linear
series of bytes, which cannot describe many kinds of real data in
an obvious way, conventions have been developed for interpreting
the information as representations of complex data. All of the
conventions for a particular “kind” of file constitute
a file format.
Some typical file formats for saving images are JPEG, TIFF, PNG and
GIF. The best file format for saving an image depends upon how the
image is intended to be used. For example, if the image is intended
for the internet, file size is a very important factor, and if the
image is intended to be printed, high resolution and quality have
GIMP uses the process of Feathering to make a
smooth transition between a region and the background by softly
blending the edges of the region.
In GIMP, you can feather the edges of a
selection. Brushes can also have feathered edges.
A floating selection (sometimes called a “floating layer”)
is a type of temporary layer which is similar in function to a
normal layer, except that a floating selection must be anchored
before you can resume working on any other layers in the image. You
can use various operations to change the image data on the floating
selection. There can only be one floating selection in an image at a
You can anchor a floating selection in various ways. First, you can
create a New layer. If
you create a new layer while there is a floating selection, the
floating selection is anchored to it. Second, you can anchor the
floating selection to the current layer, which is just below the
floating selection. To do this, click anywhere on the image except
on the floating selection. This merges the floating selection with
the layer below it. You can also anchor the floating selection to
the layer below it by clicking on the anchor button of the Layers
dialog or using the
Anchor layer command.
There are also various ways to create a floating selection. The
first is to convert an existing selection into a floating selection
with the Float command.
The “paste” operations,
Paste Named Buffer,
Paste Into, also
create a floating selection. In addition, the Transform tools,
create a floating selection when they are used on a selection,
rather than a layer. When the Affect mode is
Transform Layer and a selection already exists,
these tools transform the selection and create a floating selection
with the result. If a selection does not exist, they transform the
current layer and do not create a floating selection. (If the
Affect mode is
Transform Selection, they also do not create a
floating selection.) You can also create a floating selection by
clicking on a selection and dragging it.
Floating selections are a rest of the time when GIMP did not use
layers. They have no practical use, but you must know what you
have to do with them.
Floyd-Steinberg dithering is a method of
dithering which was first
published in 1976 by Robert W. Floyd and Louis Steinberg. The
dithering process begins in the upper left corner of the image. For
each pixel, the closest available color in the palette is chosen and
the difference between that color and the original color is computed
in each RGB channel. Then specific fractions of these differences
are dispersed among several adjacent pixels which haven't yet been
visited (below and to the right of the original pixel). Because of
the order of processing, the procedure can be done in a single pass
over the image.
When you convert an image to
Indexed mode, you
can choose between two variants of Floyd-Steinberg dithering.
Gamma or gamma correction is a non-linear operation which is used to
encode and decode luminance or color values in video or still image
systems. It is used in many types of imaging systems to straighten out
a curved signal-to-light or intensity-to-signal response. For example,
the light emitted by a CRT is not linear with regard to its input
voltage, and the voltage from an electric camera is not linear with
regard to the intensity (power) of the light in the scene. Gamma
encoding helps to map the data into a perceptually linear domain, so
that the limited signal range (the limited number of bits in each RGB
signal) is better optimized perceptually.
Gamma is used as an exponent (power) in the correction equation.
Gamma compression (where gamma < 1) is used to encode linear
luminance or RGB values into color signals or digital file values,
and gamma expansion (where gamma > 1) is the decoding process,
and usually occurs where the current-to-voltage function for a CRT
For PC video, images are encoded with a gamma of about 0.45 and
decoded with a gamma of 2.2. For Mac systems, images are typically
encoded with a gamma of about 0.55 and decoded with a gamma of 1.8.
The sRGB color space standard used for most cameras, PCs and printers
does not use a simple exponential equation, but has a decoding gamma
value near 2.2 over much of its range.
In GIMP, gamma is an option used in the brush tab
of the GIMPressionist
filter and in the Flame filter.
The display filters
also include a Gamma filter. Also see the
Levels Tool, where you can
use the middle slider to change the gamma value.
GIF™ stands for Graphics Interchange Format.
It is a file format
with good, lossless compression for images with low
(up to 256 different colors per image). Since GIF was developed, a
new format called Portable Network Graphics (PNG)
has been developed, which is better than GIF in all respects, with
the exception of animations and some rarely-used features.
GIF was introduced by CompuServe in 1987. It became popular mostly
because of its efficient, LZW compression. The size of the image
files required clearly less disk space than other usual graphics
formats of the time, such as PCX or MacPaint. Even large images could
be transmitted in a reasonable time, even with slow modems. In
addition, the open licensing policy of CompuServe made it possible
for any programmer to implement the GIF format for his own
applications free of charge, as long as the CompuServe copyright
notice was attached to them.
Colors in GIF are stored in a color table which can hold up to 256
different entries, chosen from 16.7 million different color values.
When the image format was introduced, this was not a much of a
limitation, since only a few people had hardware which could display
more colors than that. For typical drawings, cartoons,
black-and-white photographs and similar uses, 256 colors are quite
sufficient as a rule, even today. For more complex images, such as
color photgraphs, however, a huge loss of quality is apparent, which
is why the format is not considered to be suitable for those purposes.
One color entry in the palette can be defined to be transparent.
With transparency, the GIF image can look like it is non-rectangular
in shape. However, semi-transparency, as in
PNG, is not possible.
A pixel can only be either entirely visible or completely
The first version of GIF was 87a. In 1989, CompuServe published an
expanded version, called 89a. Among other things, this made it
possible to save several images in one GIF file, which is especially
used for simple animation. The version number can be distinguished
from the first six bytes of a GIF file. Interpreted as ASCII symbols,
they are “GIF87a” or “GIF89a”.
The GNU project was started in 1983 by Richard Stallman with the
goal of developing a completely free operating system. It is
especially well-known from the GNU General Public License (GPL) and
GNU/Linux, a GNU-variant with a Linux kernel.
The name came about from the naming conventions which were in
practice at MIT, where Stallman worked at the time.
For programs which were similar to other programs, recursive
acronyms were chosen as names. Since the new system was to be based
on the widespread operating system, Unix, Stallman looked for that
kind of name and came up with GNU, which stands for
“GNU is not Unix”. In order to avoid confusion, the
name should be pronounced with the “G”, not like
“new”. There were several reasons for making GNU
Unix-compatible. For one thing, Stallman was convinced that most
companies would refuse a completely new operating system, if the
programs they used wouldn't run on it. In addition, the architecture
of Unix made quick, easy and distributed development possible,
since Unix consists of many small programs that can be developed
independently of each other, for the most part. Also, many parts of
a Unix system were freely available to anyone and could therefore
be directly integrated into GNU, for example, the typesetting
system, TeX, or the X Window System. The missing parts were newly
written from the ground up.
GIMP (GNU Image Manipulation Program) is an
official GNU application.
Grayscale is a mode for encoding the colors of an image which
contains only black, white and shades of gray.
When you create a new image, you can choose to create it in
Grayscale mode (which you can colorize later, by changing it to RGB
mode). You can also change an existing image to grayscale by using
Mixer,although not all formats will accept these changes.
Although you can create images in Grayscale mode and convert images
to it, it is not a color model, in the true sense of the word.
As explained in RGB mode, 24-bit
GIMP images can have up to 256 levels of gray. If
you change from Grayscale to RGB mode, your image will have an RGB
structure with three color channels, but of course, it will still be
Grayscale image files (8-bit) are smaller than RGB files.
Guides are lines you can temporarily display on an image while you
are working on it. You can display as many guides as you would like,
in either the horizontal or the vertical direction. These lines help
you position a selection or a layer on the image. They do not appear
when the image is printed.
To create a guide, simply click on one of the rulers in the image
window and pull out a guide, while holding the
mouse button pressed. The guide is then
displayed as a blue, dashed line, which follows the pointer. As soon
as you create a guide, the “Move” tool is activated and
the mouse pointer changes to the Move icon.
The behavior of the guides depends upon the
Affect mode of the “Move” tool.
When Transform Layer mode is selected, the
mouse pointer turns into a small hand as soon as it gets close to a
guide. Then the guide is activated and it turns red, and you can
move the guide or delete it by moving it back into the ruler. If
Transform Selection mode is selected, you can
position a guide, but you cannot move it after that.
To make it easier for you to position image elements, you can
“magnetize” the guides with the
Snap to Guides
command. You can remove the guides with the
Remove all guides
command. You can enable and disable displaying the guides without
removing them by using the
Show Guides command.
A hex triplet is a way of encoding a color for a computer. The
“#” symbol indicates that the numbers which follow it
are encoded in hexadecimal. Each color is specified in two
hexadecimal digits which make up a triplet (three pairs) of
hexadecimal values in the form “#rrggbb”, where
“rr” represents red, “gg” represents green
and “bb” represents blue.
In digital image processing, a histogram is a graph representing the
statistical frequency of the gray values or the color values in an
image. The histogram of an image tells you about the occurrence of
gray values or color values, as well as the contrast range and
the brightness of the image. In a color image, you can create one
histogram with information about all possible colors, or three
histograms for the individual color channels. The latter makes the
most sense, since most procedures are based on grayscale images and
therefore further processing is immediately possible.
HSV is a color model
which has components for Hue (the color, such as blue or red),
Saturation (how strong the color is) and Value (the brightness).
The RGB mode is very well suited to computer screens, but it doesn't
let us describe what we see in everyday life; a light green, a
pale pink, a dazzling red, etc. The HSV model takes these
characteristics into account. HSV and RGB are not completely
independent of each other. You can see that with the Color Picker
tool; when you change a color in one of the color models, the other
one also changes. Brave souls can read
Grokking the GIMP, which explains their
This is the color itself, which results from the
combination of primary colors. All shades (except for the
gray levels) are represented in a chromatic
circle: yellow, blue, and also purple, orange,
etc. The chromatic circle (or “color wheel”)
values range between 0° and 360°. (The term
“color” is often used instead of
“Hue”. The RGB colors are
This value describes how pale the color is. A completely
unsaturated color is a shade of gray. As the saturation
increases, the color becomes a pastel shade. A completely
saturated color is pure. Saturation values go from 0 to
100, from white to the purest color.
This value describes the luminosity, the luminous intensity.
It is the amount of light emitted by a color. You can see
a change of luminosity when a colored object is moved from being
in the shadow to being in the sun, or when you increase the
luminosity of your screen. Values go from 0 to 100. Pixel
values in the three channels are also luminosities:
“Value” in the HSV color model is the vector
sum of these elementary values in the RGB space.
An image hose in GIMP is a special type of brush
which consists of several images. For example, you could have a
brush with footprints, which consists of two images, one for the
left footprint and one for the right. While painting with this
brush, a left footprint would appear first, then a right footprint,
then a left one, etc. This type of brush is very powerful.
An image hose is also sometimes called an “image pipe”
or “animated brush”. An image hose is indicated in the
Brushes dialog by a small red triangle in the lower right corner of
the brush's symbol.
Incremental mode is a paint mode where each brush stroke is drawn
directly on the active layer. When it is set, each additional stroke
of the brush increases the effect of the brush, up to the maximum
opacity for the brush.
If incremental mode is not set, brush strokes are drawn on a canvas
buffer, which is then combined with the active layer. The maximum
effect of a brush is then determined by the opacity, and stroking
with the brush repeatedly does not increase the effect beyond this
The two images above were created using a brush with spacing set to
60 pixels. The image on the left shows non-incremental painting and
the image on the right shows the difference with incremental painting.
Incremental mode is a tool option that is shared by several brush
tools, except those which have a “rate” control, which
automatically implies an incremental effect. You can set it by
checking the Incremental checkbox in the
toolbox for the tool (Paintbrush, Pencil and Eraser).
Indexed color mode is a mode for encoding colors in an image where
each pixel in the image is assigned an 8-bit color number. The color
which corresponds to this number is then put in a table (the palette).
Changing a color in the palette changes all the pixels which refer
to this palette color. Although you can create images in
Indexed Color mode and can transform images to
it, it is, strictly speaking, not a
Interpolation means calculating intermediate values. When you
enlarge (“digitally zoom”) or otherwise transform
(rotate, shear or give perspective to) a digital image,
interpolation procedures are used to compute the colors of the
pixels in the transformed image. GIMP offers
three interpolation methods, which differ in quality and speed. In
general, the better the quality, the more time the interpolation
takes. The methods are:
None (sometimes called “Nearest
Neighbor”): The color of each pixel is copied from its
closest neighboring pixel in the original image. This often
results in aliasing (the “stair-step” effect) and a
coarse image, but it is the fastest method.
Linear (sometimes called
“Bilinear”): The color of each pixel is computed as
the average color of the four closest pixels in the original
image. This gives a satisfactory result for most images and is a
good compromise between speed and quality.
Cubic (sometimes called
“Bicubic”): The color of each pixel is computed as
the average color of the eight closest pixels in the original
image. This usually gives the best result, but it naturally takes
JPEG is a file format
which supports compression and works at all color depths. The
image compression is adjustable, but beware: Too high a compression
could severely reduce image quality, since JPEG compression is lossy.
Use JPEG to create web graphics or if you don't want your
image to take up a lot of space. JPEG is a good format for
photographs and for computer-generated images (CGI). It is not well
digital line drawings (for example, screenshots or vector graphics),
in which there are many neighboring pixels with the same
color values, few colors and hard edges,
Black and white images (only black and white, one bit per pixel) or
half-toned images (newsprint).
Other formats, such as GIF, PNG or JBIG, are far better for these
kinds of images.
In general, JPEG transformations are not reversible. Opening and
then saving a JPEG file causes a new, lossy compression. Increasing
the quality factor later will not bring back the image information
which was lost.
JPEG files usually have an extension .jpg, .JPG, or .jpeg. It is a
very widely used format, because it compresses images very
efficiently, while minimizing the loss of image quality. No other
format comes close to achieving the same level of compression. It does
not, however, support transparency or multiple layers. For this
reason, saving images as JPEG often requires them to be exported
The JPEG Save dialog
When you save a file in JPEG format, a dialog is displayed that allows you
to set the Quality level, which ranges from 1 to 100. Values above 95
are generally not useful, though. The default quality of 85 usually
produces excellent results, but in many cases it is possible to set
the quality substantially lower without noticably degrading the image.
You can test the effect of different quality settings by checking
Show Preview in image window
in the JPEG dialog. Checking this option
causes each change in quality (or any other JPEG parameter) to be
shown in the image display. (This does not alter the image,
though: it reverts back to its original state when the JPEG dialog is
The JPEG algorithm is quite complex, and involves a bewildering number
of options, whose meaning is beyond the scope of this documentation.
Unless you are a JPEG expert, the Quality parameter is probably the
only one you will need to adjust.
After you save an image as a JPEG file, the image is no longer
considered “dirty” by GIMP, so
unless you make further changes to
it, you will not receive any warning if you close it. Because JPEG
is lossy and does not support transparency or multiple layers, some
of the information in the image might then be lost. If you want to
save all of the information in an image, use
JPEG files from many digital cameras contain extra information,
called EXIF data, which specifies camera settings and other
information concerning the circumstances under which the image was
created. GIMP's ability to handle EXIF data
depends on whether the “libexif” library is available
on your system; it is not automatically packaged with
GIMP. If GIMP was built with
libexif support, then EXIF data is preserved if you open a JPEG
file, work with the resulting image, and then save it as JPEG. The
EXIF data is not altered in any way when you do this (which means
that some of its fields are no longer valid). If
GIMP was not built with EXIF support, this does
not prevent files with EXIF data from being opened, but it does mean
that the EXIF data will not be present when the resulting image is
Some information about the advanced settings:
DCT is “discrete cosine transform”, and it is the
first step in the JPEG algorithm going from the spatial to
the frequency domain. The choices are “float”,
“integer” (the default), and
“fast integer”. The float method is very slightly
more accurate than the integer method, but is much slower
unless your machine has very fast floating-point hardware.
Also note that the results of the floating-point method may
vary slightly across machines, while the integer methods should
give the same results everywhere. The fast integer method is
much less accurate than the other two.
The Lab color space (also called the L*a*b color space) is a
developed in the beginning of the 1930s by the Commission
Internationale d`Eclairage (CIE). It includes all the colors that
the human eye can perceive. That contains the colors of the
RGB and the CMYK color spaces, among others. In Lab, a color is
indicated by three values: L, a and b. Here, the L stands for the
luminance component — corresponding to the gray value — and a and b
represent the red-green and blue-yellow parts of the color,
In contrast to RGB or CMYK, Lab is not dependent upon the
various input and output devices. For that reason, it is used as an
exchange format between devices. Lab is also the internal color
model of PostScript Level II.
You can think of layers as being a stack of slides which are more or
less transparent. Each layer represents an aspect of the image and
the image is the sum of all of these aspects. The layer at the bottom
of the stack is the background layer. The layers above it are the
components of the foreground.
You can view and manage the layers of the image through the
Representation of an image with layers:
The final image:
Marching ants is a term which describes the dotted line which
surrounds a selection. The line is animated, so it looks as if
little ants are running around behind each other.
You can disable the marching ants by unchecking the
option or by using the keyboard shortcut
GIMP has twenty-one layer modes. Layer modes are
also sometimes called “blending modes”. Selecting a
layer mode changes the appearance of the layer or image, based on
the layer or layers beneath it. If there is only one layer, the
layer mode has no effect. There must therefore be at least two
layers in the image to be able to use layer modes.
You can set the layer mode in the Mode menu in
the Layers dialog. GIMP uses the layer mode to
determine how to combine each pixel in the top layer with the pixel
in the same location in the layer below it.
There is a drop-down list in the Toolbox options box which
contains modes that affect the painting tools in a similar way
to the layer modes. You can use all of the same modes for
painting that are available for layers, and there are two
additional modes just for the painting tools, which are
described here at the end of the list of layer modes.
Layer modes permit complex color changes in the image. They are
often used with a new layer which acts as a kind of mask. For
example, if you put a solid white layer over an image and set the
layer mode of the new layer to “Saturation”, the
underlying visible layers will appear in shades of gray.
In the descriptions of the layer modes below, the equations are also
shown. This is for those who are curious about the mathematics of the
layer modes. You do not need to understand the equations in order to
use the layer modes effectively, however.
The equations are in a shorthand notation. For example, the equation
R = T + B
means, “For each pixel in the upper and lower layer, add each
of the corresponding color components together to form the resulting
pixel's color.” Pixel color components must always be between
0 and 255. Unless the description below says otherwise, a negative
color component is set to 0 and a color component larger than 255 is
set to 255.
The examples below show the effects of each of the modes. The image
on the left is the normal state and the image on the right shows the
results of the layer mode. In this image, Wilber is on the upper
layer, surrounded by transparency, and the lower layer is solid
Since the results of each mode vary greatly depending upon the
colors on the layers, these images can only give you a general idea
of how the modes work. You are encouraged to try them out yourself.
You might start with two similar layers, where one is a copy of the
other, but slightly modified (by being blurred, moved, rotated,
scaled, color-inverted, etc.) and seeing what happens with the layer
Normal mode is the default layer mode.
The layer on top covers the layers below it. If you want to
see anything below the top layer when you use this mode, the
layer must have some transparent areas.
The equation is:
R = T
Dissolve mode dissolves the upper layer
into the layer beneath it by drawing a random pattern of
pixels in areas of partial transparency. It is useful as a
layer mode, but it is also often useful as a painting mode.
This is especially visible along the edges within an image. It
is easiest to see in an enlarged screenshot. The image on the
left illustrates “Normal” layer mode (enlarged)
and the image on the right shows the same two layers in
“Dissolve” mode, where it can be clearly seen how
the pixels are dispersed.
Multiply mode multiplies the pixel values
of the upper layer with those of the layer below it and then
divides the result by 255. The result is usually a darker
image. If either layer is white, the resulting image is the
same as the other layer. If either layer is black, the
resulting image is completely black.
The equation is:
R = T × B ÷ 255
The mode is commutative; the order of the two layers doesn't
Divide mode multiplies each pixel value
in the lower layer by 256 and then divides that by the
corresponding pixel value of the upper layer plus one. (Adding
one to the denominator avoids dividing by zero.) The resulting
image is often lighter, and sometimes looks
The equation is:
R = B × 256 ÷ (T + 1)
Screen mode inverts the values of each of
the visible pixels in the two layers of the image. (That is, it
subtracts each of them from 255.) Then it multiplies them
together, inverts this value again and divides by 255. The
resulting image is usually brighter, and sometimes
“washed out” in appearance. The exceptions to
this are a black layer, which does not change the other layer,
and a white layer, which results in a white image. Darker
colors in the image appear to be more transparent.
The equation is:
R = (255 - ((255 - B) × (255 - T))) ÷ 255
The mode is commutative; the order of the two layers doesn't
Ce mode est commutatif: l'ordre des calques ne compte pas.
Overlay mode inverts the pixel value of
the lower layer, multiplies it by two times the pixel value of
the upper layer, adds that to the original pixel value of the
lower layer, divides by 255, and then multiplies by the pixel
value of the original lower layer and divides by 255 again. It
darkens the image, but not as much as with
The equation is:
R = B × (B + (2 × T × (255 - B)) ÷ 255) ÷ 255
Dodge mode multiplies the pixel value of
the lower layer by 256, then divides that by the inverse of
the pixel value of the top layer. The resulting image is
usually lighter, but some colors may be inverted.
The equation is:
R = B × 256 ÷ (256 - T)
In photography, dodging is a technique used in a darkroom to
increase the exposure in particular areas of the image. This
brings out details in the shadows. When used for this purpose,
dodge may work best on Grayscale images and with a painting
tool, rather than as a layer mode.
Burn mode inverts the pixel value of the
lower layer, multiplies it by 256, divides that by one plus
the pixel value of the upper layer, then inverts the result.
It tends to make the image darker, somewhat similar to
The equation is:
R = 255 - ((255 - B) * 256) ÷ (T + 1)
In photography, burning is a technique used in a darkroom to
decrease the exposure in particular areas of the image. This
brings out details in the highlights. When used for this
purpose, burn may work best on Grayscale images and with a
painting tool, rather than as a layer mode.
Hard Light mode is rather complicated
because the equation consists of two parts, one for darker
colors and one for brighter colors. If the pixel color of the
upper layer is greater than 128, the layers are combined
according to the first formula shown below. Otherwise, the pixel
values of the upper and lower layers are multiplied together
and multiplied by two, then divided by 256. You might use this
mode to combine two photographs and obtain bright colors and
The equation is:
if (T > 128)
R = 255 - ((255 - B) × (255 - (2 × (T - 128))) ÷ 256)
R = (B × T × 2) ÷ 256
Soft Light is not related to
“Hard Light” in anything but the name, but it
does tend to make the edges softer and the colors not so
bright. It is similar to “Overlay” mode. In some
versions of GIMP, “Overlay”
mode and “Soft Light” mode are identical.
The equation is:
Rs = (255 - ((255 - B) × (255 - T))) ÷ 255 (Screen mode)
R = ((255 - B) × T × B) + (B × Rs)) ÷ 255
Grain Extract mode is supposed to extract
the “film grain” from a layer to produce a new
layer that is pure grain, but it can also be useful for giving
images an embossed appearance. It subtracts the pixel value
of the upper layer from that of the lower layer and adds 128.
The equation is:
R = B - T + 128
Grain Merge mode merges a grain layer
(possibly one created from the “Grain Extract”
mode) into the current layer, leaving a grainy version of the
original layer. It does just the opposite of
“Grain Extract”. It adds the pixel values of the
upper and lower layers together and subtracts 128.
The equation is:
R = B + T - 128
Difference mode subtracts the pixel value
of the upper layer from that of the lower layer and then takes
the absolute value of the result. No matter what the original
two layers look like, the result looks rather odd. You can
use it to invert elements of an image.
The equation is:
R = | B - T |
The mode is commutative; the order of the two layers doesn't
Addition mode is very simple. The pixel
values of the upper and lower layers are added to each other.
The resulting image is usually lighter. The equation can
result in color values greater than 255, so some of the light
colors may be set to the maximum value of 255.
The equation is:
R = T + B
The mode is commutative; the order of the two layers doesn't
Subtract mode subtracts the pixel values
of the upper layer from the pixel values of the lower layer.
The resulting image is normally darker. You might get a lot of
black or near-black in the resulting image. The equation can
result in negative color values, so some of the dark colors
may be set to the minimum value of 0.
The equation is:
R = B - T
Darken Only mode compares each
component of each pixel in the upper layer with the
corresponding one in the lower layer and uses the smaller
value in the resulting image. Completely white layers have no
effect on the final image and completely black layers
result in a black image.
The equation is:
if (B < T)
R = B
else R = T
The mode is commutative; the order of the two layers doesn't
Lighten Only mode compares each
component of each pixel in the upper layer with the
corresponding one in the lower layer and uses the larger value
in the resulting image. Completely black layers have no effect
on the final image and completely white layers result in a
The equation is:
if (B > T)
R = B
else R = T
The mode is commutative; the order of the two layers doesn't
Hue mode uses the hue of the upper layer
and the saturation and value of the lower layer to form the
resulting image. However, if the saturation of the upper layer
is zero, the hue is taken from the lower layer, too.
Saturation mode uses the saturation of
the upper layer and the hue and value of the lower layer to
form the resulting image.
Color mode uses the hue and saturation
of the upper layer and the value of the lower layer to form
the resulting image.
Value mode uses the value of the upper
layer and the saturation and hue of the lower layer to form
the resulting image. You can use this mode to reveal details
in dark and light areas of an image without changing the
Each layer in an image can have a different layer mode. (Of course,
the layer mode of the bottom layer of an image has no effect.) The
effects of these layer modes are cumulative. The image shown below
has three layers. The top layer consists of Wilber surrounded by
transparency and has a layer mode of “Difference”. The
second layer is solid light blue and has a layer mode of
“Addition”. The bottom layer is filled with the
“Red Cubes” pattern.
GIMP also has similar modes which are used for
the painting tools. These are the same twenty-one modes as the layer
modes, plus an additional two modes which are specific to the
painting tools. You can set these modes from the
Mode menu in the Tools option dialog. In the
equations shown above, the layer you are painting on is the
“lower layer” and the pixels painted by the tool are
the “upper layer”. Naturally, you do not need more than
one layer in the image to use these modes, since they only operate
on the current layer and the selected painting tool. The two
additional painting modes are described here.
The Behind mode is only available from
the Toolbox options, not as a layer mode from the Layers
dialog. When you paint with a tool in “Behind”
mode, it paints behind objects that are
already painted on the layer. That means that this mode only
makes sense when you are painting on a layer that has
transparent areas, otherwise you wouldn't be able to see any
difference in the resulting image.
In the example image, Wilber is on the top layer, surrounded
by transparency. The lower layer is solid light blue. The
Bucket Fill tool was used, with an Affect
mode of “Selection”, and the entire layer was
selected. A pattern was used to paint with the Bucket Fill
The Color Erase mode is only available
from the Toolbox options, not as a layer mode from the Layers
dialog. When you paint with a tool in
“Color Erase” mode, it finds areas in the layer
which have the current painting color and erases them, turning
those areas transparent. Like “Behind” mode, the
layer must have a layer mask (so that transparency is allowed)
for an effect to be seen.
In the example image, the color of the Bucket Fill tool was
white, so white parts of Wilber were erased and the blue
background shows through.
Masks are special elements associated with a layer or a selection,
which determine the transparency of the layer or selection.
There are two types of masks:
Every layer can have its own mask. The layer mask represents
the Alpha channel of the layer and allows you to manage its
transparency. By painting on the layer mask, you can make parts
of the layer opaque or transparent: painting with black makes
the layer transparent, painting with white makes the layer
opaque and painting with shades of gray makes the layer
semi-transparent. You can use all of the tools to paint on the
mask. You can use the Layer mask for transition effects, volume
effects, merging elements from another image, etc. See the
Layer Mask section
for more details.
Channel Mask, also called
Channel Masks determine the transparency of a selection. By
painting on a Channel Mask with white, you remove the mask and
increase the selection; with black, you reduce the selection.
This procedure lets you create a selection very precisely. You
can also save your selections to a Channel Mask with the
Save to Channel
command. You can retrieve it later by using the
“Channel to selection” command from the
Channel menu. Channel
masks are so important in GIMP that a
special type has been implemented: the
Quick mask. See the
section for more details.
The moiré effect (pronounce “Moa-ray”)is an unintended
pattern which appears when a regular pattern of grids or lines
interferes with another regular pattern placed over it. This can
happen, for example, when you are scanning an image with a periodic
structure (such as a checkered shirt or a half-toned image), scanning
a digital image, taking a digital photograph of a periodic pattern,
or even when silkscreening.
If you discover the problem in time, the best solution is to move
the original image a little bit in the scanner or to change the
camera angle slightly.
If you cannot re-create the image file, GIMP
offers some filters which may help you with the problem. For more
information, see the
NL Filter (Non-Linear)
A Parasite is additional data which may be written to an XCF file.
A parasite is identified by a name, and can be thought of as an
extension to the other information in an XCF file.
Parasites of an image component may be read by
GIMP plug-ins. Plug-ins may also define
their own parasite names, which are ignored by other plug-ins.
Examples of parasites are comments, the save options for the TIFF,
JPEG and PNG file formats, the gamma value the image was created
with and EXIF data.
A Path is a contour composed of straight lines, curves, or both. In
GIMP, it is used to form the boundary of a
selection, or to be stroked to create visible
marks on an image. Unless a path is stroked, it is not visible when
the image is printed and it is not saved when the image is written
to a file (unless you use XCF format).
All of the functions which GIMP and its
extensions make available are registered in the Procedure Database
(PDB). Developers can look up useful programming information about
these functions in the PDB by using the
PDF (Portable Document Format) is a
file format which was
developed by Adobe to address some of the deficiencies of
PostScript. Most importantly, PDF files tend to be much smaller than
equivalent PostScript files. As with PostScript,
GIMP's support of the PDF format is through the
free Ghostscript libraries.
A pixel is a single dot, or “picture element”, of an
image. A rectangular image may be composed of thousands of pixels,
each representing the color of the image at a given location. The
value of a pixel typically consists of several
Channels, such as the Red,
Green and Blue components of its color, and sometimes its Alpha
Created by Adobe, PostScript is a page description language mainly
used by printers and other output devices. It's also an excellent way
to distribute documents. GIMP does not support
PostScript directly: it depends on a powerful free software program
The great power of PostScript is its ability to represent vector
graphics—lines, curves, text, paths, etc.—in a resolution-independent
way. PostScript is not very efficient, though, when it comes to
representing pixel-based raster graphics. For this reason, PostScript
is not a good format to use for saving images that are later going to
be edited using GIMP or another graphics program.
Linux distributions almost always come with Ghostscript already
installed (not necessarily the most recent version). For other
operating systems, you may have to install it yourself. Here are
instructions for installing it on Windows:
Look for the package gnu-gs or ghostscript (for non-commercial
use only) and go to the download section.
Download one of the prepared Windows distributions, such as
gs650w32.exe or gs700w32.exe.
Start the executable and follow the instructions of the
Copy the executable gswin32c.exe from the bin directory of the
Ghostscript installation to the Windows directory (or any other
directory that is contained in the PATH). As an alternative,
advanced users can set an environment variable, GS_PROG, to point
to gswin32c.exe (e.g. C:\gs\gsX.YY\bin\gswin32c.exe).
Now you should be able to read PostScript files with
GIMP. Please note
that you must not move the Ghostscript directories once the
installation is complete. The installation creates registry entries
which allow Ghostscript to find its libraries. (These instructions
courtesy of http://www.kirchgessner.net.)
PNG is the acronym of “Portable Network Graphic”
(pronounce “ping”. This recent format offers many
advantages ans a few drawbacks: it is not lossy and gives files
more heavy than the JPEG format, but it is perfect for saving your
images because you can save them several times without losing
data each time (it is used for this Help). It supports True Colors
(several millions of colors), indexed images (256 colors like GIF),
and 256 transparency levels (while GIF supports only two levels).
Unfortunately, Microsoft Internet Explorer recognizes only two
The “Save as PNG” dialog
Interlacing: When this option is checked,
the image is progressively displayed on a Web page. So, slow
computer users can stop downloading if they are not interested.
Save background color: If your image has
many transparency levels, the Internet browsers which
recognize only two levels, will use the background color of
your Toolbox instead. But Internet Explorer does not use these
Save gamma: informations about your
monitor will be saved, so that the image will be displayed in
the same way on other computers, provided that the display
program supports these informations, what is rarely the case.
Save layer offset: No interest. Images
with layers are flattenned before saving to PNG and layer
offset is taken in account.
Save Resolution: Save the image
resolution, in dpi (dot per inch).
Save creation time: That will be the date
of last saving.
Save comment: you can read this comment
in the Info-window.
Save color values from transparent pixels:
No intrest. If Internet Explorer would take this option in
account, it would replace semi-transparent pixels by the color
value of these pixels.
Compression level: Since compression is
not lossy, the only reason to use a compression level less
than 9 would be a too long time to compress file on a slow
computer. Nothing to fear from decompression: it is as quick
whatever the compression level.
Save defaults: If you click on this
button, your settings will be saved and can be used by other
savings by clicking on the Load defaults.
Since PNG format supports indexed images, you have better
reduce the number of colors before saving if you want to have
the lightest file for the Web. See
Computers work on 8 bits blocks named “Byte”. A
byte allows 256 colors. Reducing the number of colors below
256 is not useful: a byte will be used anyway and the file size
will not be less. More, this “PNG8” format, like
GIF, uses only one bit for transparency; only two transparency
levels are possible, transparent or opaque.
If you want PNG transparency to be fully displayed by Internet
Explorer, you can use the AlphaImageLoader DirectX filter in the
code of your Web page. See
Microsoft Knowledge Base
PSD is Adobe Photoshop's native
file format, and it is
therefore comparable to XCF in
complexity. GIMP's ability to handle PSD files is
sophisticated but limited: some features of PSD files are not loaded,
and only older versions of PSD are supported. Unfortunately,
Adobe has now made the Photoshop Software Development
Kit — which includes their file format
specifications — proprietary, and only available to a limited
set of developers approved by Adobe. This does not include the
GIMP development team, and the lack of
information makes it very difficult to maintain up-to-date support
for PSD files.
Quantization is the process of reducing the color of a pixel into one
of a number of fixed values by matching the color to the nearest color
in the colormap. Actual pixel values may have far more precision than
the discrete levels which can be displayed by a digital display. If
the display range is too small, then abrupt changes in colors (false
contours, or banding) may appear where the color intensity changes
from one level to another. This is especially noticeable in Indexed
images, which have 256 or fewer discrete colors.
One way to reduce quantization effects is to use
operations in GIMP which perform
dithering are the
(if you have enabled the dithering option) and the
Convert to Indexed
command. However, they only work on RGB images and not on Indexed
Additive color model
RGB is a color model
which has components for Red, Green and Blue. These colors are
emitted by screen elements and not reflected as they are with paint.
The resulting color is a combination of the three primary RGB colors,
with different degrees of lightness. If you look closely at your
television screen, whose pitch is less than that of a computer
screen, you can see the red, green and blue elements lit with
different intensities. The RGB color model is
GIMP uses eight bits per channel for each primary
color. That means there are 256 intensities (Values) available,
resulting in 256×256×256 = 16,777,216 colors.
It is not obvious why a given combination of primary colors produces a
particular color. Why, for instance, does 229R+205G+229B give a shade
of pink? This depends upon the human eye and brain. There is no color
in nature, only a continuous spectrum of wavelengths of light. There
are three kinds of cones in the retina. The same wavelength of light
acting upon the three types of cones stimulates each of them
differently, and the mind has learned, after several million years of
evolution, how to recognize a color from these differences.
It is easy to see that no light (0R+0G+0B) produces complete darkness,
black, and that full light (255R+255G+255B) produces white. Equal
intensity on all color channels produces a level of gray. That is why
there can only be 256 gray levels in GIMP.
Mixing two Primary colors in RGB mode
gives a Secondary color, that is, a
color in the CMY model. Thus combining Red and Green gives
Yellow, Green and Blue give Cyan, Blue and Red give Magenta.
Don't confuse secondary colors with
Complementary colors which are
directly opposite a primary color in the chromatic
Mixing a primary color with its complementary color gives gray (a
It is important to know what happens when you are dealing with colors
The most important rule to remember is that decreasing the intensity
of a primary color results in increasing the intensity of the
complementary color (and vice versa). This is because when you
decrease the value of a channel, for instance Green, you automatically
increase the relative importance of the other two, here Red and Blue.
The combination of these two channels gives the secondary color,
Magenta, which is the complementary color of Green.
You can check this out. Create a new image with only a white
background (255R+255G+255B). Open the
dialog and select the Red channel. If necessary, check the
preview box. Move the white slider to the left to decrease the
Red value. You will notice that the background of your image
gets closer and closer to Cyan. Now, decrease the Blue channel:
only the Green will remain. For practice, go backwards, add a
color and try to guess what hue will appear.
Sample Merged is an option you can set when you use the
Bucket Fill tool, the
Color Picker tool and
various selection tools. It is useful when you are working on an
image with several layers and the active layer is either
semi-transparent or has a
Layer Mode which is not set
to Normal. When you check the Sample Merged option, the color which
is used for the operation is the composite color of all the visible
layers. When the Sample Merged option is not checked, the color used
is the color of the active layer itself.
Supersampling is a more sophisticated antialiasing technique, that
is, a method of reducing jagged and stair-stepped edges along a
slanted or curved line. Samples are taken at several locations
within each pixel, not just at the center, and
an average color is calculated. This is done by rendering the image
at a much higher resolution than the one being displayed and then
shrinking it to the desired size, using the extra pixels for
calculation. The result is a smoother transition from one line of
pixels to another along the edges of objects.
The quality of the result depends on the number of samples.
Supersampling is often performed at a range of 2× to 16× the original
size. It greatly increases the amount of time needed to draw the image
and also the amount of space needed to store the image in memory.
One way to reduce the space and time requirement is to use Adaptive
Supersampling. This method takes advantage of the fact that very few
pixels are actually on an object boundary, so only those pixels need to
be supersampled. At first, only a few samples are taken within a pixel.
If the colors are very similar to each other, only those samples are
used to calculate the final color. If not, more samples are used. This
means that the higher number of samples is calculated only where
necessary, which improves performance.
TGA (TARGA Image File) is a
file format which
supports 8, 16, 24 or 32 bits per pixel and optional RLE compression.
It was originally developed by the Truevision company.
“TGA” stands for Truevision Graphics Adapter and
“TARGA” stands for Truevision Advanced Raster Graphics
TIFF (Tagged Image File Format) is a
file format which was
developed primarily for scanned
raster graphics for color separation. Six different encoding routines
are supported, each with one of three different image modes: black and
white, grayscale and color. Uncompressed TIFF images may be 1, 4, 8 or
24 bits per pixel. TIFF images compressed using the LZW algorithm may
be 6, 8 or 24 bits per pixel. Besides Postscript format, TIFF is one
of the most important formats for preliminary stages of printing. It
is a high quality file format, which is perfect for images you want to
import to other programs like FrameMaker or CorelDRAW.
A Tile is a part of an image which GIMP
currently has open. In order to avoid having to store an
entire image in memory at the same time,
GIMP divides it into smaller pieces.
A tile is usually a square of 64 x 64 pixels, although tiles at
the edges of an image may be smaller than that.
At any time, a tile may be in main memory, in the tile cache
in RAM, or on disk. Tiles which are currently being worked on are
in main memory. Tiles which have been used recently are in RAM.
When the tile cache in RAM is full, tiles which have been used
least recently are written to disk. GIMP
can retrieve the tiles from RAM or disk when they are needed.
Do not confuse these tiles with those in the
URLs (Uniform Resource Locators) are one type of Uniform Resource
Identifiers (URIs). URLs identify a resource by its primary access
mechanism (commonly http or ftp) and the location of the resource in
the computer network. The name of the URI scheme is therefore
generally derived from the network protocol used for it. Examples of
network protocols are http, ftp and mailto.
Since URLs are the first and most common kinds of URIs, the terms
are often used synonymously.
A Uniform Resource Identifier (URI) is a string of characters that
serves to identify an abstract or a physical resource. URIs are used
for the identification of resources in the Internet (such as web
pages, miscellaneous files, calling up web services, and for receivers
of e-mail) and they are especially used in the Worldwide Web.
XCF is a
file format which is
special because it is GIMP's
native file format: that is, it was designed specifically to store all
of the data that goes to make up a GIMP image.
Because of this, XCF files may be quite complicated, and there are
few programs other than GIMP that can read them.
When an image is stored as an XCF file, the file encodes nearly
everything there is to know about the image: the pixel data for
each of the layers, the current selection, additional channels
if there are any, paths if there are any, and guides. The most
important thing that is not saved in an XCF
file is the undo history.
The pixel data in an XCF file is represented
in a lossless compressed form: the image byte blocks are compressed
using the lossless RLE algorithm. This means that no matter how many
times you load and save an image using this format, not a single
pixel or other image data is lost or modified because of this format.
XCF files can become very large, however GIMP
allows you to compress the files themselves, using either the gzip
or bzip2 compression methods, both of which are fast, efficient, and
freely available. Compressing an XCF file will often shrink it by a
factor of 10 or more.
The GIMP developers have made a great effort to
keep the XCF file format compatible across versions. If you create a
file using GIMP 2.0, it ought to be possible to
open the file in GIMP 1.2. However, some of the
information in the file may not be usable: for example,
GIMP 2.0 has a much more sophisticated way of
handling text than GIMP 1.2, so a text layer
from a GIMP 2.0 XCF file will appear as an
ordinary image layer if the file is opened in
YCbCr is a color model
which was developed for the PAL television standard as a simple
modification to the YUV color model. In the
meantime, it has become the CCIR-601 standard for image and video
recording. For example, it is used for JPEG pictures and MPEG videos,
and therefore also on DVDs, video CDs and for most other widespread
digital video standards. Note that a color model is still not a color
space, since it doesn't determine which colors are actually meant by
“red”, “green” and “blue”.
For a color space, there must still be a reference to a specific
absolute color value.
There are color models which do not express a color by the additive
basic colors, red, green and blue (RGB), but by other properties, for
example, the brightness-color model. Here, the criteria are the basic
brightness of the colors (from black, through gray, to white), the
colors with the largest portion (red, orange, yellow, green, blue,
violet, or other pure colors that lie between them) and the saturation
of the colors (“gaudy” to pale). This color model is
based on the
ability of the eye to recognize small differences in luminosity
better than small color differences, and to recognize those better
than small differences in saturation. That makes gray text written on
a black background easy to read, but blue text on a red background
very hard to read, even with the same basic brightness. Such color
models are called brightness-color models.
The YCbCr model is a slight adaptation of such a brightness-color
model. An RBG color value is divided into a basic brightness, Y, and
two components, Cb and Cr, where Cb is a measurement of the deviation
from gray in the blue direction, or if it is less that 0.5, in the
direction of yellow. Cr is the corresponding measurement for the
difference in the direction of red or turquoise. This representation
uses the peculiarity of the eye of being especially sensitive to green
light. That is why most of the information about the proportion of
green is in the basic brightness, Y, an only the deviations for the
red and blue portions need to be represented. The Y values have twice
the resolution of the other two values, Cb and Cr, in most practical
applications, such as on DVDs.
YUV is a color model
which uses two components to represent the color
information, luma (the strength of the light per area) and the
chrominance, or proportion of color (chroma), where the chrominance
again consists of two components. The development of the YUV color
model also goes back to the development of color television (PAL),
where ways were sought for transmitting the color information
along with the black-and-white signal, in order to achieve backwards
compatibility with old black and white televisions without having to
increase the available transmission bandwidth. From the YUV color
model of the analog television techiques, the YCrCb color model was
developed, which is used for most kinds of digital image and video
compression. Erroneously, the YUV color model is also often spoken
about in those fields, although the YCbCr model is actually used.
This often causes confusion.
For the calculation of the luma signals, the underlying RGB data is
first adjusted with the gamma
value of the output device, and an R'G'B' signal is obtained. The
three individual components are added together with different
weights, to form the brightness information, which also functions as
the VBS signal (Video Baseband Signal, the black-and-white signal)
for the old black and white televisions.
The exact calculation is more complicated, however, since some aspects
of the color perception of the human eye have to be taken into
account. For example, green is perceived to be lighter than red, and
this is perceived to be lighter than blue. Furthermore, in some
systems gamma correction of the basic color is first performed.
The chrominance signals, and the color difference signals also,
contain the color information. They are formed by the difference of
blue minus luma or red minus luma.
From the three generated components, Y, U and V, the individual
color proportions of the basic color can be calculated again later:
Y + U = Y + ( B - Y ) = Y - Y + B = B
Y + V = Y + ( R - Y ) = Y - Y + R = R
Y - B - R = ( R + G + B ) - B - R = G
Furthermore, because of the structure of the retina of the human
eye, it turns out that the brightness information is perceived at a
higher resolution than the color, so that many formats based on
the YUV color model compress the chrominance to save bandwidth