gimp/docs/gimp.txt

594 lines
31 KiB
Plaintext

GIMP Documentation
outline of basic information for manual/tutorial
Main sections:
Introduction
what is the gimp?
a brief history -> current release
what is it used for?
what isn't it used for?
Setup
acquisition
ftp sites: main, alternatives (sunsite, ftp.x.org, linux distributions)
discussion of online resources such as homepage and homepages
of various contributors
binary vs. source distributions
installation
if binary, make install
if source, make config, give details of compilation, make, make install
gimp library directory/personal gimp directories
gimprc
location of default gimprc
the various gimprc options (descriptions for each & default values)
discussion of tiles/swap
what are tiles?
why are they useful to the gimp?
what are reasonable settings for the tile cache size?
what is the gimp swapfile/tmp directory?
where should it be? (on /tmp or in ~/.gimp)
discussion of color resolutions
what are color resolutions?
how do you find out which color resolutions are available to you?
what color resolutions does the gimp require?
why does the gimp "dither" in 8-bit color?
why do the colormaps change between grayscale and color in 8-bit
why don't indexed images use their own palette?
discuss other alternatives for viewing images under X windows (xv)
the color cube resource: how can the shades of color be controlled in 8-bit?
discussion of fonts (type1/truetype/bitmap/standard X11)
how are fonts used in the gimp? (gimp requires X to render them)
what fonts can X render? (type1/bitmap)
what fonts come standard with X?
where can more fonts be obtained?
Images
the toolbox
foreground
background
changing the colors/color select dialog
main menu
tool descriptor field
new
gimp works on "images" which are defined by a width, height, depth,
a selection mask, a set of layers, and a set of auxillary channels
image types: what are the options?
what are the image types used for
why can't an indexed image be created?
why are indexed images a pain?
how should indexed images be dealt with?
a short example of "new" command
image menus
scrollbars
rulers (options for hiding/showing rulers)
ruler units (in gimprc) {pixels, inches, centimeters}
concepts of layers
bucket fill the new layer
add a layer, blend it
transparency
what is it?
how is it used?
how can its appearance be controlled? (via preference dialog/gimprc)
what file formats support it?
undo/redo
multiple (can be set via gimprc and preferences)
on a per-image basis
loading/saving
loading and saving of images is accomplished through plug-ins
most load/save plug-ins have no knowledge of the full set of
attributes which the gimp needs for an image. In fact, most are
limited to saving only 1 layer, and in some cases, an alpha channel
such plug-ins are limited by their respective file formats (jpeg, tiff, etc)
these plug-ins will save only the active layer of an image
the xcf file format, however, can save all pertinent information about
an image so that it can be stored and retrieved in full
viewing
viewing an image can be efficiently accomplished through a variety of features
the image can be zoomed in or out with the magnify tool and the zoom
hot-keys and menu-items
the zoomed image can be panned using the scrollbars at the sides of the image
or the middle mouse button
<possibly discuss the navigator here, if implemented>
multiple views can be created to view the image at different zoom ratios
the image window can be expanded to reveal hidden areas
and if necessary, the shrink wrap hot-key/menu-item can be invoked
to limit the window borders to the image extents.
the "always resize" option of the magnify tool and the gimprc can
be enabled to cause the window to resize appropriately for each successive zoom
resizing/scaling
images can be resized or scaled
resizing involves changing the width and height of the image by either
cropping or adding area to its contents
scaling involves actually changing the contents of the image, either
"stretching" or "compressing" the layers and channels to fit to the
new size
Selections
introduction
what are selections?
so is a selection just another channel? (what makes it special)
how can selections be created?
why are selections useful--they constrain operations by
allowing regions of an image to be specified
denoted by a boundary of marching ants--speed controlled in gimprc file
selection mask extends over entire image
if a layer in the image covers only part of the image extents, then
the parts of the selection boundary which lie outside the active layer will be
drawn, but will not march.
brief description of selection tools
pointer to channels section for information on how to "paint" selections
<possibility of "quit edit" mode for masks here>
common operations on selections
feathering -- like a gaussian blur (can also be done with selection tools)
bordering -- description of the algorithm
sharpen -- gives all points of the selection values of 0 or 255
invert -- gives each point of the selection the value = (255 - value)
stroke -- uses the current brush and foreground color and "strokes"
the outline of the current selection
clear -- if there is a selection, this clears the selected area of the active layer.
if the active layer has an alpha channel, this clears to transparent.
if not, this clears to the background color.
if there is no selection, this affects the entire image
fill -- similar to clear, except that it always fills to the background
color.
Tools
what are tools?
typically activated from toolbox, hot keys or menu items
only one active at a time
discussion of context sensitive cursors
the tools dialog
what is it?
double clicking on the tool icon
selecting from the dialogs menu
what are the common features of their usage:
always invoked with the left mouse button, and can be cancelled
by pressing the right mouse button before you release the left
shift, control keys are often used to modify the behavior of the current
tool
selection tools
rectangular: creates rectangular regions of selection which can be
added, subtracted, or used to replace the current selection.
holding down the shift and control keys constrains the region to a square
elliptical: creates elliptical selections similar to rectangular. holding the
shift and control keys creates circles. this selection tool has an antialiasing
option which is on by default and creates selections which appear smoother
but contain intermediate selection values. that is, there are points in the
selection which aren't completely "on" or "off".
freehand: can create any kind of selections, but requires a steady hand
possibility here for the (dynadraw option for much smoother input). when
the left mouse button is released, the created polygon is closed automatically.
there is an antialiasing option for this tool which produces smoother selections.
bezier: the bezier selection tool can be used to create either polygonal or curved
regions of selection. There are typically two stages to bezier selection. the first
involves placing bezier control points. control points are point which the final
curve will be constrained to pass through. these should be placed on the edges
of the boundary which is desired. no selection can be made until a final control
point is place atop the first control point. a selection can be made by clicking
in the interior of the polygon. each control point has two additional points which
control the curve segment which passes through. these two additional points can
be manipulated by dragging on the control point. holding down shift while
dragging constrains the movement to just one of the additional control points.
holding down the "control" key while dragging one of the original control points
translates that point. the bezier tool has an antialiasing feature which is useful
for producing smooth output.
fuzzy: the fuzzy selection tool is invoked by left-clicking in a region of interest
it works using a seed fill algorithm and expands outwards from the initial point
to adjacent pixels with similar color. its expansion is limited by a threshold
parameter which limits the acceptable difference between neighboring pixels.
this tool also contains an antialias option which, depending on the underlying
image, can produce smoother selections. this tool also has an option "sample merged"
which specifies that the underlying pixel information is to be taken from the image
as it appears to the user, and not just from the active layer or channel
holding down the left mouse button and dragging allows the threshold parameter
to be dynamically adjusted. moving the mouse left or up decreases the threshold;
moving the mouse right or down, increases.
intelligent scissors: take the stuff i wrote a long time ago
by-color: this tool is invoked not through the toolbox, but through the menus
it selects in a manner similar to the fuzzy selection tool, except that it searches the
entire image for pixels of color similar to the initial point, instead of doing a seed
fill.
give an explanation of the by-color-select dialog and its uses
for all selection tools, the produced region replaces the current selection by default
however, holding down shift will "add" the new selection to the current
holding down control will "subtract" the new selection from the current
all selection tools and the move tool will translate the current selection if
the "alt" key is depressed.
magnify: allows easy zooming on portions of the image. a mouse click centers
at the mouse click and zooms in one step. a shift-click centers at the
mouse position and zooms out one step
clicking and dragging with the mouse highlights an approximate
rectangular area which will be fitted as closely as possible to the
size of the window.
move tool: primarily deals with layer translation including floating selections
can be used to move selections when the "alt" key is depressed
crop: crops through all layers, channels and selection mask
<possibly an option for cropping only the current layer>
operation of cropping extents (dragging vs expanding/contracting)
transform tools:
discussion about the method of operation of transform tools
they float the current selection if there is one and transform the contents
if there is no current selection mask, they simply transform the active layer
if there is no selection mask and a channel is active, they have no effect
with the exception of the flip tools, all transforming tools pop up a dialog
box which gives information on the current transformation.
with the exception of the flip tools, all transformations can be done with
or without smoothing. smoothing results in a better final result, but
takes additional time and can possibly introduce intermediate shades
of the original colors to the final product. Smoothing cannot be used
with indexed images.
the degree (quality) of the smoothing algorithm can either be linear
or cubic. cubic takes substantially more time but yields superior results.
the degree of smoothing can be set as an option in the gimprc file.
flip horz: flips horizontally
flip vert: flips vertically
rotate: rotates. shift constrains the angle of rotation to 5 degree intervals
scale: scales. shift constrains the current height. control constrains the
current width. shift-control constrains the original aspect ratio
shear: shears. the direction of the shear (either horizontal or vertical)
is determined by the direction the mouse is first dragged in
perspective: allows the rectangular region to be mapped into any arbitrary
quadrilateral. the corners of the rectangle can be dragged to any new locations
and the resulting region is the mapping of the rectangle into the new quadrilateral
bucket fill: if there is a selection, fills inside the selection bounds. if there is no selection,
executes a seed fill from the point the mouse was initially clicked in. in the case of
no selection, an antialiasing parameter may be set which can yield better
results in the final fill. This option is similar to the one for the fuzzy selection tool.
the bucket fill tool fills an area with the foreground color. If the shift key is depressed
then the background color is used as the fill color.
patterns are used in conjunction with the bucket tool and are tiled to fill the
area. pattern dialog allows the selection of an active pattern
the behavior of the bucket fill tool can be selected from the tools dialog
to specify either pattern filling or color filling.
the pattern dialog can be displayed from the dialogs submenu
patterns are found in the directories listed in the gimprc
new patterns can be defined by saving an image in one of the searched
directories in the gimp pattern format.
the list of directories can be reparsed at any time with the refresh button
description of paint modes--tie in to paint tools/blend (they all use them too)
bucket fills can also be done with a specified opacity which controls how the
paint being applied from the tool interacts with the existing image
blend: blend typically works by filling from the foreground to the background color
the tool is activated by clicking and dragging in the desired direction and
then releasing. while dragging, a rubber-banded line will appear which
has different meanings depending on the current blending mode
there are a number of different blend modes including
linear: perpendicular lines to the drag line from foreground to background
bilinear: perpendicular lines to the drag line from foreground to background
in both directions
radial: concentric circles outwards from the starting of the drag line to the end
square: concentric squares outwards starting with FG and ending with BG
conical (asymmetric): cone with rays extending outwards from the
drag line's starting point. length of drag line has no meaning
conical (symmetric): cone of rays goes from fg to bg in both directions
shapeburst (angular, spherical, dimpled): shapeburst assigns a color
to pixels somewhere between fg and bg depending on how
far the pixel is from the nearest edge of the selection
blend, like bucket fill has both a paint mode and an opacity
blend does not necessarily work between the foreground and background
colors. the gradient editor can be used to construct arbitrarily complex fades
between as many colors as desired.
complete description of gradient editor
maybe a note about how all of the blend modes assign a value between 0 and 1
for each considered pixel and in the case of a simple blend, this corresponds to
a color intermediate to bg and fg. in the case of a custom blend, it corresponds
to a point on the collection of segments defined in the gradient editor.
blends can be done in RGB space, HSV space, and from the fg color to transparent
painting tools
all painting tools are applied using the left mouse button and dragging
paint is then applied as though the mouse cursor were a paintbrush
applied paint can be blended with the underlying image using a number
of different modes (as in bucket fill/blend)
normal: the color being applied is independent of the existing image
addition: the color is added to the existing
subtract: color is subtracted
multiply: color is multiplied, result is always darker
screen: inverse of multiply, result is always lighter
color only: hue and saturation of color are used, existing lightness kept
hue only: only hue of color is used
value only: only value is used
saturation only: only saturation is used
behind: color is applied "behind the existing image" meaning that
painting only affects parts of the image which have some
transparency
lighten only: color is applied only if its value exceeds existing
darken only: color is applied only if its value doesn't exceed existing
whatever sort of paint the painting tool is applying (this can even be "erasing"
paint, in the case of the eraser tool), the application is limited by a brush mask.
a brush mask is a grayscale image which determines the amount of paint to
apply at each stroke.
there are a number of brush masks that come with the gimp
brush masks are found from a list of directories specified in the gimprc
brush masks can be created by saving a grayscale image with the gimp brush
file loader/saver
brush masks can be selected using the brush dialog which is invoked from
the dialogs submenu.
complete description of the brush dialog
set paint mode
paint opacity
brush name, width, height, appearance
brush spacing
refresh feature for researching the brush paths
when using indexed images, all brush masks are converted from grayscale
to black and white so that additional colors are not introduced into the image
the paint tools are as follows:
paintbrush: applies the foreground color. a fade out parameter can be
set in pixels which specifies how long a stroke can be before it begins to fade
to transparency
pencil: similar to paintbrush, but it uses non-antialiased brush masks for
precise work which requires absolute pixel accuracy
eraser: the eraser tool does not apply paint, but removes it. if an image has
an alpha channel, paint is removed in favor of transparency. if an image does
not have an alpha channel, the eraser tool is similar to the paintbrush, but
paints in the background color.
airbrush: the airbrush tool paints in the foreground color, but unlike the
paintbrush which is constrained to apply an even coating of paint during
a single stroke, the airbrush can apply as much paint as desired. a pressure
parameter controls how quickly paint is applied, and a rate parameter controls
how quickly paint is applied when the mouse is stationary.
clone: the clone tool works by transfering paint from a source to a specified
destination. the source can either be an image or a pattern. the type of source
is selected through the clone tool dialog. if the source is an image, the actual
source coordinates can be specified by shift clicking the mouse
at the appropriate location. then, when painting, the image at the
source coordinates is effectively copied from the source to the destination.
cloning from image to image automatically adjusts between the different
image types (ie grayscale-> RGB, RGB->indexed). The RGB->indexed
conversion is complicated and can slow down the operation of the tool
at times. when cloning from image to image, the source and destination
can either be the same image or different images. if the clone source type
is a pattern, the active pattern is used as a source of pixels to paint to the
destination. the active pattern is selected the same way as with bucket fills.
convolve: convolving can involve either sharpening or blurring. a
pressure parameter in the convolve dialog allows the degree of sharpening
and blurring to be set.
Floating Selections
floating selections are temporary layers which are "floating" above another layer
or channel. only one floating selection at a time may exist in an image. floating selections
appear in layers dialog as a temporary icon, but can be treated in many respects
like a normal layer in the image (ie, opacity, combination mode)
can't be raised or lowered, given layer masks, etc.
created by pasting, floating a selection
introduce concept of global edit buffer
the edit/copy command saves region inside current selection
the edit/cut command save region inside current selection and extracts
the contents from the active drawable. if the layer has an alpha channel, extraction
yields transparent pixels. otherwise, extraction yields the background color.
if there is no selection, edit/cut copies and removes the entire layer or channel
edit/paste copies the contents of the global edit buffer into an image.
this stage automatically takes care of any conversions between image types
when the contents of the global edit buffer are pasted, any current selection
is removed
edit/paste has a special option "paste into" which does not remove the current selection
the appearance of floating selections--marching ants
if there is a selection with a floating selection, the boundary of the selection does not march
the selection mask acts as a mask when compositing the floating selection to the image. that is:
any areas of the floating selection which lie within the selection mask are visible,
while any areas which lie without the selection mask are masked out.
if edit/paste or select/float are invoked while a floating selection exists, the current
floating selection is "anchored" to the image
the floating selection can be anchored at any time (which simply means it is composited
permanently with the image) by clicking outside its boundary, or by invoking select/anchor
the move tool translates floating selection exactly as it translates layers
assuming no selection exists, transform tools transform the entire floating selection,
yielding a new one.
named buffers allow multiple edit buffers which can be recalled by name
Layers
introduction
layers are the basic building block of images
when a new image is created, a default layer is created
layers can have any extents relative to the image
image is like a frame, layers are cellophane overlays
layers have transparency, except for the bottommost layer, which doesn't necessarily
layers are composited with composite modes and opacities to form the final image
layers allow separate elements of an image to remain independent
for a final image, layers are composited into a final image
the layers dialog (blending mode, opacity, ordering, visibility, activation, attributes)
the layers dialog controls the layer compositing
the active layer is the one which is highlighted
when a tool is used, the active layer is the one affected
the active layer can be set by clicking in the list box item
the option menu controls the blending mode, (list all modes, uses)
the opacity slider allows the overall transparency of a layer to be set
each layer item has two icons on the left
the first is a small eye, which controls the visibility of a layer
the visibility allows layers to be hidden for clarity
the second is a fleur, which controls layer "linkage"
layer linkage is a feature which allows multiple layers to
be translated simultaneously with the active layer
layer list items are highlighted when active and their order in the
list denotes the composite order. composition proceeds from bottom to top
layer select feature
like the windows 3.1 task manager
shift-tab switches between layers
works like a stack. last visited layer is the next to pop up
layer previews
the size of layer previews in the layer selector and the layers dialog
can be controlled via the gimprc with the preview size setting
smaller preview sizes are often not quicker to calculate, but do require
less memory and allow more layers to be displayed simultaneously
set layer preview size to "none" for efficiency
merging
layers are merged by combining the visible layers
can be constrained by: image size, bottommost layer, or expanded as necessary
flattening
flattening is similar to merging, but removes transparency, and limits to image size
delete
removes the active layer
duplicate
makes a duplicate of the active layer, new layer is inserted above the active layer
new
new layer: specify size, name
resizing
layer can be resized so that additional border space is available, or the
layer's contents can be cropped.
dialog allows the contents of the layer to be positioned within the new layer
scaling
the layer is scaled to a new size
lowering/raising
allows the layers to be reordered
preserve transparency
an option which restricts all paint operations so that they have
no affect on the transparency of the layer
layer masks (all the modes of operation)
layer masks are additional alpha channels for layers.
they are essentially grayscale images with the same extents as the layer
a layer is limited to at most one
they allow independent localized changes to transparency
control-click in the layer mask preview toggles the activation
this creates a red border around the layer mask preview
shift-click makes the layer mask visible instead of the layer
this creates a green border around the layer mask preview
to toggle between editing the layer mask or the layer itself, click the previews
a border denotes which is currently active
layer mask creation
a layer mask is created for the active layer by selecting the layer mask menu item
the initial layer mask can be completely white (all areas of the layer will be visible)
completely black (the entire layer is invisible)
or set initially to the layer's alpha channel (useful for some effects)
layer mask application
choosing the layer mask apply menu item removes the layer mask
a choice is given between apply or discard.
apply causes the layer mask to be composited with the layer's alpha channel
discard removes the layer mask as though it never existed
layer mask restrictions
cannot transform a layer with a layer mask
layer masks cannot be given to floating selection layers
layer alpha to selection operation
the select/layer alpha command causes a layer's alpha channel to be
used as the basis for the image's selection mask
Channels
introduction
the two kinds of channels (image components, auxillary channels)
selection masks are channels
image components
for each type of image, the components are the independent
channels. for RGB: red, green, blue. grayscale and indexed have 1 component
alpha channels of images are not considered a component for the purposes of
discussing channels
auxillary channels
auxillary channels are not involved with the image.
they are used for storing selection masks and editing them as though
they were images.
they can be made visible by giving them a color and opacity.
they are then composited on top of the layers composite
interaction between active layer/active channel(s)
any combination of image components may be active at a given
time. the activation of image components acts to constrain the
application of paint to the active layer.
if a layer is active, no auxillary channels can be active
if an auxillary channel is active, no layers are active, and the
image components are insensitized
the channels dialog
invoked from the dialogs menu.
for each image, the image components are displayed first
auxillary channels are then displayed
each item has one icon on the left of a small eye and controls the
visibility of the channel or component
new
creates a new auxillary channel
image components cannot be created
duplicate
duplicates the current channel. this can be done for image
components and auxillary channel. the result is an auxillary channel
delete
deletes the active auxillary channel
image components cannot be deleted
double clicking an aux channel list item allows it's attributes to be edited
auxillary channel attributes include name and composite color & opacity
Image Correction
image correction deals with changing the characteristics of whole regions of an image
image correction tools operate on the contents of the current selection or the entire layer
if there is no selection
hue/saturation
allows adjustment of hue saturation--more on this
brightness/contrast
more on this stuff
levels
histogram
curves
maybe by 1.0!
color balance
slider controls the relative color shift
Plug-ins
introduction (why are they cool, what are they)
how they work
registration
interface
effects
extensions
file loaders/savers
Very Large Images
tiles
tile size is compiled into the gimp
tiles can be used to distribute gimp processing
tiles can be used for quick loading of images at varying resolutions
tiles are a nice mechanism for dealing with images >> physical memory size
performance issues (getting the most out of GIMP)
compositing is time consumptive--hiding layers when possible is a good idea
viewing only a small portion of the image at 1:1 makes redisplays after
big edits much faster
tile cache size should be set as high as possible, but should still be
sufficiently small that the system does not swap processes.
tile cache swap should be located on a partition with sufficient space
<give approximate rules of thumb here>
programming issues and approaches (plug-ins, extending GIMP)
plug-ins can use put, get-pixel abstractions
for efficiency, actual tiles should be dealt with
in both plug-ins and GIMP programming, the pixel regions are useful for
programming algorithms with tiles.
register_regions, process_regions
get_row, set_row
get_col, set_col
get_area, set_area
Scripting--The Procedural Database
introduction
scripting is extremely useful for automating image creation or common editing functions
there is no specific scripting language for the GIMP, instead procedural database
this way, there are no limits to what can be done with GIMP functionality: imagine
remote process servers which crunch through CPU intensive imaging tasks and return
data. the flow of image information is controlled through the interface to the pdb.
the procedural database
all procedures which handle the GIMP's functionality are registered here
all plug-ins and extensions must register themselves with the main database
once registered, plug-ins & extensions can call other plug-ins or internal gimp procs.
procedural database contains information on each procedure including:
blurb, help, author, copyright, input & output arguments (with descriptions)
all procedures are blocking
internal procedures
internal procedures constitute the inner workings of the main application.
images, channels, layers, etc. can be created here and the functionality
of almost all the tools can be accessed through internal procedures
in fact, the entire plug-in interface simplifies calls to internal procedures
<give header file location for internal procedures>
<data structures for arguments, procedure definition>
<give call here for registering>
<invocation procedures>
<short example>
plug-ins/extensions
plug-ins and extensions are expected to register themselves in the
procedural database upon initialization which occurs if the executable
is found by the gimp in an appropriate directory.
once registered, a plug-in or extension can be accessed in the same
way as every other procedure.
<registration example--all plug-ins must do this>
<plug-in interface to procedural database>
interface documentation
can be automatically generated with procedural_db_dump call.
then an elisp script creates a tex info file.
registering a procedure with full information is important for these docs
available scripting extensions
give a list--probably will contain something for the siod interpreter
siod-scheme in one defun
all procedures in pdb are queried, scheme wrappers are auto-generated