Updating the cursor information involves sampling the active image/
drawable, which can be expensive, especially if there are filters
active that also respond to cursor motion, e.g., while using the
warp tool.
Move the actual updating to an idle function. This dramatically
improves responsiveness in these situations.
Remove the invert-linear and invert-non-linear variants and simply add
"gboolean linear" to gimp-drawable-invert. This should actually be an
enum but I didn't find a good name right now...
Pass through mode uses the same compositing logic as REPLACE mode,
however, it's a special case of REPLACE, where the layer is already
composited against the backdrop. This allows us to take a few
shortcuts that aren't generally applicable to REPLACE mode.
Add a dedicated op class for pass through mode, derived from the
REPLACE mode op, implementing these shortcuts.
and add gimp_drawable_invert_linear(). Also, finally deprecate
gimp_invert() and port all its uses in plug-ins and scripts to
gimp_drawable_invert_non_linear() so the result is the same.
Fix the default brush name -- "Round Fuzzy" was gone for a while :P
The fact that "Hardness 050" was the default brush regardless is a
conincidence; see the bug report for more details.
glib-genmarshal was rewritten in glib 2.53.4, and as of now (2.53.6)
it has a bug where it unconditionally generates marshaler bodies,
even for standard marshalers, even with --stdinc. This causes
libgimpwidgets to define and export g_cclosure_marshal_VOID__INT()
and g_cclosure_marshal_VOID__OBJECT(), which upsets defcheck, and
breaks the build.
Work around this for now by using --header --body when generating
the marshal.c files, which includes the prototypes in the source,
instead of including the header ourselves. This is the only code
path where the new glib-genmarshal doesn't generate bodies for
standard marshalers. Note, however, that this usage is deprecated,
so we'll probably want to change it back once it's fixed.
...in both the core and libgimp.
Images now know what the default mode for new layers is:
- NORMAL for empty images
- NORMAL for images with any non-legacy layer
- NORMAL_LEGAVY for images with only legacy layers
This changes behavior when layers are created from the UI, but *also*
when created by plug-ins (yes there is a compat issue here):
- Most (all?) single-layer file importers now create NORMAL layers
- Screenshot, Webpage etc also create NORMAL layers
Scripts that create images from scratch (logos etc) should not be
affected because they usually have NORMAL_LEGACY hardcoded.
3rd party plug-ins and scripts will also behave old-style unless they
get ported to gimp_image_get_default_new_layer_mode().
gimp_layer_update_mode_node(): when showing the mask, set mode to
NORMAL, and make sure that the composite space is PERCEPTUAL for
LEGACY layers, and LINEAR (or whatever is chosen in layer attibutes)
otherwise.
this commit changes just those which make no difference to
functionality: property and object member defaults that get overridden
anyway, return values of g_return_val_if_fail(), some other stuff.
Looked a bit deeper into heal: while I didn't try to understand what
it's actually doing, this is strange: there is a comment that says
that healing should done in perceptual space, and the code uses
R'G'B'A float (at least it completely did before the last commit).
On the other hand, the code adds and subtracts temporary buffers,
which screams "gamma artifacts" unless done in linear space.
This commit changes everything to use linear float buffers,
and removes the comment. It "looks" right to me now, please test.
Split libappgegl into libappgegl-generic and libappgegl-sse2, and
move the SSE2 code (part of the newly added smudge code) to the
latter, so that the rest of the code can be compiled without SSE2
compiler flags. This allows building GIMP with SSE acceleration
enabled, while running the resulting binary on a target with no
SSE accelration.
Commit 3635cf04ab moved the special
handling of bottom-layer compositing to GimpOperationLayerMode.
This required giving the op more control over the process()
function of its subclasses. As a temporary workaround, the commit
bypassed the subclasses entirely, using "gimp:layer-mode" for all
modes. This is the reckoning :)
Add a process() virtual function to GimpOperationLayerMode, which
its subclasses should override instead of
GeglOperationPointComposer3's process() functions. Reinstate the
subclasses (by returning the correct op in
gimp_layer_mode_get_oepration()), and have them override this
function.
Improve the way gimp_operation_layer_mode_process() dispatches to
the actual process function, to slightly lower its overhead and
fix some thread-safety issues.
Remove the "function" field of the layer-mode info array, and have
gimp_layer_mode_get_function() return the
GimpOperationLayerMode::process() function of the corresponding
op's class (caching the result, to keep it cheap.) This reduces
redundancy, allows us to make the ops' process() functions private,
and simplifies SSE dispatching (only used by NORMAL mode,
currently.)
Move the blend and composite functions of the non-specialized
layer modes to gimpoperationlayermode-{blend,composite}.[hc],
respectively, to improve code organization.
Move the SSE2 composite functions to a separate file, so that they
can be built as part of libapplayermodes_sse2, allowing
libapplayermodes to be built without SSE2 compiler flags. This
allows building GIMP with SSE acceleration enabled, while running
the resulting binary on a target with no SSE accelration.
Add a "blend_function" field to the layer-mode info array, and use
it to specify the blend function for the non-specialized modes.
This replaces the separate switch() statement that we used
previously.
Remove the "affected_region" field of the layer-mode info array.
We don't need it anymore, since we can go back to using
GimpOperationLayerMode's virtual get_affected_region() function.
Last but not least, a bunch of code cleanups and consistency
adjustments.
Paint tools in straight line mode (shift click) will now display the
angle in status bar. Angle 0 is considered as the horizontal line from
left to right, and angle is measured counterclockwise from there, which
is the most common convention.
Use GIMP_LAYER_MODE_NORMAL (not NORMAL_LEGACY) when falling back from
gimp_paint_core_replace() to gimp_paint_core_paste() for layers
without alpha. Adapt the format of the used paint buffers accordingly.
... and fix flatten-image along the way. *And* do some cleanup.
Currently, gimp_image_merge_layers() combines the layers on its own,
one by one. This is incompatible with pass-through groups, because
the group's buffer is rendered independently of its backdrop, while
we need to take the backdrop into account when mergeing the group.
Instead, render the subgraph of the parent graph, corresponding to
the set of merged layers, directly into the new layer. Since the
layers we merge are always visible and continuous, all we need is a
minor massage to the parent graph to make it work. This takes care
of pass-through groups intrinsicly.
This commit also changes the behavior of flatten-image: Currently,
the flattened layers are rendered directly on top of the opaque
background, which can make previously-hidden areas (due to layers
using composite modes other than src-over, or legacy layer modes)
visible. This is almost certainly not desirable.
Instead, construct the graph such that the flattened layers are
combined with the background only after being merged with one
another.
GimpFilter's is_last_node field only reflects the item's position
within the parent stack. When a layer is contained in a pass-
through group, it can be the last layer of the group, while not
being the last layer in the graph as a whole (paticularly, if
there are visible layers below the group). In fact, when we have
nested pass-through groups, whether or not a layer is the last
node depends on which group we're considering as the root (since
we exclude the backdrop from the group's projection, resulting in
different graphs for different groups).
Instead of rolling our own graph traversal, just move the relevant
logic to GimpOperationLayerMode, and let GEGL do the work for us.
At processing time, we can tell if we're the last node by checking
if we have any input.
For this to work, GimpOperationLayerMode's process() function needs
to have control over what's going on. Replace the derived op
classes, which override process(), with a call to the layer mode's
function (as per gimp_layer_mode_get_function()) in
GimpOperationLayerMode's process() function. (Well, actually, this
commit keeps the ops around, and just hacks around them in
gimp_layer_mode_get_operation(), because laziness :P)
Keep using the layer's is_last_node property to do the invalidation.
GimpTileHandlerProjectable is similar to GimpTileHandlerValidate,
except that it calls {begin,end}_render() on its associated
projectable before validating.
In pass-through mode, the group layer-stack's input is connected to
the backdrop. However, when rendering the group's projection, we
want to render the stack independently of the backdrop.
Unfortunately, we can't use the stack's graph as a subgraph of two
different graphs.
To work around that, the next few commits add a mechanism for a
projectable to be notified before and after its graph is being
rendered. We use this mechanism to disconnect the stack's graph
from the backdrop before rendering the projection, and reconnect
it afterwards. Yep, it's ugly, but it's better than having to
maintain n copies of (each node of) the graph (each nesting level
requires an extra copy.)
This commit adds {begin,end}_render() functions to GimpProjectable.
These functions should be called right before/after rendering the
projectable's graph.
When any of the children of a pass-through group excludes its
backdrop, the group itself should exclude the backdrop too. Override
get_excludes_backdrop() to follow this logic, and call
update_excludes_backdrop() when this condition might change.
Note that we always composite pass-through groups using src-over mode,
so to actually hide the backdrop, we need to disconnect it from the
group's mode node's input pad (and reconnect it, when the backdrop is
no longer hidden).
Override GimpDrawable::get_source_node() for GimpGroupLayer. Use
a node that contains both the drawable's buffer-source node, and the
layer stack's graph node. Choose which one of these to connect to
the source node's output based on the group's layer mode: the stack
graph for pass-through mode, and the buffer-source node for all the
rest.
When in pass-through mode, connect the source node's input (which
receives the backdrop) to the stack graph's input. Keep maintaining
the projection in pass-through mode. ATM, the projection uses the
same graph as the source node, so it's rendered against the group's
backdrop -- we don't want that. The next few commits fix it.
Update the group's drawable directly upon filter stack update in
pass-though mode, because the group's graph doesn't go through the
projection.
TODO: if any of the group's children (or a child of a nested pass-
through group, etc.) uses dst-atop/src-in, this needs special
attention.
Make sure the input of the layer's filter node is connected to its
source node (when it has an input pad), so that, once we implement
pass-though mode, the group's source node can see the backdrop.
For pass-through groups, we want to use the group's layer-stack
graph directly in its filter node, in place of the drawable's
buffer-source node. Add a get_source_node() vfunc to GimpDrawable,
which defaults to returning the buffer-source node, and use it in
gimp_drawable_get_source_node() instead of using the buffer-source
node directly. We'll later override this function for
GimpGroupLayer.
... causing compilation to fail on 32 bit targets
Use SSE2 compiler flags when building libappgegl, since it's used by
the new smudge tool code.
Avoid using SSE for the smudge tool if SSE acceleration is disabled
at runtime, or if the buffers are not properly aligned.
Add "gboolean with_filters" to gimp_drawable_calculate_histogram(),
which is passed as FALSE in almost all places, except the histogram
dockable where we want to see both the drawable's unmodified histogram
*and* the histogram after filters are applied.
During constrained motion, round the slider value before clamping
it, so that the slider limits are always enforced. Additionally,
snap the slider to 1/12-ths of the line length, rather than
1/24-ths.
Make sure that sliders can never have negative-zero values, which
can result in a -inf base for spiral.
Shift-click should actually toggle only within a given group. The new
capability of toggling only a sub-item, brought by commit 970e9ac is
still feasible in 2 steps: first toggling the parent (item group), then
the desired child.
It brings now a third possibility with exclusive toggle among many
children items, without touching other groups and top-level items.
... so that when the base and balance sliders overlap, the base
slider is the one that's picked, since the balance slider is
constrained by the base, but not the other way around.
When loading tiles from an XCF, reject tiles whose on-disk size is
greater than 1.5 times the size of an uncompressed tile -- a limit
that is already present for the last tile in the buffer. This
should allow for the possibility of negative compression, while
restricting placing a realistic limit.
Currently, no limit is placed on the on-disk tile data size. When
loading RLE- and zlib-compressed tiles, a buffer large enough to
hold the entire on-disk tile data, up to 2GB, is allocated on the
stack, and the data is read into it. If the file is smaller than
the reported tile data size, the area of the buffer past the end
of the file is not touched. This allows a malicious XCF to write
up to 2GB of arbitrary data, at an arbitrary offset, up to 2GB,
below the stack.
Note that a similar issue had existed for earlier versions of GIMP
(see commit d7a9e6079d3e65baa516f302eb285fb2bbd97c2f), however,
since prior to 2.9 the tile data buffer was allocated on the heap,
the potential risk is far smaller.
...with known plugins
Add new plug-in file-raw-placeholder.c that registers itself for
loading all RAW formats from file-raw/file-formats.h, but does nothing
except returning an error message pointing to darktable and
RawTherapee.
When no real RAW loader is installed, this plug-in is selected
automatically as RAW loader, otherwise the first installed RAW loader
is used. Selecting another in prefs still works as before.
... "threads" property.
Actually there is no need of having a public GEGL_MAX_THREADS as written
in the previous commit. We can just retrieve the max for a GObject
property.
Raise GIMP_MAX_NUM_THREADS to 64, following the changes in GEGL (see
GEGL commits 6d128ac and f26acbb). This is still considered unstable and
to be used at one's own risk (cf. GIMP commit 1f5739d) but at least, it
could allow discovering and fixing bugs.
It would be nice if GEGL_MAX_THREADS could be public so that to not have
to edit this by hand at each change.
- trailing whitespaces cleaned out;
- vectors are called "path" in all visible strings;
- do not check for floating selection and active channel: oppositely to
layers, a vector can be selected in the same time as a channel, and
while there is a floating selection.
and update the grid as soon as a constraint is changed, not only on
the next motion. Change GimpTransformTool to forward the events to the
widget if it exists, but still handle them if it doesn't (yes this
code duplication is ugly, but the widget can hardly handle events if
it doesn't exist...).
More than 2000 lines of code less in app/, instead of
if (instance->member)
{
g_object_unref/g_free/g_whatever (instance->member);
instance->member = NULL;
}
we now simply use
g_clear_object/pointer (&instance->member);
Change the text selection to draw an outline around each selected
glyph. It looks just as ugly as before but at least keeps the text
readable regardless of its color.
if did revert to the previous selection and thus break stuff like
enbaling quick mask or inverting the selection, because I merged the
undo magic it does into gimp_rectangle_select_tool_halt(), whereas
before it was done by the former gimp_rectangle_tool_cancel(), so only
on explicit cancel not HALT from whatever source.
Do the same in the new code and move the undo magic from halt() to
rectangle_response(CANCEL), which is exactly the same distinction as
with the old GimpRectangleTool code.
When the user provides a filename without an extension in the save
dialog, we add one for them, update the filename in the dialog, and
retry. However, the updated filename is made up of only the
basename, leaving out the dirname part, if specified. This means
that if the user enters "/somedir/somefile", the new filename
becomes "somefile.xcf", which refers to the current directory,
instead of "somedir".
Fix this by maintaining the dirname when adding a file extension.
Don't unconditionally call COMMIT in rectangle_response(), because
that now implicitly HALTs the tool. Instead, check if we got here
because of a click, and call our commit() directly.
which is the same as g_object_class_list_properties() but filters
out the properties for which we don't want to create a GUI.
Use it in gimp_prop_gui_new().
GimpOperationTool's aux inputs were not properly ported to the new way
filter tools work (complete destruction and re-creation of the tool
dialog).
Split creating the operation GUI and adding it to the dialog into
separate functions, and call them at the right places.
Call HALT generically in gimp_tool_control() after calling COMMIT, and
remove all hacks in tools that call both COMMIT and HALT or call
halt() from commit().
Some tools interact with their subclasses (e.g. filter tool and
operation tool), and it's essential that COMMIT runs through the
entire class hierarchy before HALT.
Probably breaks something, please test.
Need to keep around the operation's name and its description, so
everything can be re-created when an image is clicked.
Instead, completely shut down GimpOperationTool when GimpGeglTool is
halted, so the next click will bring up a dialog with only the
operation selection combo.
which allow to override stuff from GimpToolInfo for dynamic tools like
GimpFilterTool and friends. When NULL, the getters are falling back to
GimpToolInfo strings.
...and "export_dialog_title"
It's ridiculous to keep this code around for strings that are only
marginally different (and not better) than the strings we can generate
from other strings we have anyway.
to ::can_undo() and ::can_redo(). They still return description
strings and the new naming is slightly off :) but get_undo_desc() will
be needed for something else soon, and half of the time the functions
are indeed used to check whether there are undo/redo staps at all.
Switch the gegl:spiral prop gui from using a line controller to
a slider-line controller, and use sliders to control the "balance"
and "base" properties.
Add supprt for placing sliders on a GimpToolLine -- handles that can
be dragged over the line. The sliders are accesible through a new
"sliders" property, and via the gimp_tool_line_{get,set}_sliders()
functions.
Add a slider-line controller, which works like a line controller,
but whose callback also supplies/takes an array of sliders.
Note that the data type for individual sliders is called
GimpControllerSlider (in particular, it's not line specific), so
that we may use it with other controller/tool-widget types in the
future.
Use the new gimp_filter_tool_get_drawable_area() instead of always
using gimp_item_mask_intersect() which is only right when the
operation is applied to the "selection" region. Also call
gimp_operation_tool_sync_op() when the region is changed in the UI.
Don't offset the ui range of op properties that use pixel-distance
units to the top-left corner of the region, since they're relative;
only do that for pixel-coordinate units. Let their ui range be
[0, width/height].
Pass a "GimpCreateControllerFunc" to all gimppropgui-*.[ch]
constructors which takes a callback (to update the config object when
the on-canvas GUI) and a controller type that determines the
callback's signature, and returns another callback (to update the
on-canvas GUI when the config object changes).
In GimpOperationTool, pass such a GimpCreateControllerFunc that
handles creating and adding on-canvas controller via the new
gimpfiltertool-widgets.[ch]. So far, a simple line like in the
blend tool is supported.
Add a custom GUI for gegl:spiral, and have its origin, radius and
angle controlled by such a line.
which return's the used drawable's offsets and a GeglRectangle
where the filter is applied according to GimpFilterOptions::region
(either the selection or the whole drawable).
and call it from GimpFilterTool's "notify" callback. Remove signal
connections from all subblasses and instead implement ::config_notify().
The config object belongs to GimpFilterTool, and only GimpFilterTool
should know when it's created and can be connected to.
so things from the tool's previous use get destroyed, including their
(maybe dangling) signal connections. Also shut down more stuff in
halt(), including destroying not just hiding the GUI.
because widgets are bound to one GimpDisplayShell. Also, chain up
unconditionally in gimp_color_tool_draw(), we always want to draw the
widget even while picking colors.
The widget is fed events by GimpFilterTool, the actual interaction
with the filters operation and config will be done by subclasses.
The order of precedence when there are multiple possible canvas
interactions is: moving the split preview guide, color picking,
widget.
- introduce new state "boolean is_first" which tracks if the currently
drawn rectangle is the first with this instance
- cancel the widget if there was no movement when creating the first
rectangle
- undo to the previous rectangle if the user created a zero-extent
rectangle
- also undo to the previous rectangle if a newly drawn rectangle is
canceled with button-3 release
We can't rely on g_object_unref() in halt() for breaking all property
GBindings between the tool options and GimpToolRectangle, because we
might be in the middle of a signal emission which also refs and keeps
the rectangle alive until the last callback returns. So we had
dangling rectangles interacting with tool options.
Remember all bindings in a list, and break them explicitly when we
shut down the rectangle in halt().
Also, forgot to unset the display's highlight in the rectangle
selection tool.
because of bailing out early after emitting "response". Instead, don't
ref the object around this function, and move the "response" emission
to the end of the function.
The tool manager still keeps an active tool which it unrefs on
destruction, triggering a final HALT on the tool, which may want to
lookup tool options to reset something. Happened with the new
widget-ported rectangle select tool.
- enable the setting code in gimp-gegl.c again
- but set the default to one thread in GimpGeglConfig, with a CPP warning
- rename "processors" to "threads" in the GUI
- add a warning box about unexpected results when increasing #threads
We were leaking all tool widgets set with gimp_draw_tool_set_widget(),
and those having signal connections to e.g. the display shell were
doing things when they were supposed to be gone. Fixes make check.
- factor out widget creation to new start() function
- and tool shutdown to new halt() function
- connect to "response" and remove key_press()
- remove oper_update(), it was doing the same as the draw tool impl
- unset "rect_adjusting" before bailing out on button_release()
- update the integer rectangle when the double properties change
- don't try to show handles with a size of < 3
- remove unused members
- shorten some variable names
which is a replacement for GimpRectangleTool. It's a massive piece of
code and I'm not sure everyting works as it should, but it seems to do
crop stuff without any glitches.
and a default key_press() handler that emits CONFIRM, CANCEL and RESET
responses. Remove code with the same purpose from all subclasses.
Change tools feed key_press() to the widget and connect to its
"response" instead of implementing key_press() themselves. This will
only be better and less code after the tool side of this is done
generically.
Rename gimp_tool_widget_snap_offsets() to set_snap_offsets(),
and add gimp_tool_widget_get_snap_offsets().
Also rename gimp_tool_widget_status() to set_status(), and
add new function and signal set_status_coords().
When toggling visibility of a child in an item group, we should also
toggle the visibility of other items in the same group, as well as
top-level items. Otherwise toggling exclusive visibility of any item in
a group is identical to toggling the parent's exclusive visibility,
which is simply absurd.
We still don't touch visibility of items in other groups.
which is the code that calculates handle size based on pointer
proximity. Use the new function in GimpToolHandleGrid and
GimpToolLine, and clean up some stuff in GimpToolLine.
Simply use g_object_bind_property() to connect the grid properties of
GimpTransformOoptions and GimpToolTransformGrid and remove all other
grid property setting code.
The GimpLayer implementation of the GimpItem transform functions,
and the GimpDrawable convert_type() function, apply their operation
to both the layer and its mask. The subclasses of GimpLayer --
GimpGroupLayer and GimpTextLayer -- override some of these
functions, providing their own logic for the layer part, and
duplicating the mask part.
Avoid this duplication by adding a set of virtual transform and
type-conversion functions to GimpLayer. Have the GimpLayer
implementaion of the corresponding GimpItem and GimpDrawable
functions use these functions to apply the operation to the layer,
while taking care of the mask themselves. Have GimpLayer's
subclasses override the new virtual functions, instead of the
GimpItem and GimpDrawable ones.
Note that the existing implementation of convert_type() in
GimpTextLayer neglected to convert the mask, hence text layer masks
retained their old format after conversion. This issue is fixed as
a side effect of this commit.
Humanize action names to make them readable while
preserving their original grouping. Mark for translation
the missing ones. Use absolute values to make
"increase/decrease more" less cryptic since we hardcode
those values anyway.
so widgets can return which handle was clicked. The values boolean
semantics stay the same so if(retval) gives the same result. This is
useful for the upcoming transform tool widgets.
The GEGL ops sanity check causes all ops to be initialized. The
strings used by their properties will pick the translation selected
at the time of the check. It must therefore run after language
intiailization, otherwise the selected translation would correspond
to the system locale, even if the user selected a different language
in the preferences.
Split the sanity check into early and late stages. The early stage
is run before the call to app_run(), as it did before, while the
late stage is run during app_run(), after the configuration has been
loaded. Currently, the GEGL ops check is the only late-stage check;
all other checks are performed during the early stage.
The identity parameter checks added to the raster brush
transformation functions in the previous commit are unnecessary,
since we're already testing for the identity matrix. Remove them.
Check if the brush parameters match the identity parameters, and
return the original brush mask/pixmap if they do, in the actual
mask/pixmap transformation virtual functions, instead of in their
wrappers. While the identity parameters for raster brushes are
always scale=1, aspect-ratio=0, angle=0, and hardness=1, for
generated brushes they depend on the specific brush
parameterization.
Thanks to Lionel N. who bugged me with his Windows installation where
the XML file was not found by GIMP. We should output a warning when this
happens so that packagers can detect the issue and the expected
installation path for this dependency.
The reporter wants me to call him "Padawan Lionel" but I won't fall for
it! Or did I? :-D
This option behaves similarly to the other transform tool, however
it's limited to "adjust" and "clip" only. Now that the flip tool
can reflect across guides, this option is meaningful.
... when clipping, if they have an alpha channel
Right now, this case is only reachable through PDB, but it will become
relevant for the flip tool soon.
When compressing tool motion events, only compress motion events
at the front of the event queue, targeted at the same widget as,
and having similar characteristics to, the initial motion event;
stop compressing upon the first mismatched event.
Previously, all pending motion events targeted at the canvas were
compressed, stopping only at a button-release event targeted at the
canvas. As a result, when adding a guide by dragging from a ruler,
there could be a situation where a pending button-release event
targeted at the ruler would be followed by motion events targeted at
the canvas, with a cleared state mask. These motion events would
get compressed to the initial motion event targeted at the ruler,
resulting in a motion event with a cleared state mask being processed
before the said button-release event. This, in turn, would cause the
guide tool's cursor_update function to be called, while the tool is
still active, emitting a CRITICAL. Sheesh.
The moral of the story is: let's play it safe.
which encapsulates a bunch of GimpCanvasItems plus their interaction,
using GimpTool-like virtual functions like button_press(),
button_release() motion() etc. Also has GimpDrawTool-like API to
manage the canvas items of its subclasses.
Call gimp_canvas_group_remove_item() and don't just unref them, so
their state gets restored and signals get disconnected. They may not
be owned by the group, or have other external references.
Simply increase its change_count in dispose(). There is really no
reason to build expensive update regions and emitting signals when we
are about to go away.
Add a log handler so that GIMP can display errors outputted by GEGL.
Since third party code may run in threads and we have no control on
these, we have to be sure GTK+ code is run in a thread-safe way, hence
the usage of gdk_threads_add_idle_full(). This was the case here for
GEGL, and handling GEGL logs the same way as other GIMP logs would
result in crashes.
This way, you can increment repeated messages even when not the last
one and you don't overflow the error dialog needelessly when 2 errors
repeat one after another.
by encoding them directly in the string attached to all filter
actions. The code now supports both "gegl:some-operation" and
"gegl:some-operation\n<serialized config>".
Add "default_settings" to GimpGeglProcedure to store the settings of
the invoking action, much like the "default_run_mode" member.
Change filters-commands.c to parse the new operation string, create
GimpGeglProcedures with the deserialized settings, and use those
settings when the procedures are ran.
Change the filter history to be smarter about what is already in the
history, there can now be several different procedures with the same
name.
Remove the dilate and erode actions from the drawable group, and add
them to filters, they are just special cases of value-propagate with
fixed settings.
Allow NONE(0) for "wrapmode" (which translates to GEGL_ABYSS_NONE) in
the plug-in-edge PDB compat wrapper. The old plug-in accepted this
(undocumented) value and used a GimpPixelFetches in NONE mode.
which determines if a filter is applied directly (RUN_NONINTERACTIVE)
or asynchronously using GimpOperationTool (RUN_INTERACTIVE).
Split filter actions in two groups, one for direct apply and one for
interactive apply, which have separate callbacks that create
GimpGeglProcedures with the right default_run_mode set.
(After doing this distinction automatically based on the existance of
editable properties, I figured will might want direct apply also for
filters that do have properties, such as e.g. dilate and erode, which
are just value-propagate with some constant property values)
which gets added automatically by procedure_commands_get_display_args().
Move the non-interactive and run-with-last-vals code to
gimp_gegl_procedure_execute() (not execute_async()) because it makes
more sence to call it synchronously anyway (not implemented yet).
This commit should change no behavior.
Since commit bc4589968c, GimpFilterTool
assumes that color picking is only ever enabled through color picker
toggles, created using gimp_filter_tool_add_color_picker(). However,
the curves tool enables color picking using the color tool functions
directly. CRITICALS ensued.
Use the new gimp_filter_tool_enable_color_picking() function, added
in the previous commit, and handle picking through the filter tool
interface, instead of the color tool interface. This fixes the
issue.
Subclasses of GimpFilterTool should use these functions to control
picking, instead of using gimp_color_tool_{enable,disable}()
directly. This makes sure that the tool's picking state is
consistent, and allows the caller to control the pick identifier,
and use abyss picking (not currently needed by any subclass, but
maybe in the future, who knows.)
Return only the config object's GType and do the g_object_new() in the
caller (one caller only needs the type, there is no need to create a
dummy object just to get to its type).
around calling the actual execute vfuncs, Seen a crash when quitting
GIMP while a script-fu window was open, the progress was already
finalized. Now it simply prints the error message about the failing
script on the concole.
so they show up in recent filters, and don't need their own callbacks.
This has the problem that they now show a GUI with no options, but
that simply puts on more pressure to fix this general uglyness of ops
without editable properties.
We don't search recursively but only at the first level. If a plugin is
in its own subdirectory, the entry point has to be named the same (minus
extension) as the directory. For instance my-plugin/my-plugin for a
binary, or my-plugin/my-plugin.(py|exe|…).
This way, a plugin can load shared objects (libraries, other script
files, etc.) without polluting the main plug-ins directories, and in
particular without interfering with other plug-ins.
This is a first step to fix bug 757057 (DLL files which were used in
various plugins).
Commit 1e6acbd4e1 modified the
generated enum recipes to run gimp-mkenums from the source
directory, instead of the build directory, so that only the
basenames of the corresponding header files would appear in
the comment at the top of the generated files. This was a
mistake -- $(GIMP_MKENUMS) is expecting to be invoked from the
build directory.
Switch back to running gimp-mkenums from the build directory. To
avoid including the relative path from the build directory to the
source directory in the generated file, add a @basename@ production
variable to gimp-mkenums, which exapnds to the basename of the
input file, and use it instead of @filename@ in the recipes for the
generated enum files.
When regenerating an enum file, don't copy it back to the source
directory if it hasn't actually changed. This allows using a read-
only source directory where the enum header is newer than the
generated file, as long as they're not really out of sync.
OTOH, *do* touch the generated source-dir file even when unchanged,
in order to avoid re-running its recipe on the next build, however,
allow this to silently fail (which is harmless).
The layer blend space, composite space, and composite mode
properties have a special AUTO value, which may map to different
concrete values based on the layer mode. Make sure we can change
this mapping in the future, without affecting existing XCFs (saved
after this commit), by encoding these properties as follows:
When saving an XCF, if the property has a concrete (non-AUTO)
value, which is always positive, encode it as is. If the property
is AUTO, which is always 0, encode it as the negative of the value
it actually maps to at the time of saving (note that in some cases
AUTO may map to AUTO, in which case it's encoded as 0).
When loading an XCF, if the encoded property (stored in the file)
is nonnegative, use it as is. Otherwise, compare the negative of
the encoded property to the value AUTO maps to at the time of
loading. If the values are equal, set the property to AUTO;
otherwise, use the concrete value (i.e., the negative of the value
stored in the XCF).
Note that XCFs saved prior to this commit still load fine, it's
simply that if we change the AUTO mapping in the future, all their
AUTO properties will keep being loaded as AUTO, even if the
resulting concrete values will have changed.
... to the warp tool
The interpolation and abyss policy options control the sampler type
and abyss policy of the map-relative node. The high quality preview
option determines whether to use the same sampler for map-relative
during preview as the one used during commit, or whether to use a
fast nearest-neighbor sampler.
A bit too much? Maybe :)
... and to disable/control the rate of the periodic stroke.
The warp tool is now fast enough to enable stroking directly in
the motion handler, which gives better-quality response to motion
than stroking periodically. It's not quite fast enough to enable
exact motion, though :/
Allow individually enabling/disabling stroking during motion and
periodically, and allow controlling the rate of the periodical
stroke.
because the scale is useless otherwise with its range of 2..10000,
and because I needed a test case for the next commit (the
gimp_scale_entry_set_logarithmic() API was unused).
When using the MOVE behavior, don't append the current cursor
position to the stroke path in the timeout proc if the cursor
hasn't moved since last time. It has no effect, except for
requiring an unnecessary update.
When undoing a warp stroke, don't disconnect the current warp node
from its predecessor; rather, keep the graph as-is, and only
reconnect the render_node to the previous node. This avoids
invalidating the undone node, so that redoing it (which, likewise,
only involves reconnecting the render_node) doesn't require
reprocessing, making it much faster.
The spacing parameter controls the stroke spacing of the warp op.
It's similar, but not identical, to the brush spacing parameter of
the paint tools. It provides a tradeoff between speed and quality.
Change the UI range of the hardness parameter from [0, 1] to [0, 100],
to match the other parameters.
"layers-text-tool" action shows as "Text Tool" while "vectors-path-tool"
shows as "Path Tool". That's very confusing with tools-text and
tools-vectors respectively.
These actions are mostly about entering in edit mode with the active
layer or path. For text layers, it will enter text edition on canvas,
whereas just open the attributes edition dialog on other layers. For
consistency, layers-text-edit is renamed as well too layers-edit-text.
This also fix the side effect of commits 10099bd and 526918b where I
didn't realize that layers-text-tool was also working on non text layers
on purpose (being very badly named). Now there is a separate layers-edit
and layers-edit-text.
Thanks to Pat David for English corrections. :-)
Limit the number of lines drawn in the transform tool canvas grid to
one line every 5 image pixels. This should probably be done in display
pixels, didn't change that yet.
When transform tools are applied on a path, set their bounding box as follow:
- if a selection exists, use the selection bounds
- else if the path has a valid bounding box, use it
- else use the image canvas bounds
Also disable transform tools on an empty path (path without strokes) since
there is no data to transform.
Add support for The following GEGL op property keys, which shall
contain GUM expressions of the specified type, controlling the
corresponding UI attributes of the property's widget:
- sensitive [boolean]: controls widget sensitivity.
- visible [boolean]: controls widget visibility.
- label [string]: controls widget label (or the label of the
associated label widget).
- description [string]: controls widget tooltip text.
When any of above keys are present, the values they evaluate to take
precedence over the static values the corresponding attributes would
otherwise have.
GUM is a small DSL, used in some property keys of GEGL operations
to dynamically control UI attributes based on context. This commit
only adds an interpreter for the language; see the next commit for
the actual handling of the relevant keys.
See the comment at the top of gimppropgui-eval.c for a description
of the language.
Note that the interpreter is licensed under the LGPL.
If focus is on the layer list for instance, running this action from
right-click menu raises the on-canvas toolbar, ready for edition, with a
visible text cursor. But if the canvas has no focus (since you clicked
on the layer list, it has the focus), you still have to move your mouse
over and click the text on canvas. That doesn't make sense and there
would be barely any reason to use this over selecting the text tool then
clicking the canvas.
... non-text layers.
Since commit 10099bd, the action will be non-sensitive with non-text
layers, so if it happens, there is a bug somewhere. Therefore
g_return_if_fail() on this condition.
Moreover opening the edit attributes dialog was absolutely confusing on
what this action is supposed to do. We should not have these kind of
random behaviors.
When a layer is removed from a layer stack, and its backdrop needs
invalidating, emit the UPDATE signals for the backdrop only after
the layer has been removed from the container.
A subclass of GimpDrawableStack, for layer stacks. Invalidates the
layers' backdrop as/when necessary, according to the value of their
excludes_backdrop property.
Make gimp_drawable_stack_update() protected, instead of private, so
that we can use it in GimpLayerStack.
A boolean flag, specifying whether the backdrop is clipped to the
layer. That's the case when the layer's composite mode is dst-atop
or src-in.
This is a read-only property, derived from the other attributes of
the layer. We compute its value through a virtual function, so that
GimpGroupLayer will eventually be able to specialize it for pass-
through groups.
The next commit uses this property to actually do something useful.
Takes a layer mode and a composite mode, and returns the region
included in the composition.
Use this function in GimpOperationLayerMode, instead of testing
for specific composite modes directly. Will also be used by
the next commit.
Indentation cleanup in gimp_layer_modes.h
... so that we can use it for other functions that involve
compositing regions (which we do in the next commit).
Rename gimp_operation_layer_mode_get_affect_mask() and
friends to _get_affected_region().
With bigger and denser default images, the older 18px font default is
very small (~ 4pt font at 300PPI).
I decided to settle for a 15pt font, i.e. 62 px at 300 PPI, which seems
like quite an acceptable relative size for our FullHD defaults.
It is bigger than a default font size for —say— a text processor, but
GIMP is not really designed to process large walls of small text anyway.
Rather shorter texts at bigger sizes (i.e. designs, etc.). So this seems
like quite an ok default.
Note that as any defaults, this only goes as far and won't be what most
people need most of the time anyway. But at least we should get rid of
this ridiculous 18px default.
... tool when font unit is not pixel.
The default size is 18px. If for instance one is on a 300 PPI image, and
set the unit to mm, hitting "Reset to default values" sets to 213 px
(which corresponds to 18mm at 300PPI). I'm not 100% sure of the source
of the problem, but resetting the font size before the unit fixes it and
properly sets back to 18px. Let's go the lazy way and just do this.
Add the additional enum values to enum GimpSelectCriterion, and
the few needed lines to gimppickable-contiguous-region.c.
It's horribly slow, but works.
We check them into git, so this makes it easier to keep them in
sync when using a separate build directory.
Case in point -- this commit also syncs a few enum files that went
out-of-sync with their headers.
gimp_unit_get_digits() won't return relevant information when using
higher resolution. gimp_unit_get_scaled_digits() will provide the right
amount of details relatively to the actual print resolution of the
active image.
Add "import-raw-plug-in" to gimprc, and a new procedure
gimp_register_file_handler_raw(). On startup, remove all load
procedures that are marked as "handles raw" but are not implemented by
the configured plug-in. Add the list of available plug-ins to prefs ->
import/export. Register all file-darktable procedures as handling raw.
... brush features.
My previous commit only fixed aspect ratio reset, but I realize that
much more brush options are affected by defaults of generated brushes
(size, hardness, aspect ratio, angle and spacing). Let's reset all these
after the basic property reset.
Generated brushes can have specific aspect ratio, hence resetting to
fixed property defaults may be wrong. Therefore GimpPaintOptions needs
to redefine the reset() method from GimpConfigInterface.
Note: the specific "Reset aspect ratio to brush's native aspect ratio"
button was alright. But the broader "Reset to default values" was not.
Determines if the tool can pick at a given coordinate, without
actually picking. The default implementation uses
gimp_image_coords_in_active_pickable(); GimpFilterTool overrides
this function, to return TRUE whenever the active picker has
pick_abyss == TRUE.
Use this function when updating the cursor, and when determining
whether to draw the sample-average region indicator.
Add a boolean 'pick_abyss' parameter to GimpCreatePickerFunc. When
this parameter is TRUE, the picker should pick outside the bounds
of the drawable. Use FALSE for color pickers, and TRUE for position
pickers.
and use them for the new image in "Paste as new image". We were using
the resolution and unit of the image the paste command was invoked
from, which is entirely random and useless.
Change wording in Advanced Color Options drop-down menu to not imply
that GIMP isn't color-managed.
Slightly changed wording from Elle's original patch (mitch).
Also set the combo box' "ellipsize" property to END because it's too
wide. As mentioned before, this doesn't work for whatever reason,
hints are appreciated :)
when a click is detected by gimp_tool_check_click_distance(), only do
a motion() back to the button_press() coordinates if there has been a
motion at all since button_press(). This should not change any tool's
behavior, it's only an optimization to keep tools from doing useless
work.
The halftones transfer mode of dodge/burn uses pow(), which produces
NaN for negative input values. This tool doesn't really have OOG
values in mind, but using an odd power function fixes this issue,
and is in line with the behavior of the other modes w.r.t. OOG
values.
Spotted by José Americo Gobbo: when using a pattern source
for the clone tool, the top left corner was picked
as starting control point, which made the "fixed" aligment
rather useless. Using the pattern center allows the
better control of a pattern source.
Add a "mask-only" property to GimpBrushClipboard. When TRUE, only
create a brush mask (not a pixmap brush with mask and image).
Keep two clipboard brushes around: one classic "Clipboard Image" one
to be used as "stamp", and one new "Clipboard Mask" one that turns the
clipboard into a brush mask.
Fix the edit-paste and edit-paste-as-new-image PDB wrappers to use
gimp_get_clipboard_object(), not just clipboard_buffer(), and deal
correctly with entire images in the clipboard.
Fix brush shrinking used to compensate for the blur: avoid over-
shrinking the brush and changing its aspect ratio.
Change the way hardness maps to blur radius: hardness == 0 maps to
the largest radius such that, when the kernel is applied to the
middle pixel of the brush, the kernel is completely within the brush
bounds, taking brush shrinking into account, *assuming the brush is
a circle*.
Use the dimensions of the unrotated brush when calculating the blur
radius, so that rotation doesn't affect the blur amount (the blur
itself is not isotropic, though, and is applied after rotation, so
while the blur amount remains uniform, its effect does depend on the
brush angle.)
Get rid of the blur-radius upper limit -- it's fast enough to handle
large radii now.
A few additional minor speedups.
Also, make sure we don't overflow for large blur radii. Not a
problem yet, since the blur radius is capped, but soon...
Add a specialized convolution algorithm for the hardness blur. It
uses the same kernel as before, but performs the convolution in
(amortized) O(1)-per-pixel time, instead of O(n^2), where n is the
size of the kernel (or the blur radius).
Note that the new code performs the convolution in the input color
space, instead of always using a linear space. Since our brush
pixmaps (but the not masks) are currently perceptual, the result is
a bit different.
Add a debug procedure group, living in 'debug.pdb', which would host
useful debug helper functions. Functions in this group are not part
of the stable API, and may be changed at any point.
All procedures added to 'debug.pdb' should have a 'debug_' prefix,
and use the new std_pdb_debug() macro, which adds the proper "here be
dragons" warning to their description.
Add two debug procedures: gimp-debug-timer-start() and
gimp-debug-timer-end(), which measure elapsed time, a la
GIMP_TIMER_{START,END}, and can be used to profile script-fu
commands.
If the transformed item is a layer, and we are transforming the entire
layer (if there is now selection), hide the original layer during the
interactive transform. Based on a 2.8 patch from saulgoode.
Commit 9d4084c82f skips conversion and
blending of (some) transparent source and destination pixels. When
`blend_out == blend_layer`, it banks on the fact that the alpha values
of `blend_out` would be the same as those of `blend_layer`, and hence
the same as those of `layer`; thing is, we only copy those values from
`layer` to `blend_layer` for the pixels that we *don't* skip, so this
assumption is just wrong :P This leaves us with bogus alpha values in
`blend_out` for the skipped pixels, when the above equality holds.
For composite modes that use the alpha values of `blend_op` (aka `comp`)
even for transparent input pixels (i.e., src-atop and src-in), this may
result in artifacts.
Fix this by simply initializing the alpha values of `blend_out` for
skipped pixels unconditionally.
The expression `src_offset_x - coords->x + origin->x` is parsed as
`(src_offset_x - coords->x) + origin->x`; since floating point
arithmetic is not generally associative, even when
`coords->x == origin->x` (in particular, when there is no active
symmetry), it may still yield a different result than plain
`src_offset_x` if there's not enough precision for the intermediary
result (which is usually the case when `{origin,coords}->x` is
noninteger.) Since `src_offset_x` is an integer, and since the result
of this expression is rounded to an integer, if the error happens to
be in the direction of the rounding, it's magnified to a whole pixel,
which causes visible "jitter". (Ditto for `src_offset_y` and co.)
Regardless of this issue, we want to individually round `origin->[xy]`
and `coord->[xy]` down before taking their difference, since the
original offset is calculated according to rounded-down coordinates.
This solves the original issue along the way.
We don't support subpixel source sampling, so there's no use in
pretending that we do. Demoting everything to int as soon as
possible helps guarantee that these values are at least rounded
properly and in fewer places.
Make sure we always round coordinates down, and not toward zero.
Keep using floats only in the signatures of the relevant PDB
functions.
Copy TransInfo arrays around using memcpy(), use memcmp() to
compare them, add a function to allocate one. Clean up some
logic in gimp_transform_tool_check_active_item().
and a GimpSettingsBox. This brings savable settings to all ops, also
the automatic ones in the GEGL tool. It also makes the code cleaner
and more general.
... have been saved.
No need to keep a list of 0 images when the creator requested a quit or
close-all actions and manually went through the list to save all
remaining images. Yet one can still cancel the quit/close-all action by
hitting Esc (or Cancel button) during the last save, since it is an
idle source action.
GimpFilterTool::get_settings_ui() is no longer needed, replace
it by a simple utility function in gimpfiltertool-settings.c.
Also, use the GimpFilterOptions functions added earlier, and some
random cleanup.
and add them as return values to GimpFilterToool::get_operation(), so
the tools is configured entirely per-instance now.
This makes get_operations()'s signature more evil, but helps making
GimpOperationTool less conplicated and convoluted.
We were still saving channel colors in 8 bit, this additionally
saves/loads the color as float values. Still save the old PROP_COLOR
for compatibility.
from gimpdir/tool-options/ to gimpdir/filters/, and only if moving
fails try reading from the old location as fallback. We don't normally
move files around, but this one-liner nicely avoids cluttering
gimpdir.
Filters settings used to be serialized and deserialized only
when a filter tool's GUI was shown, too late for the code that
re-runs/re-shows filters with previous values.
Move the entire loading/saving code to gimp-operation-config.c, even
adding/removing the dummy separator item between timestamped automatic
history and manually saved settings. Load the settings automatically
when a settings container is requested, but still trigger saving from
the few places the container is changed in the GUI; could also
automate that later.
This commit also moves all settings of filters that have their own
tools from gimpdir/tool-options/ to gimpdir/filters/. Add compat code
to try the old filename if the new doesn't exist, so files are
migrated automatically.
WIP, but this step already fixes the bug.
Both in the GimpImage API and in the GUI. The toggle in the save
dialog now controls ZLIB compression directly. Changed the various
info labels accordingly. Ditch the XCF parasite that saved the XCF
compat mode.
Commit 4beff2f was basing it on the screen y PPI but that is not really
consistent or logical actually. Since the actual stroke dialog uses the
y resolution of the current image, it makes sense that the generic
stroke defaults in the preferences should use the y resolution of the
default image.
Enable 64 bit file offsets in XCF files, starting with newly added XCF
version 11.
We use at least version 11 if:
- we would use the previous version 10 (essentially skipping 10)
- the in-memory size of the image is larger than 4 Gig
Change the xcf_read_foo() functions to take the XcfInfo* instead of
a GInputStream*, and make them advance the info->cp offset by
themselves. Makes xcf-load.c a lot more readable.
Step one, without changing anything in the saved XCFs yet:
Abstract reading and writing of file offsets away into their own
xcf_read_offset() and xcf_write_offset() functions, which take
"goffset" instead of "guint32". Also change xcf_seek_pos() to take a
goffset argument.
Change all file offset variables in xcf-load.c, xcf-write.c and struct
XcfInfo to goffset, and add new member "bytes_per_offset" to XcfInfo,
which is currently always 4.
This value could be based on either the x or y resolution, or maybe some
kind of mean values. It could also be based off the print resolution of
any image (if the user sets a physical dimension, the actual pixel width
will vary depending on the print resolution). There is no actual "good"
answer here. But any of these values will be better than a default 96.0.
The abbreviated commit hash we show in the shell and the about
dialog is currently just the last 7 characters of 'git describe',
based on the assumption that abbreviated hashes are always 7-digits
long. When the hash is longer than that, we're just showing a
nonsense commit.
This was never a good idea, since users can override this, and
since disambiguation can result in longer hashes, but since git
2.11, the default abbreviated hash length is determined based on
the size of the repository, which currently results in 10 digits
for us.
Let's just do it right.
It is a little fuzzy whether expected or not, architecturally-wise. On
one hand, I can see some core/ header includes under config/. Though on
the other hand, app/Makefile.am explicitly sorts config/ below core/.
Anyway let's just use g_cclosure_marshal_VOID__VOID which is the same as
gimp_marshal_VOID__VOID and get rid of the include.
This fixes builds from scratch.
It was agreed that we should write "plug-in" consistently. Only possibly
user-visible strings were updated.
Thanks to scootergrisen for a first patch which could not make it
after changing decision on the canonical writing.
Current code only guess resolution for a single monitor. Ideally
the widget sizes could be different depending on the window where a
given widget is on. But that's a start.
In particular all rotate/flip actions can apply to an image or drawable.
Let's make it clear, especially when it is run out of the menu context,
for instance in the action search.
which unlike HSL Lightness is actually physically meaningful and
also generally speaking much more useful than HSL Lightness.
Change "Lightness" to "Lightness (HSL)" to make it clear that
the "Lightness" in the Colors/Desaturate/Desaturate menu is not the
same as "Lightness" in LAB/LCH.
For completeness add the option to desaturate to "Value (HSV)".
Add links in app/operations/gimpoperationdesaturate.c
to the Wikipedia article with definitions of L/I/V in HSL/HSI/HSV.
Use gimp_item_translate() instead of gimp_item_set_offset() to set
a pasted item's position, so that the offsets of pasted layer group
children are updated correctly. Otherwise, only the group itself
moves, while its children remain positioned relative to the top-left
corner of the image.
... since that's the color space it actually works in.
Keep the legacy "Color (HSV)" mode's name as is, wrong as it is,
since, well, that's what it used to be called...
No need to do full back and forth RGB/HSV conversions.
Change the behavior such that fully desaturated values remain
desaturated, instead of saturating towards red.
Pixels whose source or destination alpha is zero are not blended, and
therefore do not need to be converted between the composite and blend
spaces (assuming a conversion is necessary to begin with.) When there
is a large enough segment of consecutive pixels that don't need
blending, split the conversion/blending process around it, so that
we don't convert too many unblended pixels unnecessarily.
For layers with lots of transparency, this can dramatically reduce
compositing time; for layers with no transparency, the added
overhead is rather negligible.
When adding an item to a filter stack that doesn't have a graph yet,
calling gimp_filter_stack_update_last_node() may ultimately lead to
the invocation of gimp_filter_stack_get_graph(), which would create
a new graph, and add the item's node to it; gimp_filter_stack_add()
would then erroneously attempt to re-add the node to the graph.
Fix this by calling gimp_filter_stack_update_last_node() after
(potentially) adding the node to the graph in
gimp_filter_stack_add().
The function has been unused since commit b5cc2a9.
Its work is now divided into gimp_tool_palette_set_toolbox() and
gimp_tool_palette_hierarchy_changed().
Argh! Always triple-check commits before pushing!
Commit e30c92c had completely wrong big values, which I was using
to make tab border update really visible while testing.
Allow overriding icon sizes set in themes from the preferences.
This initial commit updates only toolbox icons. More to come.
4 options are available: small, medium, large and huge (the later would
likely be useful for HiDPI screens).
Uses a new widget GimpIconSizeScale.
Merge mode lays the source layer on top of the destination, same as
normal mode, however, it assumes the source and destination are two
parts of an original whole, and are therefore mutually exclusive.
This is useful for blending cut & pasted content without artifacts,
or for replacing erased content in general.
Calculates the dot product of the two input colors, and uses that
as the value for all the output color's components. Basically,
a per-pixel mono mixer.
Useful for custom desaturation, component extraction, and crazier
stuff (bump mapping!)
Include erase mode in the menu for layers and general paint tools.
This makes the eraser tool somewhat unnecessary, but allows for
interesting use cases (e.g., airbrush eraser, etc.)
... possibly due to small win32 stack
Limit the number of samples processed in one go by gimp_composite_blend()
so that we don't overflow the stack when we alloca() buffers on it.
... and get rid of the dedicated op. This gives us support for all
the blend/composite options for this mode.
Rename COLOR_ERASE to COLOR_ERASE_LEGACY, with perceptual blending/
compositing and immutable everything, and add a new COLOR_ERASE
mode, defaulting to linear blending/compositing, with mutable
everything. Modify affected code.
These are more general, and more expensive, versions of the non-
subtractive compositing functions. They are used with modes that
specify the SUBTRACTIVE flag. This doesn't affect anything yet, but
the next commit ports color-erase mode to a blendfun.
Most modes only modify the *color* of overlapping dest/src regions,
however, erase and color-erase may also reduce their *alpha*, i.e.,
eliminate some of the overlapping content. Flag these modes with
the new SUBTRACTIVE flag, as they require more general compositing
code. The next commit adds the said code.
The mode group switching combo box is hard to discover, until we use the
default group instead of legacy group as default - it is better to make legacy
resemble the full old set to.
Try to sort all GIMP_ICON_* defines into FDO categories like in
https://specifications.freedesktop.org/icon-naming-spec/latest/ar01s04.html
Add defines for all icons we override, rename some icons to their FDO
standard names, and mark the ones we duplicate with a comment so we
don't forget to rename those to standard names in 3.0.
In DEs which use a global menu, such as Unity, updating the menu
can be expensive. This particularly affects canvas scaling and
rotation, for which updating the menu synchronously causes notable
lag.
Remove all stock items added since 2.8, restore accidentially removed
ones, and rename the newly added GIMP_STOCK_* defines to GIMP_ICON_*.
(will move to having GIMP_ICON_* defines instead of magic hardcoded
strings for all icons).
being exported to libgimp, and having a non-exported value, this is a
horrible mess like with GimpLayerMode, but at least the cruft value
names are deprecated now.
Need to provide the pixels in a format that matches the profile,
simply using "RGBA float" here was a brain bug of mine. Two profiles
and two formats are parameters the used GimpProfileTransform needs to
work correctly.
.. due to gdk_pixbuf_scale() with themes using the pixbuf engine
Make GimpDisplayShell a subclass of GtkEventBox, so that it gets its
own window, isolating its events from those of its ancestors.
In particular, the "expose" event handler of GtkNotebook, which the
shell is a child of in SWM, is particularly slow with themes that
use the pixbuf engine. If the notebook and the shell use the same
window, this can cause notable, and somtimes severe, lag when the
rulers or scrollbars are updated frequently, such as when rapidly
moving the cursor.
which got renamed from create_map() in this commit too. "Completely"
means including insane options like color_managed and gamma_hack, they
are confusing enough so they should at least work correctly.
Need to convert both from the drawable's profile to the filter's input
format and from the filter's output format back to the drawable's
profile. This change fixes things for the case where the filter's
input and output formats are different.
which returns an array of modes in the order they would appear in a
GimpLayerModeContext's UI (like tool options or the layers dialog),
without the separators.
Use it in context-commands.c and layers-commands.c instead of static
(and outdated) arrays for the actions that cycle through modes.
In particular, this enables grids whose points of intersection
are at the middle of the image's pixels, which is useful for
undistorted painting with odd-sized brushes using tools other than
the pencil.
This commit also changes the grid visibility behavior, so that the
the visibiltiy of horizontal and vertical grid lines (depending on
the zoom level) is independent.
Instead, add a gimp_layer_mode_get_format() function, which takes
the layer mode, composite space, and blend space, and returns the
I/O format.
Currently, we always use the composite space format as the I/O
format. This simplifies gimp_composite_blend(), and gives us
composite-space support for the "special" layer mode ops for free.
Replace the 'with-behind' and 'with-replace' properties with a
single 'context' property, and use it to select the included
layer modes, according to their context mask.
Add a dummy GIMP_LAYER_MODE_SEPARATOR value to the GimpLayerMode
enum, and use it to explicitly mark the menu separators in the
layer-mode group arrays; add separators to the layer-mode menu
accordingly.
Update the rest of the code to use 'context' instead of 'with-behind'
and 'with-replace'. In particular, in the layers and layer options
dialogs, select the right context based on whether or not the
selected layer is a group.
A bitmask, specifying in which contexts a layer mode is applicable.
Can be a combination of:
- LAYER: usable as a layer mode for actual layers.
- GROUP: usable as a layer mode for layer groups. Currently, all
modes that specify LAYER also specify GROUP, and vice versa,
but the planned pass-through mode will be GROUP only.
- PAINT: can be used as a paint mode.
- FADE: can be used for fading.
Add a 'context' field to _GimpLayerModeInfo, and provide context
masks to all the modes.
Use the context mask for validation when setting a layer's mode.
The next commit will use the mask when populating the layer mode
menus.
Add a "paint_composite_mode" field to GimpLayerModeInfo and set the
mode of the eraser to SRC_ATOP. Defaulting to SRC_OVER for all
painting didn't quite do it for all modes.
set all legacy modes to completely immutable and the LAB modes'
blend mode to immutable. Change GimpLayer setters and the UI
accordingly. Remove the LAB color spaces from the GUI, they can
only be used with the LAB blend modes anyway and not changed.
Nobody has them anymore, and they are deprecated in GTK+ 3.x. This
also fixes all conflicting mnemonics except those I missed, but we can
fix them now.
Fix cage transform to refuse to work on locked, invisible and group
layers. Add GimpTool::initialize() implementation so the generic
"drawable has changed" mechanism triggers the right response.
They can be affected by the same problem described in
commit 4c3a772cd8, although in the
case of SRC_ATOP, the affected pixels are always fully transparent.
When the source alpha is zero, we don't calculate the blended color,
so `comp[b]` can be infinite or NaN, in which case the expression
`in[ALPHA] * (comp[b] - layer[b])` is NaN, rather than the expected
value of zero.
and to operations/layer-modes/, respectively.
Add gimp_layer_modes_init() which asserts on the correct order of the
GimpLayerModeInfo array, and switch to accessing the array directly in
gimp_layer_mode_info().
Similar to the Photoshop mode of the same name. Assigns
either 0 or 1 to each of the channels, depending on whether the
sum of source and destination channel values is less than, or
greater than (or equals to), one, respectively.
This is equivalent to inverting the source, and using it to perform
per-pixel, per-channel threshold against the destination, which is
useful for various effects.
Stuff like passing "input" directly if "aux"'s opacity is 0, etc.
Used to be partly handled by normal mode, even though it applies
to other modes too.
Adjust the logic for the new compositing modes.
Add a GimpLayerModeAffectMask enum, and a corresponding
get_affect_mask() function to GimpOperationLayerMode, which
specifies which of the op's inputs, if any, are affected by the
mode, apart from the overlapping regions. Most modes affect only
the overlapping regions, but dissolve and replace also affect the
rest of the input. This information is used for determining if
the optimizations are applicable.
Some of the generic files still contain SSE2 code, in particular
gimpoperationlayermode.c. The reason why it often works without is that
native gcc will usually pre-define SSE macros by default.
To check this: gcc -dM -E - < /dev/null | grep SSE
Yet I had a case on a small netbook where the SSE macros were not
pre-defined even though supported. Consequently the build failed.
Largely based on a patch by Ell, with the enum type renamed and
various small changes. Adds another axis of configurability to the
existing layer mode madness, and is WIP too.
The colon-separated list used in the plugin originally comes from
gimp_help_get_locales() anyway. My previous code was using different
lists of locales in different places, which was inconsistent.
I know this looks absolutely horrible, please spare me comments about
that. This commit has the purpose to let everybody experiment with the
new modes, and suggest improvements of the GimpLayerModeBox widget; we
need *some* way of controlling the new layer mode madness.
These variations on darken only and lighten only have the advantage over the
componentvise versions that they always use the full triplet of either original
or new layer - meaning no new colors/hues will be introduced. This is similar
to how these modes operated/operates in picture publisher and photo-paint.
- set the filter_tool->drawable before showing the tool gui.
- set the sensitivity functions for channel combobox of threshold,
levels and curves tools dialogs only once during dialog creations.
- use the filter_tool->drawable inside the sensitivity functions
C++ won't allow us to use GimpLayerMode in the API where we used to
have GimpLayerModeEffects.
Move GimpLayerModeEffects to libgimpbase/gimpcompatenums.h so it's
not in the API any longer, and instead typedef and define stuff in
libgimp/gimptypes.h, and adapt the compat enum registering code
accordingly.
The porting from 8bit per component scaled some 8bit fractions up to huge
floating point numbers, this works for most values but causes trouble for near
transparent pixel values. This commit copies the inner blend loop from the new
divide layer mode, but keeps the old compositing logic.
... is fully transparent, instead of just src.
The blend func results only affect the intersection of dest and src.
Run time is currently dominated by the compositing step for most modes,
so the difference in performance is pretty negligible, but it does make
a difference for the more expensive modes, like the HSV ones.
The print size displayed in image property and title format should use
gimp_unit_get_scaled_digits() instead of gimp_unit_get_digits() and
adding 1, which is quite random or magic number-y.
For operations needing to override default behavior sub-classes should still be
used.
This commit also enables pinligh, vividlight and linearlight blend mode modes
We can't just switch to a GimpOperationTool by using the normal
gimp_context_set_tool() or gimp_context_tool_changed() because it
needs additional initialization like setting an operation at all.
In gimp_gegl_procedure_execute_async(), g_object_set_data() the used
procedure on the newly created tool.
In gimp_display_shell_initialize_tool(), when we re-create the active
tool because of a drawable change, check for the procedure and invoke
it again, instead of simply creating an empty operation tool by
calling gimp_context_tool_changed().
Optionally convert all imported (not XCFs) images to 32 bit linear
floating point, and optionally add a little noise in order to
distribute the colors minimally. The new options are on a new "Image
Import & Export" prefs page that needs a new icon. Original dithering
patch by pippin.
Since CIE Lab is one of the supported color spaces for doing the blending -
this enum is not only about transfer functions/curves or gamma. This finishes
already started cleanups.
... in status bar.
Follow-up of commits f836892 and d1c3c3d. With high resolutions, the
distance displayed in the status bar when shift-clicking in a paint tool
may lack digit precision and show the same value (in non-pixel unit) for
several consecutive pixels. Compute a more accurate precision than what
gimp_unit_get_digits() can provide in such cases.
When working with high resolution, you may have cases where measured
length won't be displayed with enough digits; i.e. several pixels length
would show up as 0. For instance at 4000 PPI, up to 7 pixels show up as
0 mm, then at 0.1 mm from the 8th pixel (actually reaching over 0.05 mm,
approximating as 0.1), then 0.2 at 24 pixels (actually: 0.152), and so
on. At such a resolution, 3 digits are needed for 'mm' instead of the
1 digit returned by gimp_unit_get_digits() so that we display reliable
lengths.
Therefore we need to compute ideal digit precision. Configured digits
for a given unit will now be used as a minimum value, but actually used
digits may be higher.
Compute the ideal decimal precision for cursor position and length
status so that you get the best precision on physical units depending
on the current resolution, yet avoiding over-precision (which can be
misleading). The unit's "digits" value is now used as a minimum
precision only.
It is used both for blending and compositing, the repeated use of the word
BLEND in code made the logic involving both blending and compositing hard to
read.
Note that in some cases the alloca may be unnecessary, but this keeps
the code clean, and we have to be able to *potentially* do the alloca
anyway, so what the hell.
Implement a common utility function gimp_blend_composite that uses utility
functions for implementing layer modes, with separate (possibly SIMD) optimized
loops for blending and compositing, with configured linear TRC, perceptual
gamma TRC or even using CIE Lab as the space.
from gimp_applicator_new() and gimp_gegl_mode_node_set_mode().
Compositing doesn't depend on the layer format any longer, only on the
layer mode. Painting with "use applicator" unchecked is still broken
in some cases and needs more fixing.
like all other function typedefs and add GimpBlendFunc typedef which
will be needed soon. Also rename get_layer_mode_function() to
gimp_get_layer_mode_function().
...layer group cause a bug in the existing layer size
Change gimp_group_layer_get_size() to return FALSE if there are no
children (there is no content).
In gimp_group_layer_update_size(), skip children where get_size()
returns FALSE. Fixes bogus size calculation.
Temporarily slower - but permits paint modes like overlay/softlight to work at
all for linear TRC pixel encodings. Should be reverted when the non-graph
approach works properly again.
with proper value names. Mark most values as _BROKEN because they use
weird alpha compositing that has to die. Move GimpLayerModeEffects to
libgimpbase, deprecate it, and set it as compat enum for GimpLayerMode.
Add the GimpLayerModeEffects values as compat constants to script-fu
and pygimp.
Make overlay, Lch color, Lch hue, Lch saturation and Lch lightness mode handle
alpha more like how normal does it. This is a change that we ideally might want
applied to other layer modes as well to get rid of MIN() calls.
when parsing of an object property fails, we need to set *expected to
G_TOKEN_NONE to tell the config parser that something has gone wrong,
or it will continue parsing and run into trouble with the inconsistent
state (it will try to set an error over the already set error, causing
a warning).
The "accumulated" (color value) adjective does not look like the best
word in both descriptions. Respectively, the first is an averaged
value whereas the second would be a "merged" or "composited" value.
If native speakers have comments or better propositions, they are
welcome to speak up.
...regardless of image color precision
Split enum value GIMP_COLOR_FRAME_MODE_RGB into RGB_PERCENT and RGB_U8,
which display the current % values, and values in a range of 0..255.
Move the entire drawing control logic to gimp_color_tool_oper_update()
which gets invoked on hovering, and don't mess with it in
button_press() and button_release(). Tested to work with the color
picker, paint and filter tools.
Disable the new "automatic window tabbing" feature introduced on macOS
Sierra. It breaks GTK+ applications and we would need proper support for
this in GTK+ if we want to use it.
Change GimpHistogram to take a "gboolean linear" parameter and always
honor that parameter, so both kinds of histograms can now be created
for all drawables.
Add a horrible "Linear" toggle to the histogram dockable which always
defaults to the active layer's actual pixel format, but can be
switched at any time. This UI is ugly and needs to change.
On the PDB, default to gamma-corrected if the plug-in is unaware of
higher precision, and to the drawable's native pixel format otherwise.
Other places using histograms (e.g. levels, curves) are unchanged.
Set gimp_tool_control_set_preserve() to TRUE and set an appropriate
dirty_mask, just like all tools which have a permanent on-canvas state
outside of a simple press-drag-release.
The white, gray and dark sliders of GimpHandleBar have a black contour.
This makes the white and gray slider visible even with similar colored
background. On the other hand, the black slider is barely visible on a
dark background (and could even be made totally invisible using the same
color). So let's use a light-gray contour on the dark slider, making now
all sliders working with any background color.
gimp_dialog_config_fill,stroke_options_notify(): ignore notifications
on the fill and stroke option's parent class properties, they are not
serialized and completely irrelevent here.
This reverts commit eab4929a78.
Oups it would seem gtk_accel_map_change_entry() could return FALSE even
when the expected shortcut is correctly set (my guess is that it was
already the same shortcut, so indeed no "change" happened, though it is
not a failure either; yet I haven't checked if that is the actual
reason).
Let's just revert this. It's not always a good thing to be too thorough!
Sorry for this!
These are used for display switching, and even though you could
remap these shortcuts, it would work only until you close an image,
open a new one, or reorder tabs in SWM, in which case your shortcut
would end up forcefully overrided, which is a bad user experience.
If we want to give more flexibility and allow one to map these
shortcuts, we must also make sure the display showing actions won't
override customized shortcuts. In the meantime, it is better to
simply forbid these in our preferences.
... for "windows-display-*" actions.
I should not happen, but let's be thorough and properly handle failure
with a message since this is a runtime issue.
gtk_action_group_remove_action() removes the action from the group while
not actually cleaning any accelerator. This is a problem for transient
actions which have only a meaning within the current session, such as
the display switching actions named with the display ID (unique within
the session only).
Current commit, combined with the previous one (commit c0ee959), fix
"windows-display-*" actions being saved inside menurc.
Some actions are not meant to be saved, in particular the
"windows-display-*" which have only a meaning during a same session
since display IDs are session-dependent. So let display deletion
happen first so that proper cleaning of action is done when writing
menurc.
Revert "app: action search should search accross all available actions."
This reverts commit 53b3673bd8.
Had to revert these two commits, quoting comment 6 of the bug:
Thinking again, that entire change is unfortunately wrong, the only
right UI manager is in fact
gimp_ui_managers_from_name ("<Image>")->data
because it's the global popup <Image> UI manager which is independent
of a display and it always in the right state for the currently
active image, all other UI managers are wrong.
Current defaults are from another time. Acceptable defaults could be
common screen resolutions. 1366x768 is apparently the most common
(according to various stats on the web), but since we target advanced
graphics artists, let's go for 1920x1080, which is the second most
common resolution, also known as Full HD.
For unstable builds, let's have at least one odd number, uncommon ratio
and higher values, encouraging tests with less common numbers and bigger
images. I chose 2001x1984. Feel free to update to any other funky values
following these "unstable" rules.
... with something more suitable.
72 PPI is from a time where people thought this was a common screen
resolution. This is not the case nowadays, and anyway images targetted
for screen display should not bother with PPI resolution at all, only
with actual pixel dimensions.
PPI resolution is more useful for printing. And for this case, 300 is
quite an accepted OK value for most cases. So this is likely a better
default for GIMP.
Add GimpMenuFactory can always be found in a GimpDialogManager, use
gimp_dialog_factory_get_menu_factory() where we have a dialog factory
instead of accessing "global_menu_factory" or redundantly passing a
menu factory around where we already pass around a dialog factory.
<Dockable> has this whole list of actions named similarly to dialogs-*
actions, and we don't want these to take precedence, especially since
they would always create a new dock instead of just showing the existing
one if already present.
Also fix the redundancy check on already added actions.
There is no reasons not to translate the "occurs" text which will be
in the color names in palettes extracted from images. It indicates on
how many pixels a given extracted color was occuring. We should also
translate the full string (not just "occurs") since some language will
have to reorder words, and may even use different bracket characters.
So I also add a translator comment to make sure the translators get
what the %s and %d stand for in this string.
Duplicate accelerators are not supposed to happen. It is not possible
to set them through the GUI in particular. Nevertheless
gtk_accel_map_load() would apparently let duplicates pass, which could
happen after editing the menurc directly, or using the development
version (no action name migration happens there), or simply after a
potential bug. This is then very annoying because you may see several
actions displaying the same shortcut but only one actually work. And
trying to re-set through GUI the shortcut to the one action you wish to
run does not fix the duplicate issue (you have to laboriously find which
other action use the same accelerator and delete it first).
Better be safe than sorry and make a quick check at startup, then delete
the accelerator on one of the duplicates (you can't guess which one was
actually wanted, but at least you will facilitate manual reset through
the GUI).
Redundant accelerators were:
- <Primary><Shift>y on dialogs-mypaint-brushes and edit-strong-redo.
Since the <Primary>z vs <Primary>y has quite a strong history for
undo/redo actions, and dialogs-mypaint-brushes is quite new, let's
unmap the latest.
- <shift>l on tools-seamless-clone and tools-unified-transform.
Since the Seamless clone tool is still in the playground and we
don't even know if it will make it out quite soon, let's give
priority to the Unified Transform tool.
Limit selection shrinking to MIN (sel_width, sel_height) / 2, larger
values make no sense.
Limit selection bordering to the same value and growing to
MAX (image_width, image_height).
Introduce virtual function GimpViewable::is_name_editable() and class
member "gboolean name_editable" for the default value. Default to
FALSE and only return TRUE if the name can actually be edited by the
user.
When attemting an edit, check the new API and beep instead of starting
the edit.
Don't rely on the exact modifier being pressed or released. Instead,
check if only the right modifier is pressed after *each* modifier
change, and switch to color picking if it is; disable color picking
otherwise. This greatly reduces the risk of missing the user's wish to
pick colors because of other modifiers being pressed and released in
whatever order.
Probably fixes bug #734743.
We handle multi-selection by letting GtkTreeView handle button press
when the widget is in GTK_SELECTION_MULTIPLE mode. Change that code to
only do that when one of the participating modifiers (shift and
control on Linux and Windows, shift and cmd on macOS) is pressed.
This makes sure that the same thing is not randomly handled by two
different pieces of code, and probably fixes bug #738440, tho I can't
be sure.
When calculating an overlap between two ranges, interpolate the hue
adjustment from config->hue[primary_range] and
config->hue[secondary_range] BEFORE mapping it to the input value.
This fixes odd edge cases where only one of the ranges crosses the
red/magenta wraparound, or if adjustments to different channels yield
more than 180 degree difference from each other.
In the "New Image" and "Convert Precision" dialogs. The "Gamma/Linear"
switches get adjusted automatically to the new defaults, but can be
changed manually.
Don't offer dithering options when converting to 16 bit, it doesn't
make much sense to dither for anything but 8 bit. Thanks to Elle for
testing that this assumption is indeed true.
Use a properly commented #define instead of just a hardcoded "8 bits"
to make clear that this is a pure GUI "restriction".
Disable the convert precision dialog's dithering controls when
converting to higher bit depths o to anything > 16 bit.
Make sure the disabled dithering widgets always says "None", which
implies duplicating the bit depth checking logic in both the dialog
and its callback, in order to protect the values in GimpDialogConfig
from being overwritten by NONE.
The new properties are "virtual", they share their storage with the
"precision" property and are not serialized, they are meant for GUI
property widgets.
...instead of transforming it
Add gimp_matrix3_will_explode() which determines if a transform
matrix will blow up something in a rectangle to infinity, and use
the function so set both the GIMP and GEGL code paths to clip the
transform to the input size.
This got lost when dropping the file-uri plug-in and switching to
interal download using GIO. The file-uri plug-in would just download
whatever remote thing and open it locally, using local magic matching.
We now do the same internally and try the magic matching again on the
downloaded file.
Don't migrate any paths from an older gimprc. This unfortunately kills
all manually added paths, but makes sure all paths default to the
files of the new GIMP version, could be improved by parsing the path
and replacing only the right elements.
Add color management to GimpDrawableFilter and GimpFilterTool, GEGL
ops applied to drawables can be applied in color managed space
now. Sadly, this is very slow, so disabled by default.
I'm sure the profile guessing based on the operation's format doesn't
always work, but this general bug counts as fixed now.
Fix cursor rotation jump when tablet pen is tilted horizontally.
This changes to the correct values for the special case of tilt_x == 0.0.
Reviewer (Jehan)'s note: computing the convergence of the tilt function
in the `else` block, when tilt_x approaches 0, tilt value indeed
converges to 0.25 with tilt_y > 0 and 0.75 (or -0.25 which is the same)
with tilt_y < 0.
Initialization was failing when a paint buffer could not be returned for
a stroke (which was bound to happen in tiling). Instead it must just
ignore the coordinates which won't result in painting, and continue to
the next ones.
gimp_drawable_equalize(): is mask_only is FALSE, suspend the selection
around gimp_drawable_apply_operation() so the operation affects the
entire drawable.
The code was still checking for "plug-in-recent-*". Also, rename the
actions to "filters-recent-*" instead of "filter-recent-*" for
consistency with the other filters actions.
so the threshold can now be based on the GimpHistogramChannel enum.
Add a channel menu to the threshold dialog and a channel argument to
the PDB procedure (which is new in 2.10).
If I hadn't forgotten what the "RGB" channel is supposed to do I would
have implemented the RGB mode in GimpOperationThreshold correctly.
Right now I'm just guessing. Anyone?
They used to be 0..255, inherited from the old gimp_histogram() and
gimp_threshold() procedures. This commit deprecates these old
procedures and changes the ranges in the new gimp_drawable_histogram()
and gimp_drawable_threshold() to double with a 0.0..1.0 range.
Add property "color-tag" of type enum GimpColorTag to GimpItem so all
layers, channels and paths can be tagged with a color.
For interoperability, use the color list from Krita which is a
superset of Photoshop's colors.
Features a "Color Tag" submenu in the layers, channels and paths
menus, a row of color radio buttons in the properties dialogs,
undo and PDB API.
As a side effect, some common code is now factores out into
items-actions.[ch] and items-commands.[ch] which adds visible, linked
and lock actions for layers and channels.
Introduce item-options-dialog.[ch] which abstracts this away and use
it from the layer, channel, vectors options dialogs. This is all
pretty ugly but better than duplicating that code three times. The
vector-options-dialog is now completely pointless but I kept it anyway
for now, let's see what unique path options we come up with.
Convert GimpRGB to CMYK using a color transform to the configured CMYK
profile instead of the naive gimp_rgb_to_cmyk().
Add gimp_color_frame_set_color_config() and call it on all color
frames in the GUI (color picker tool, cursor info, sample point view).
Keep a GimpColorTransform around that does the conversion.
Also color manages the frame's color area now (visible in the sample
point view), which was forgotten earlier. Addresses bug #467930.
Fix the check that keeps events on overlay widgets from entering the
tool event mechanism, they have no business there.
gimp_overlay_child_realize(): set the embedding widget's event mask on
all overlay children, so their windows will be used as event window,
so their events become distinguishable from events on the parent (the
canvas).
gimp_display_shell_canvas_tool_events(): fix the check for events on
overlays, and skip them for real this time.
Call gimp_context_tool_preset_changed() on the global user context,
not on the tool preset list's or tool preset editor's local
context. Fixes restoring of presets in MWM. Tracked down by Jose
Americo Gobbo.
Have "Save" and "Restore" buttons in both the tool preset list/grid
and the tool preset editor. The save button stores the active tool's
options in the preset, if possible.
This fixes restoring of brush properties (size, spacing angle etc.)
from presets, which was utterly broken before. The fix consists of
two parts:
- In tool_manager_preset_changed(), always copy the brush properties
again after setting the preview on the tool options, in order to
override brush properites that get copied from a linked brush when
that brush gets set on the tool options.
But no amount of copying stuff again and again would help without:
- In gimp_context_set_by_type(), don't use g_object_set() to set the
object (brush, pattern etc.). Instead, build a GValue and call
gimp_context_set_property(). This may seem odd, but avoids a
g_object_freeze_notify()/thaw_notify() around the g_object_set(),
which was causing "notify" to be emitted at the very end, after
everything this context change has triggered. GimpContext is an
essential core object and there is an expectation of a reasonable
order of signal emissions and callbacks being called. The "notify"
at the end was keeping any callbacks of the context's "foo-changed"
signals to override anything an earlier callback had done, if a
"notify" callback was overriding that overriding again.
This was probably the reason for a lot of odd behavior observed over
the years. In fact, I have been searching for this for at least 5
years.
...are not submitted to respective Tool Options sliders
Treat brush angle and aspect ratio like all other paint options values
that can be linked to brush defaults and take their default values
from the brush. They were special casing their defaults to constants,
and GimpBrushGenerated was adding the passed dynamic radius and aspect
values to its own. This was totally incomprehensible.
Now GimpBrushGenerated's transform_size() and transform_mask()
implementations just translate between these APIs value ranges and the
brush's own value range and only use the passed values (not the
brush's native values), which makes the editor <-> tool options
interaction and the painted brush shape predictable.
Also connect the active brush's property notifications to the paint
options properties, so the paint options follow a brush edit live if
the respective "linked" toggles are checked.
And some cleanup.
Add a GimpFillType argument to GimpItem::resize() and fill type
widgets to the canvas and layer resize dialogs. Fill the new parts of
the drawable according to fill type in gimp_drawable_resize(). Make
sure places that need the old behavior get GIMP_FILL_TRANSPARENT
passed by hardcoding it in the GimpItem::resize() implemetations of
channel, mask, selection etc.
We didn't convert patterns to the target drawable's profile when using
gimp_drawable_fill().
Introduce gimp_drawable_fill_buffer() as single filling utility
function that does things right and use it from gimp_drawable_fill()
and gimp_fill_options_create_buffer().
Use a GimpSpinScale in channel-options-dialog.c and clean up the
dialog layout. Affects the "Quick Mask Attributes", "New Channel" and
"Channel Attributes" dialogs.
Add new setting GimpGuiConfig:devices-share-tool. When TRUE, we never
copy any properties between the user context and the GimpDeviceInfo's
context, so no tool or anything changes.
We do however still keep track of the active device so the setting can
be enabled/disabled at any time. Also hide GimpDeviceStatus' tool,
brush etc. indicators in "shared" mode.
Otherwise the quit dialog is still in front yet it has no focus and
one doesn't see this immediately.
Hitting ctrl-d to exit and ignore any subsequent files for instance
duplicates the current active image instead.
making its external API "complete". Remove the redundant
"new_base_type" and "new_precision" from the internal (vfunc) API (the
Babl format has the same information).
Rename "dialog" variables to "private" like in the other recently
cleaned up dialogs. Use less complicated code when connecting to the
container of dirty images.
Otherwise keyboard interaction (like hitting Escape or ctrl-d) ends up
broken and you may not realize immediately the dialog lost focus since
it is still the top window.
"There is one image" should not necessarily be translated to '1' in
other languages. It depends if their singular applies only to 1, like
in English. For instance in French, it can also apply to 0, so I fixed
the French translation.
...when moving guides
Show relative coordinates when moving guides and sample points, the
cursor display in the statusbar already shows absolute coordinates.
Add GimpCellRendererButton and use it to add a "Save" icon to each row
of dirty images. Click invokes the "edit-save" action, shift-click
invokes "edit-save-as". Also add a tooltip for the icon button.
Involves minor changes to GimpContainerTreeView to allow
GimpCellRendererButton to be added, and to allow external
"query-tooltip" handlers to run.
and use gimp_file_new_for_config_path() and _get_config_path() when
dealing with them. We used a weird mix of config paths and plain
(filesystem encoded) paths, waiting to to break on umlauts or
whatever. The code in gimpcolorconfig.c was particularly bad.
Add a SELECT_SOFTPROOF_PROFILE mode to the color profile dialog and
use it to select a profile from a newly added "Soft-Proofing Profile..."
menu item in view -> color management.
In a blocking operation we don't give the main loop time to lay out
the statusbar correctly after showing the progress bar. Force a size
on the progress bar using gtk_widget_size_allocate(). This sucks.
and call gimp_image_convert_indexed() in image-commands.c where it
belongs. Also remember the default values there, they will to to
GimpDialogConfig next.
and move the actual stroking code to select-commands.c,
vectors-commands.c and gimpvectorstool.c. Remember the active drawable
so the stroke always happens on the drawable the dialog was invoked
with.
and move the actual filling code to select-commands.c and
vectors-commands.c. Remember the active drawable so the fill always
happens on the drawable the dialog was invoked with.
Don't change the paste_type to NEW_LAYER just because there is a
complex thing in the clipboard. Instead, honor the request to paste
FLOATING or FLOATING_INTO and reduce the pasted thing to a simple flat
layer. Also remove the message and paste_type change from edit_paste()
in edit-commands.c, it had the same purpose, just with a user
notification.
In order to paste the full layer one needs to explicitly invoke
"Paste as new layer" now.
It still changes the paste_type to NEW_LAYER if it's impossible to
attach a floating selection to the target, the menu item was simply
insensitive before.
Instead, provide a custom GimpAddMaskCallback, connect to "response"
internally and call the callback. Takes clutter out of layers-commands.c.
Also attach the dialog to the layer so we don't show multiple add
mask dialogs. Should do the same to all dialogs with public structs,
the custom callback approach keep things more encapsulated.
the edge of a dock in multi-window-mode
when the pointer is grabbed do not process a
motion-notify-event having the wrong 'device' member.
This avoids a harmful device change.
They have a size of 1x1, treat them a image-sized when pasting layers
on top of them.
(my commit below was also about gimp_edit_get_paste_offset(), not
about gimp_edit_paste() as the log claims)
Use the newly added clipboard for entire images to copy/paste layers
(we only create single-layer clipboard images, and use only the first
layer of any recieved image, the layers can be arbitrarily complex
though):
- change gimp_edit_copy,cut,paste() to return/take a GimpObject
that can be a GimpImage or GimpBuffer
- cut/copy the whole layer if there is no selection
- always paste layers as new layers, not floating selections
- always paste news layers on top of the active layer, where
we would attach a floating selection
- add enum GimpPasteType { FLOATING, FLOATING_INTO, NEW_LAYER }
- add GimpPasteType parameter to gimp_edit_paste() and handle all
three cases there because there is now a lot of common code
involved
- change all callers accordingly, use only legacy buffer pasting
from the PDB for now
which can be used to load/save XCF from/to arbitrars streams. Use the
new functions from the load/save procedure invokers.
This commit also fixes a leaked open file ans input stream for each
loaded XCF.
...when threshold > 0
Add an "Antialias" toggle to the bucket fill options and set it on the
GimpFillOptions. In gimp_drawable_bucket_fill(), pass it to
gimp_pickable_contiguous_region_by_seed() instead of always defaulting
to TRUE.
The position of the toggle and its huge tooltip may need some
adjustment.
Add GimpFillOptions and GimpStrokeOptions to GimpDialogConfig and use
them in the Fill/Stroke Selection/Path dialogs and for the "with last
values" commands. Add GUI for them to Preferences -> Dialog Defaults.
This requires most of the stuff in my last few commits, and some
more changes:
GimpFillOptions is a GimpContext which has all sorts of connections to
everything, including a Gimp pointer. Hack around in GimpDialogConfig
to add a Gimp property, and add "gimp" parameters to quite some GimpRC
functions. Treat the Gimp* as a GObject* in all public API because
core/ stuff is not known in config/.
Move fonts, data factories, document list, paint methods and user
context creation to gimp_init() or gimp_constructed() so that most
members are created when gimp_new() is done. This does not load any
data earlier, it just makes sure that all containers exist when
gimp_load_config() is called. It's also cleaner and less fragile,
and initialize units in gimp_init(). This was completely
over-engineered but in the end boils down to a bad hack that needs a
static "the_unit_gimp" pointer anyway, so let's at least have the hacks
in one file.
Namely dialogs_attach_dialog(), _detach_dialog(), _get_dialog()
So we don't create the same dialogs multiple times, but raise the
already existing ones. Port some places that did this manually
to the new functions, and use it for many other dialogs.
...after entering single window mode
In GimpToolPalette, implement GtkWidget::hierarchy_changed() and
re-set the tooltips using the new toplevel's actions.
In gimp_widget_set_accel_help(), make sure we don't connect to the
same action twice, and make sure we disconnect everything if either
the widget or the accel_group go away.
It's deprecated in GTK+ 3.x, so we get rid of some deprecation
warnings. Also, it has always only been a utility function that hides
what is really going on in the widget hierarchy.
Try harder to show help when webkit is missing or the help browser
won't start:
- offer to switch to the web browser when the help browser is
present, but won't start for some reason
- in prefs, don't bother showing the help browser GUI when webkit
is missing, otherwise show a warning label if the help browser
is not installed even though webkit is there. Switch to the web
browser in both cases (modified patch from lisanet)
- add OS X replacement for gtk_show_uri() in plug-in-web-browser
(modified patch from lisanet)
(gimp-2.9:3): Gimp-Core-CRITICAL **: gimp_marshal_VOID__OBJECT_POINTER: assertion 'n_param_values == 3' failed
Conflict between signal marshal function
and signal emission.
AddressSanitizer output (paths manually shortened)
Direct leak of 96 byte(s) in 1 object(s) allocated from:
#0 0x7f543247d1d0 in realloc (/lib64/libasan.so.3+0xc71d0)
#1 0x7f542d3ad02f in g_realloc glib/glib/gmem.c:159
#2 0x57918d in gimp_color_history_set_property gimp/app/widgets/gimpcolorhistory.c:169
#3 0x7f542d681ec0 in object_set_property glib/gobject/gobject.c:1423
#4 0x7f542d681ec0 in g_object_new_internal glib/gobject/gobject.c:1817
#5 0x7f542d6839ed in g_object_new_valist glib/gobject/gobject.c:2042
#6 0x7f542d683c73 in g_object_new glib/gobject/gobject.c:1626
... standard icon names and GTK+ icon names as second choice.
We should only use GIMP specific icon names as last resort, when there
is no standard or GTK+ names dedicated to the function.
This is made possible thanks to commit 3cc77b0.
s/gimp-document-recent/document-open-recent/
s/gimp-indent/format-indent-more/
s/gimp-next/go-next/
s/gimp-previous/go-previous/
s/gimp-save/document-save/
s/gimp-save-as/document-save-as/
s/gimp-revert/document-revert/
s/gimp-open/document-open/
s/gimp-document-recent/document-open-recent/
s/gimp-quit/window-close/ ou s/gimp-quit/application-exit/
s/gimp-warning/dialog-warning/
s/gimp-edit-clear/edit-clear/
s/gimp-justify-.*/gtk-justify-.*/
s/gimp-font/gtk-select-font/
s/gimp-color-palette/gtk-select-color/
s/gimp-cancel/gtk-cancel/
Add gimp_prefs_box_set_page_scrollable() which does what it says, and
make the tallest pages of the prefs dialog scrollable. This method
allows the dialog's smaller pages to still enforce a minimum height
for the window.
Quite heavily modified by Mitch to address the more general issues
mentioned in comment 35 of the bug:
- Remember the added mask's type and the invert boolean in
GimpDialogConfig
- Add new prefs page "Dialog Defaults" which will contain
only stuff from GimpDialogConfig and can be reset separately
- Remove static mask variables from layers-commands.c and
use the new config values for both interactive mask adding
with the dialog, and for the shortcut based method
- Add a button to the layers dialog which supports add, add
with last values, delete, apply
- Add modifier-click shortcuts on the layer preview with the
same modifiers as on the button
... starting with toolbox buttons.
This is particularly a problem in single window mode (and sometimes in
multi-window mode) where canvas easily loses focus, hence key events.
Place the rulers' origin at the top-left corner of the canvas
(screen space) bounding box, and set their scale to the image-
space scale along the screen-space horizontal/vertical directions
(in other words, measuring a distance using the rulers should
give the same results as the measure tool; note that rotation
comes into play here only when the horizontal and vertical
image or screen resolutions are different, since otherwise the
scale is direction invariant.)
Make scrollbar step match ruler step under the new behavior.
...after entering single window mode
This commit fixes dangling pointers in GimpDockColumns, it doesn't fix
the bug but it's a prerequisite. Original patch by Massimo.
Separate clearing/creating the image's cached color transforms from
clearing/creating its color profile. Clear the transforms when the
color profile changes, and when image type or precision change. Create
the transforms only on demand, so clearing them multiple times doesn't
trigger any redundant (and expensive) transform creations.
... not screen space
Flip and rotate the canvas around the center of the viewport,
instead of the center of the image. Scroll the display as
necessary to keep the center of the viewport fixed during
these operations. This applies to both the corresponding
menu items, and rotation using Shift+Space.
Likewise, flip the canvas across the designated axis in
screen space, instead of image space. Rotate the display as
necessary, such that the reflection appears to happen in
screen space, regardless of the current rotation angle of the
canvas.
As proposed by Mitch. This completes first renaming attempt (commit
08ffc10) with even less ambiguous names.
"horizontal-axis-position" and "vertical-axis-position" are now
respectively "mirror-position-y" and "mirror-position-x".
"Horizontal position" and "Vertical position" could be mistaken as
meaning respectively the x and y coordinates, hence the vertical (resp.
horizontal) guide's positions.
The more accurate "horizontal-axis-position" and "vertical-axis-position"
namings are less misleading.
...selection on the canvas
if the editor dialog is active do not copy to the
clipboard the text selected, copy it instead when
selected with keys <Shift>-left/right.
Don't remove the GDK_KEY_space in GimpSearchPopup because that's
impossible and removes it from the entire GimpPopup class. Instead,
don't handle the "space -> confirm" binding manually if the focus
widget is a text widget.
Pass a GeglRectangle to all gimp_prop_gui_*() functions, and set
the soft limits to the rectangle's extents if the GParamSpec
metadata's "unit" is "pixel-coordinate" or "pixel-distance".
In GimpSymmetryEditor, pass the image extents, in GimpOperationTool
pass the drawable extents.
do not access progress object after destruction.
silence warnings like:
(gimp-2.8:1): GLib-GObject-WARNING **: gsignal.c:2635: instance '0x17083f0' has no handler with id '56146'
(gimp-2.8:1): Gimp-Plug-In-CRITICAL **: plug_in_icc_profile_apply_rgb: assertion 'progress == NULL || GIMP_IS_PROGRESS (progress)' failed
Mitch gets a better idea to deal with soft limits (i.e. min/max values
different from the property min/max) applied to a spin scale created by
gimp_prop_spin_scale_new().
So let's just remove the current implementation (using locale data on
the GimpConfig object). The symmetry spin scales are back with crazy
huge maximums, which makes quite a horrible GUI, but this is only
temporary until Mitch commits his new implementation.
... in gimp_drawable_bucket_fill()
gimp_drawable_apply_buffer() already takes the selection mask into
account. Intersecting the selection mask with the fill region
prior to that leads to wrong (too low) alpha values, in general,
when partial selection is involved.
This commit eliminates the intersection of the fill region and
selection mask data before the apply_buffer() call, although it
does calculate the intersection of their bounds, to avoid
processing regions that won't be visible anyway.
Add an icon button to the dynamics editor's "mapping matrix" page. It
looks just as ugly as the icon button of the tool preset editor, the
two should be visually improved together.
The code to reorder buttons was actually already there, yet the
alternative order was not taken into account unless the setting
"gtk-alternative-button-order" is TRUE.
Previously, activating quick mask while the selection was empty
would use the image's channel format for the mask, instead of its
mask format; these formats are different for sRGB images.
...XCF channel and layer properties
The properties PROP_ACTIVE_LAYER, PROP_FLOATING_SELECTION,
PROP_ACTIVE_CHANNEL saves the current object pointer the @info
structure. Others like PROP_SELECTION (for channel) and
PROP_GROUP_ITEM (for layer) will delete the current object and create
a new object, leaving the pointers in @info invalid (dangling).
Therefore, if a property from the first type will come before the
second, the result will be an UaF in the last lines of xcf_load_image
(when it actually using the pointers from @info).
I wasn't able to exploit this bug because that
g_object_instance->c_class gets cleared by the last g_object_unref and
GIMP_IS_{LAYER,CHANNEL} detects that and return FALSE.
Currently a GimpSpinScale created with gimp_prop_spin_scale_new() will
use the associated property's lower and upper values.
Unfortunately these generic values may not be always relevant and we
may want to construct a spin scale UI adapted to the current image.
For instance, several symmetry painting have a x/y property which has
to stay within the image's dimension, but changing the property's lower
and upper values would affect the symmetry on the class level (i.e. for
all similar symmetries on all images).
Let's allow setting data on object with key "property-name:min|max" to
provide locale min/max values specifically for this object.
This is used only on the symmetry dock for now, but could be used as
well on GEGL op UIs.
This operation produces NAN (0 / 0) and division
by zero when a layer component is zero.
Inline CLAMP to make sure it swallows also NAN.
A NAN is mapped to 1 for backward compatibility.
[the 2variate function comp (in[b], layer[b]) is
discontinuous at layer[b] = 0]
Whereas some IME would just cancel any preediting on IM context reset,
others would commit their preedit (for instance Ibus-Hangul).
Unfortunately it is not necessarily immediate, even after
gtk_im_context_reset() and gtk_im_context_focus_out(), so when switching
text layer, any preedit text is committed on the new layer.
Let's recreate the IM context on abort to make sure we don't share any
input between text layers.
The IM should be the one taking care of it by sending the proper
signals, and I would assume this should be done when we run:
gtk_im_context_focus_out ().
But it doesn't and leaves a dangling styled preedit text on leaving
a text layer, either because the IME is not properly implemented (not
reacting to the context leaving the focused widget), or GTK+2 is bugged,
or maybe we just don't understand exactly how this all works. Anyway
let's explicitly clean preedit text, not waiting for preedit-end or
commit signals on IM abort.
Honor them in gimp_widget_get_color_transform() and make them
configurable in prefs. The code and GUI avoid the negation in
the lcms "NOOPTIMIZE" flag.
With left gravity, the mark is moved back 1 character to the left of the
cursor, whereas with right gravity, it is right after newly-inserted
text. This resulted sometimes into non-deleted preedit characters.
I don't understand all the subtleties since the problem only occurred
for the first character of a text layer, at least in my tests.
Anyway now it should be good.
Japanese IME emits the preedit-commit before preedit-end (Hangul or
Unicode don't, but we could assume other IMEs may be doing the same).
As a consequence, the undo step was not pushed, thus breaking undo
history and consistency.
The solution is to simulate preediting end then restart when a commit
happens during preediting.
Since commit b593462, gimp_text_tool_apply() is a semi-public function
(only to be used from gimptexttool-editor.c). Therefore move it to the
"public functions" section.
Commit cd147a4 reintroduced the crash. The culprit was the idle function
gimp_text_tool_apply_idle() which may not have been processed when
gimp_text_tool_move_cursor() is called, resulting in inconsistencies
between the text in the text tool's pango layout and its buffer.
Force any pending text commits to be applied before moving the cursor.
to gimp_color_config_get_simulation_color_profile(), along with
various internal "printer" strings. Also reduce the number of
user-visible "print" stuff from the preferences color management page.
Use cmsFLAGS_NOOPTIMIZE only for actual image buffer or single color
transforms, but not for previews or the image display. Makes things a
lot more responsive again.
In the text tool editor code, connect to GtkIMContext::preedit-start
and introduce a boolean text_tool->preedit_active which indicates that
a preedit is going on.
Remove the new preedit-removal code from gimp_text_tool_reset_im_context()
because it was not reflecting the IM's internal state and made things
worse. Instead, added gimp_text_tool_abort_im_context() which really
gets rid of any ongoing preedit by force.
In the main text tool code, check for preedit_active and if TRUE,
apply any edits directly widhout pushing undo steps. Factored out
gimp_text_tool_apply_list() for that purpose in order not do
duplicate a lot of code.
On undo and on button_press, force-abort any ongoing preedit. This is
the right thing to do on undo, but not really on button_press, but I
don't see another way to keep states consistent.
The new gimp_text_tool_im_delete_preedit() also makes sure that
text_tool->preedit_string is non-NULL when modifying the buffer,
so it can be used as indication whether or not a preedit is
going on (in order to skip undo stuff while preediting).
Rather than just discovering them by chance, a simple grep and some
search and replace are much more efficient! :-)
Cleaning only done on C and automake files.
and update their color transforms with the new monitor's color
profile. A widget is considered changing monitors when its toplevel
window's center crosses, in order to let widgets within one window
have consistent colors.
In particular, I don't want the preedit foreground colors to be
displayed in the text style editor UI. They are only temporary hints
and should not be taken into account as user-chosen style.
When gtk_im_context_get_preedit_string(), we have to inspect the
returned Pango attributes, so that the preedit string can be displayed
in the expected fashion (i.e. as in other programs).
Some input methods in particular would even break the preedit strings
in several chunks of text displayed differently (for instance Japanese),
depending on the cursor position within the preedit string.
There is nothing in our UI to set background color to text, but this
can be used internally, for instance to display input method's preedit
texts with specific colors.
Now all previews and the display shell connect to the color config
itself, there is no need any longer to connect to the global color
config's "notify" any longer from GimpImage. Also, the settings there
are for display purposes only, so nothing in the image itself needs
to be notified of the config change.
Need to connect_after() to the shell's GimpColorConfig "notify" so the
profile transform cache in gimp_widget_get_color_transform() is
already invalidated when we call gimp_color_managed_profile_changed()
in the shell's notify callback.
This is a hack. Actually the shell should create its transform only on
demand, and a config change should simply get rid of the previous
transform, just as all other widgets are doing it.
... and image_resolution != screen_resolution.
Partially revert commit 26ea7a3530.
Also, reset shell->dot_for_dot in gimp_display_shell_fill(), so that new
images that use the first display don't inherit the dot-for-dot state of the
previous image that used the same display.
Which creates a buffer from GimpFillOptions that can be applied to a
drawable. Eliminates three slightly different copies of the same
code. Also adds to the color history for each color fill, we missed
two places before.
Now the max parameter from Windows is taken into account and takes
precedence over the saved session state if set.
This can be applied either through the "run" property of a shortcut,
or by command line: `start /max gimp-2.9.exe`.
There is a start for min handling as well but I can clearly see the
window being minimized and immediately re-raised. There must be a call
later which deiconifies the window. This will have to be fixed.
Automatically connect to the global GimpColorConfig unless a different
one is set manually (like for previews that belong to a
GimpDisplayShell and should follow its color management settings. Add
public API for subclasses to get to the renderer's color
transform. Automatically color manage rendered pixbufs, just as we
already manage temp_bufs.
Which takes two profiles and returns TRUE if converting between
them works correctly without a GimpColorTransform. Use it in
gimp_color_transform_new() to return a NULL transform if none
is needed. Took the code from gimp-gegl-loops.c.
which encapsulates a cmsHTRANSFORM and does all the pixel format
conversion magic. It has API to create transforms and proofing
transforms, and to convert pixels arrays and GeglBuffers.
Before, each place which has a transform had to keep around the
transform and its input and output Babl formats, and had to implement
lots of stuff itself. Now all that lives in GimpColorTransform,
removing lots of logic from many places, and pretty much removing lcms
from the public API entirely.
This removes including <lcms2.h>, LCMS_LIBS and LCMS_CFLAGS from
almost all directories and potentially allows to replace lcms by
something else.
which isn't really for "picking", but it just fits too nicely into
GimpPickable to not put it there.
Also add utility function gimp_pickable_srgb_to_image_color() which
takes a "real" (sRGB) GimpRGB value, transforms it to the pickable's
colorspace and puts it into an "image color" GimpRGB.
...when a color profile is active
Add GimpPickable::pixel_to_srgb() which puts a picked raw image
pixel into a GimpRGB. Default to gimp_rgba_set_pixel() but implement
pixel_to_srgb() in GimpLayer, GimpProjection and GimpImage and
run the pixel through gimp_image_color_profile_pixel_to_srgb().
Replace the 3-state "off", "display" and "softproof" radio items by
two toggles "enable" and "softproof". Also add separate controls for
display and softproof options.
Not only the logic was broken, a local variable was also shadowing the
"dest_profile" variable and preventing the broken logic to be applied
at all. Double fail.
They are unreliable because every type checking cast discards them,
they are useless anyway, visual clutter, added inconsistently, and
generally suck. Wanted to do this a long time ago, it was a bad idea
in the first place.
... layers with alpha channel
add an operation that selectively outputs aux
or source if out/inside the rectangle (x,y,width,height),
independently of the alpha channel and use it
in filters split preview.
mostly copied from gimp:mask-components
Instead, store only the action's name, and look it up upon searching.
Keeping a reference to the actions in the history extended the lifetime of
actions that sohuld have otherwise died, leading to all sorts of nastiness. In
particular, the Alt+[n] actions to switch between images would remain bound to
their accelerator after the image has been closed, holding a dangling display
pointer, which would lead to a segfault, or fail with a CRITICAL message, when
using the accelerator again.
... when the button is clicked
Previously, the old color retained its previous value, so when the color
selection dialog was canceled, the button's color would go back to a
potentially outdated value, instead of the color it had when it was
clicked (for continuously-updated color buttons, at least.)
Add rendering intent, black point compensation and gammut warning menu
items to View -> Color Management. They set the respective values of
the active color management mode, so both "color managed display" and
"print simulation" are almost completely configurable per-display
now. Setting the simulation profile is still missing.
Add a View -> Color Management submenu that allows to change
the color management mode per-display.
Internally, keep a GimpColorConfig object around per-display that
is synchronized with the global config except for the properties
that have a per-display GUI (currently the mode).
Also provide an "As in Preferences" menu item to follow the global
settings again.
Use the shell's color config for color managing the display and
various auxiliary widgets attached to it, like the notebook tab widget
and navigation popup.
The config is currently just a reference to the global prefs config,
so no behavior changed.
The values from the global GimpColorConfig are for displaying images
on screen, not default values for profile conversion. Instead, default
to intent = RELATIVE_COLORIMETRIC and bpc = TRUE.
First of all, derive GimpFilterOptions from GimpColorOptions, not the
other way around, which was a sick way of magically showing the right
options for various tools. That approach has failed and needed
additional hacks to do the right thing. Simply have the right class
hierarchy and call the right tool options GUI constructors and done.
Remove gimp_histogram_options_gui(), we don't need histogram scale
controls in tool options when we already have them next to the
histograms in the tool dialogs.
Also remove gimp_histogram_options_connect_view() and use a simple
g_object_bind_property() instead.
Remove the filter area calculation code from gimp_image_map_apply().
Instead, automatically update the area when adding the filter to the
drawable, and when the selection changes. Also, connect to the
drawable's "removed" and remove the filter when the drawable gets
removed from the image.
Reduces members and all sorts of duplication and is a much better
abstraction of what it does. Also make it a lot smarter and
self-updating, chop up the apply() function and move its parts where
they belong. Also, it's now aware of selection changes and does the
right thing.
Don't abort GimpImageMapTool on selection changes, it now nicely
handles that.
see commit below, they are not needed any longer. Also call
gimp_image_map_tool_preview() from gimp_image_map_tool_create_map()
and remove more other calls to preview().
- call it in gimp_image_map_tool_get_operation() so
gimp_operation_tool_set_operation() doesn't need to call it
- gimp_image_map_tool_settings_import() doesn't need to call it because
importing settings changes the config object anyway
Instead, emit "notify" in the config object, which is a hack,
but a temporary one. If should be the right code tho, if only
Gegl pads would be normal properties...
Keep GimpStatusbar progress messages at the front of the list.
Previously, gimp_statusbar_push() and friends would push progress
messages behind temporary messages, and would push normal messages,
added after a progress had started, in front of progress messages.
This resulted in the progress bar text not being updated if a
temporary message was active at the time the progress started, or
if a normal message was pushed after the progress started.
This commit fixes this by always keeping progress messages at the
front of the message list.
Add "gboolean color_managed" and "GFile *color_profile" to
GimpTemplate. Add a toggle and profile combo to GimpTemplateEditor.
Honor the new template properties in gimp_image_new_from_template().
Using a GFile property instead of a GIMP_TYPE_CONFIG_PATH is
preliminary, see the previous commit. I'd like to use GFile more
directly when dealing with config files, this is for testing that.
instead of the "Feather border" toggle, and pass the selected style
to gimp_channel_border().
The border style defaults to "Smooth", whereas the previous default
behavior corresponds to "Hard".
isntead of the feather parameter, and pass it to
gimp_gegl_apply_border().
Make the necessary changes to the rest of the code to maintain the
current behavior.
Mass parameter alignment changes to gimpchannel.h. Sigh #2...
instead of the feather parameter.
The BORDER_STYLE_HARD and BORDER_STYLE_FEATHERED styles are implemented
using the "gimp:border" operation, as was done previously. The
BORDER_STYLE_SMOOTH style is implemented by performing a "gimp:grow" and
a "gimp:shrink", and subtracting the shrunk image from the grown image
using "gegl:substract".
gimp_channel_border() is modified to pass either BORDER_STYLE_HARD or
BORDER_STYLE_FEATHER, depending on its feather parameter, to maintain
the current behavior. The next commit replaces it with a style
parameter as well.
Mass parameter alignment changes to gimp-gegl-apply-operation.h. Sigh...
Will be used as a parameter to control the behavior of
gimp_channel_border(), and the corresponding "Select -> Border..."
action, instead of the feather flag.
Can be one of:
- GIMP_CHANNEL_BORDER_STYLE_HARD: Current behavior, unfeathered.
- GIMP_CHANNEL_BORDER_STYLE_SMOOTH: Smooth border, better handling
partial selection, implemented as explained in the next commit.
- GIMP_CHANNEL_BORDER_STYLE_FEATHERED: Current behavior, feathered.
This doesn make any difference because a NULL profile would do the
same, but it's safer to pass the actual profile instead of relying on
the magic meaning of NULL in the call to
gimp_layer_new_from_gegl_buffer().
Don't use the (wrong) global display color managment switch to
determine whether or not to convert the buffer to the image's
profile. Use the image's "is color managed" switch.
We used to depend on the global color management OFF switch from
prefs, but that's meant for display color management. Now, don't do a
profile transform if the target image's color management is disabled.
Its API was too limited, and adding parameters for what we need next
would be equally random. It's a rarely used and mostly internal
widget, so simply use g_object_new().
This essentially reverts 1bf8eef14f
which tried to prefer file procedured that are GIO-enabled over
indirect loading via the file-uri plug-in. But file-uri is gone now so
we can treat all file procedured the same again, restoring the
original logic.
Now a file procedure which specifies only an extension is again
preferred over magic matching.
The bottom visible layer must be rendered in normal mode because
every other mode on top of nothing renders nothing.
Before, we would display a stack's last layer in normal mode, which
was a braindead attempt to make the layer stack look like in 2.8. Now
we set GimpFilter's "is-last-node" property the right way by looking
at the filters' visibility in GimpFilterStack.
When converting between RGB and GRAY and the image has a profile,
invoke the convert profile dialog with the newly added image type
conversion feature.
Get rid of individual constructors for assign and convert, instead
have one constructor that takes a dialog type enum. No new features
yet in this commit.
instead of just a boolean "convert_profile". This takes the logic to
figure the right target profile out of gimp_layer_convert_type(), it
can't possibly know everything about how to convert anyway, and having
the logic in the callers conveniently splits it up and distributes its
parts to the places they belong.
This commit should cause no behavor change and is just preparation for
fixing bug 765176.
Change file magic matching from using a simple boolean "magic matches"
logic to using a matching quality. The quality is measured by the
number of bytes that matched.
Matching a single file procedure's magics now tries all magics and
returns the best match quality.
Searching a file procedure for a given file now tries all file
procedures and returns the one with the best match quality.
This fixes raw camera files being opened as TIFF, given a better magic
than just the generic TIFF magic is provided.
g_free the lru 'unit' when removed from the cache
and keep track of the list length and last item
while looking for a duplicate already present in
the cache.
Don't skip the first 10 bytes. That code was there to skip the magic
"GIMP_XMP_1" of the old "gimp-metadata" parasite. Instead, properly
check for that magic in xcf_load_image() and pass only the actual XMP
to gimp_metadata_set_from_xmp(). Also remove the +10 hack in file-exr.
... to avoid long pause on start
On non-Linux operating systems the fontconfig cache is often not
initialized by default. The first time GIMP was launched, this led
to a non-responding application, confusing many users.
The initialization of fontconfig has now been moved to a separate
thread. The main thread will wait for this fontconfig thread to
complete, regularly pulsing the UI.
This patch was partly based on an earlier patch by Tor Lillqvist.
Startup functions can now make the progress bar pulse in order
to report that something of unknown duration is in progress.
Call status_callback (NULL, NULL, same_percentage)
We must not dither the alpha channel if it is fully opaque or fully
transparent. 2.8 did the same but the dithering of these values was
apparently consumed be putting the result into 8 bit values.
Disable "new-style" fullscreen mode on OS X, we need support in
GTK+ in order for this to work properly. The currently available
fullscreen option in View > Fullscreen works fine and we plan to
migrate to the "new-style" fullscreen support in the future.
... and improve input method UI. IM should not use some floating overlay
frame to display the preedit text. Instead it is now directly inserted
into the text tool box, as selected text. It permits redimensionning of
the text box while typing, and push the rest of the text when inserting
in the middle (instead of superimposing an overlay box over the current
text box, making everything unreadable).
Input methods still have a few more issues, but this fixes the main UI
weirdness as well as the crash with Hangul IM.
We needed to get rid of these images at a later point. This fixes (at
least) a crash seen on Mac OS X, where the images were being unreffed
before the last GimpActions (with a reference to the image) were
unreffed.
which connects the properties of a config GObject and a GeglNode, much
like g_object_bind_property() which we can't be used because GeglNode
properties can't be registered with the GObjectClass.
This fixes a problem where in some cases the item next to the edited
item was changed inadvertently. This was caused by the fact that
the change of the label within the tree_name_edited callback triggered
a re-ordering of the items, which triggered a stop editing signal, which
finally caused a re-entrance in the tree_name_edited callback. By this
time the item below "path_str" has changed, which caused the name of
another object to be edited.
Plus, implement GimpColorManaged so the view renderer can ask the
buffer for its profile.
This also fixes colors in cross-application copy & paste when GIMP
is the source.
Change gimp_viewable_get_[new]_preview() to return buffers of the
image's and layers' colorspace, but in u8. This way layer and image
previews can transform them correctly to the display profile.
Note: this makes plug-ins receive thumbnail buffers in that
pixel format too.
Also change gimp_viewable_get_[new]_pixbuf() to always return sRGB
buffers that can reasonably be put to screen directly, or put into DND
buffers. This is at least more correct now.
Keep around transforms not only to/from the image's pixels to
"R'G'B'A double" (GimpRGB's format), but also to/from "R'G'B'A u8"
(GdkPixbuf's format). Also add API to access all cached transforms
and the Babl formats expected when calling lcms.
...in "alt-pressed" mode
Menu activation doesn't cause a focus-out becaus menu keyboard
grabbing is implemented with a simple gtk_grab_add() (the menu popup
never gets the focus). Therefore, the canvas never gets a focus-out
event and the pressed modifiers are stuck.
Fixed by connecting to "grab-notify" on the canvas, and artificially
releasing all modifiers when the canvas is shadowed by a grab.
Change things so the guide is preferred over any color picking
interaction of the parent GimpColorTool. Needs a minor change in the
GimpCurvesTool subclass too.
...for the preview?
Change to allow for split previews in all 4 directions. Remove the
direction controls from the filter dialog, and instead implement
shift- and control-clicking on the split guide to switch
original/filtered sides, or orientation.
Rename profile constructors to say "d65_gray" instead of just "gray",
"srgb_trc" instead of "srgb_gamma", and drop the "srgb" from
"srgb_linear" because we now say "d65". This should be a naming scheme
that doesn't conflict with whatever future functions we might add.
Add more user friendly description to the PDB op.
Partially keep the technical description for the PDB op and add a
reference to the developer wiki.
Use the new "reference" key for the GEGL op.
After discussion in IRC some weeks ago give the Selection-Flood items
the more meaningful name Selection-Remove holes.
modified: app/actions/select-actions.c
modified: app/core/gimpselection.c
Add two new tools, GimpGuideTool and GimpSamplePointTool. They are
one-trick-ponies and can only create new or move existing guides and
sample points. They can't be selected from the toolbox, only
temporarily pushed as active tools on top of any active tool using
their public start() APIs.
Use that API to enable them when the rulers are clicked, and replace
the entire guide and sample point moving code in GimpMoveTool and
GimpColorTool by simple calls to that API.
This might look like overkill but can easily be used for other
features like moving guides from within the paint tools (mirror
painting) or gegl filters (preview curtain).
Gamma correction refers to correcting how an image is displayed on
various devices. Gamma adjustment is a more general phrase that also
applies to adjusting gamma as a step in image enhancement in a
color-managed workflow.
Pass the display scroll offset down to gimp_cairo_stipple_pattern_create()
and set it as offset on the created cairo pattern, so stipple patterns
look the same no matter how the display is scrolled.
Themes from 2.8 and before are not fully compatible with 2.10.
In particular, embedded icons would not work and you would end up with
the Symbolic icon theme (light colors), which may not work well with a
custom theme.
It is better to reset to the new defaults upon migration and users can
still try and configure the theming afterwards if they wish.
Add generic tool actions for spacing, hardness and force, and the
GimpToolControl infrasctructure to redirect them to tool-specific
actions. Add tool-specific actions for GimpBrushTool, GimpMybrushTool
and GimpWarpTool as redirect targets. Also fix some existing tool
action callbacks to use the right increase/decrease steps.
Since we have many themes now, this new name better indicates that it
is meant to follow your desktop theme settings.
Also it will likely not remain the default theme.
which are essentially a copy of the stroking GUI. We now can fill the
exact shape outlined by stroking selections and paths. Suggestions for
the menu item labels are welcome.
For selections, it's different from gimp_edit_fill() because it
ignores the selection while filling, just as stroking does. Currently
unused, stay tuned...
Fix GimpAirbrush to not keep a GimpSymmetry instance around across
paint strokes. The symmetry refs its drawable but lacks the code of
dropping the reference when the drawable is removed from the image, or
the image is closed.
Also cleaned up gimp_airbrush_paint() a bit.
When set, it crops the effect to that rectangle, using a cache for
quickly changing the previewed area. When disabling the preview rect,
make sure we don't lose the cached result by processing it into the
output cache through the mode and affect nodes which is cheap-ish.
and remove lots of labels from calls to gimp_prop_foo_new(). Also
had to manually remove some unwanted labels that are now added
automatically, fixes bug #761880.
I am still not sure whether custom guides should follow snapping rules.
Yet I could easily imagine you could want some normal guides with
snapping and in the same time symmetry without snapping to axis of
symmetry. So for the time being, let's disable snapping to custom guides
all the time and see if logic could be improved later.
This is not ideal since the scale widget is crazy huge, thus
impractical, but the solution used on properties of other symmetries
(updating min/max of the class property) is wrong since it applies to
the whole class.
For the time being, it avoid setting obviously bad values until we
figure out the ideal automatic UI construction.
Since commit 9c4dd96, a NULL label to gimp_prop_check_button_new()
generates a label from the property nick. Trying to add a GtkImage in
the button outputs a warning because a GtkBin can have only one child.
Fixes the warning:
Gtk-WARNING **: Attempting to add a widget with type GtkImage to a
GtkCheckButton, but as a GtkBin subclass a GtkCheckButton can only contain one
widget at a time; it already contains a widget of type GtkLabel
...behind main window on windows
Windows requires that the process with the
foreground window explicitely allows long
lasting background processes showing their
windows as foreground/focused
Remove GimpSymmetry::get_settings() and instead tag the properties that
should have a GUI with the GIMP_SYMMETRY_PARAM_GUI flag. Also use plain
g_object_class_install_property() because that allows for separate nick
and blurb. Finally, use gimp_prop_gui_new() to generate the GUI,
This reverts commit 6878b49491.
I didn't take into consideration that different themes could apply
various fonts, font sizes or label-scale for the widget. So the font
description needs to be reinitialized at every theme change.
Use a GType for the PROP_SYMMETRY property of GimpImage, and create
a default "identity" symmetry for an image.
I still use a GimpIntComboBox but store the property value in the
user-data column because gpointer isn't a subset of gint.
Adds in libgimpwidgets:
- gimp_int_combo_box_set_active_by_user_data()
- gimp_int_combo_box_get_active_user_data()
- gimp_int_store_lookup_by_user_data()
- gimp_prop_pointer_combo_box_new() to create a GimpIntComboBox and
attach it to a gpointer property.
Thanks Massimo and Mitch for reviewing my code.
You can now set any paint tool to mirror painting relatively
horizontal/vertical axis or a central point (any combination of these 3
symmetries).
This has been implemented as a new multi-stroke core, where every stroke
is actually handled as a multi-stroke (default of size 1).
This is also the first usage of custom guides for symmetry guiding.
Current version has to be activated in the playground.
With gimp_guide_custom_new(), you can create a custom guide with a different
style on canvas (other pattern/color/width). A custom guide won't be saved
and could be used, for instance, for specific GEGL op guiding.
Use a proper "progress" signal instead of a property "notify" one
to update the on-canvas progress widget.
This way the graph is not invalidated while processing it
which essentially means some cleanup and no more regressions:
- add state POINT_BOTH for moving the entire line
- use the move cursor for moving the entire line
- reorder some functions
- set the tool to handle click events
- make clicks and no-motion releases behave properly
- #if 0 the code for "shapeburst handles", they don't work yet
- return to gimp:shapeburst until we have figured progress for GEGL ops
- indentation, stuff...
While the tool should definitely use the new op, we want the code in
gimpdrawable-blend to use the old one for compat reasons. The new op
will be exposed via a separate pdb call in the future.
This reverts commit 8af3fec689.
This operation assigns to each pixel the minimum of the
maxima of all paths from it to the outside, as if the
input image represents a height map, and the operation
floods it with water.
Commit the old tool before even creating the new tool. Old and new
tool might be the same and share tool options, and we don't want
the new tool's initialization to mess up the old tool's state.
Fixes changing from one GEGL operation to another without explicitly
confirming the first operation. The bug only killed the cached filter
result, but that's bad enough.
The "Settings:" label should never be aligned with a label in the UI
below, because that suggests that are related. Also consistently
rename all GimpImageMapTool parameters and variables to "im_tool".
seems I didn't consistently group things in indented GimpFrame widgets
last time I cleaned up here. Should be better now. Also change the
"Auto" button to "Auto Input Levels" to make things clearer.
In gimp_display_shell_expose_region(), invalidate each rect rather
than rebuilding the cairo_region_t as GdkRegion and then invalidating
the entire GdkRegion. Improves performance for tools that invalidate
complex regions.
Use a GQueue instead of a GList in GimpCanvasGroup, so adding items
takes constant time instead of O(n). Also, removing items was
traversing the list twice. This should generally speed up tools with
lots of canvas items.
Also, add a private pointer to the instance struct instead of using a
GET_PRIVATE() macro.
Add new string members to GimpImageMapTool and use them instead of the
resp. fields of GimpToolInfo. Change ::get_operation() to return the
operation name and a lot of strings for the UI, and create both the
GeglNode and the config object in GimpOperationTool. Lots of various
cleanups in GimpImageMapTool subclasses. This is an intermediate state
on the way of making the whole filter applying mechanism more generic
and less depending on subclasses.
which registers config classes for operations that don't want their
config objects to be auto-generated. Register all color tools' config
objects. Change gimp_gegl_config_sync_node() to correctly handle ops
that have an object property of the right config object type.
Old themes should not be copied over to the GIMP 2.9/2.10 config at
first launch since they won't be compatible anyway. Let's go the less
painful way by dropping old themes rather than trying to salvage themes
which we know will be broken by design.
Add "title" and "help_id" parameters and members/properties to
GimpToolDialog, GimpToolGui, GimpOperationTool. Also make the order of
parameters more consistent.
When checked, diagonally neighboring pixels are considered connected
when calculating the affected area.
This commit also adds a corresponding diagonal_neighbors parameter to
gimp_drawable_bucket_fill(), and modifies the callers, other than the
bucket fill tool, to pass FALSE for this parameter, to retain the
current behavior.
...gimp_pickable_contiguous_region_by_seed(), in preperation
for adding a similar option to the relevant tools.
When this parameter is TRUE, all eight neighbors of each pixel
are considered when calculating the resulting region, instead
of just the four orthogonal ones.
This commit also modifies all callers to pass FALSE for this
parameter, to retain the current behavior.
There are still unused variables in gimp_mybrush_get_checksum() and
gimp_mybrush_get_memsize() but these seem to be because of pending
implementations. So I leave the warnings.
Fixed "gimpicons.c:30:45: fatal error: icons/Default/gimp-icon-pixbufs.c:
No such file or directory" after commits :
ee512b610dfa9e5b8f98
Fixed "ln: failed to create symbolic link ‘12/gimp-default-colors.png’:
File exists" by forcing new symlinking on creating get-default-colors.png
for Symbolic-inverted to prevent error during make.
Create GIMP_CONFIG_DEFAULT_(ICON_)?THEME in gimpguiconfig.h to set
defaults in a single place, except for libgimpwidgets/ which cannot
include from app/.
Call the suspend()/resume() from the previous commit around
gimp_display_shell_appearance_update() so the cancas is not
repositioned for each individual widget configuration.
Implement a mechanism to suspend/resume GimpImageWindow's "keep canvas
pos" logic which is used to keep the image in place across widget
changed such as show/hide rulers.
gimp_display_shell_fill(): call suspend()/resume() around
gimp_display_shell_appearance_update(), and center the image after
calculating the initial scale factor, so the image jumping at least
starts at the approximately right position.
because this should happen generically in the zoom model's "zoomed"
callback, and is done explicitly in a few special cases. The removed
calls should all be redundant.
...several different zooms tiled together
In gimp_display_shell_scale() don't shortcut things in the case where
we resize the window, but call gimp_display_shell_scale_resize() like
all other scaling functions do in the end.
Need to keep GimpPDBContext's stroke options (which is a context too)
in sync with the PDB context itself, so setting any of the context's
properties properly affects subsquent strokes, and also avoids running
into inconsistent states like happened in this bug.
gimp_plug_in_procedure_set_icon(): look at the passed icon_type,
not proc->icon_type. Introduced this when splitting the function
into set_icon() and take_icon() in d5255dc.
Change GimpFileDialog's file-procs properties to enum
GimpFileProcedureGroup and get them via the newly added API. Set
properties common to all dialogs in the parent class.
..."Operation not supported"
Add argument validation in the GimpPlugInProcedure impls of
GimpProcedure::execute() and ::execute_async().
When a file procedure is executed and it handles URIs not filenames,
run the resp. arguments through file_utils_filename_to_file() which
turns both URIs and filenames into a valid GFile which we can ask for
its URI.
Remove the code that links size, spacing etc. to brush defaults from
the GimpPaintOptions class impl and add it to gimppaintoptions-gui.c
as a callback instead.
It's not a core thing and that magic with side effects should only be
applied to the GimpPaintOptions behind the tool options GUI.
Call gimp_display_shell_scrolled() before gimp_overlay_box_scroll()
so the rotate transform is updated before the actual scrolling.
The explicit call to gimp_display_shell_rotate_update_transform() got
lost during the last commits, but this solution is cleaner and less
redundant anyway.
by changing it to always deal with the center of the viewport instead of
its upper-left corner.
Fix gimp_display_shell_scroll_center_image_coordinate() to use the
proper transform functions like in the last commit.
gimp_display_shell_scale_to(): calculate the point that should not
move with GimpDisplayShell's untransform/transform functions before
and after scaling, then scroll to the right point. Just using the
scale functions doesn't work any longer when a rotation is
active. Other functions are affected too, but this most important
issue can be fixed by fixing just this function.
Add PDB sample point API similar to how the guide API works. Add core
API similar to the core guide API to make guide and sample point APIs
as similar as possible.
Add GimpGeglProceure to keep track of recently used GEGL operations in
the filter history. The new procedure also takes care of running the
op in the GEGL tool, so filters-commands.c is almost empty now.
Change gimp-filter-history.c to find procedures by name instead of
comparing pointers.
The only thing missing now is rerunning a GEGL op with the last
settings (not just showing its UI).
file-procedure.[ch] is gone and its functions moved to
GimpPlugInManager where they belong (the manager keeps around the
lists of load, save and export procedures).
Utility functions from file-utils.[ch] that have nothing to do with
image files moved to core/gimp-utils.[ch].
This preparation commit only moves code around and renames it, the
history is still a list of plug-ins only:
- move app/core/gimp-filter-history.c
to app/plug-in/gimppluginmanager-history.c and clean it up
- move the actions that create the submenus under "Filters"
from the "plug-in" to the "filters" action group
- move the code that creates and updates the history actions
to the "filters" action group
- add menu setup code for the "filters" menu
- move the "history-changed" signal from GimpPlugInManager to Gimp
This adds a boolean to GimpCoords struct that is true for enabled
extended non-mouse devices and false for all the rest allowing
the mypaint brush to override the the pressure sent to the paint library.
New GimpColorHistory widget, replacing the code in GimpColorDialog, and
added to GimpColorEditor to have the color history accessible in the
color dock as well.
Thanks to Thomas Manni for the initial implementation attempt.
- GimpContext API and property
- a GimpDataFactory
- List and grid views with GimpDataFactoryView
- actions and a context menu
None of this is connected to the actual tool yet, or depends on
libmypaint in any way.
Metadata presence implies XCF version 6 or higher. Yet metadata are
stored in parasites which are perfecly supported in older GIMP versions
and are passed along through saving.
Therefore there is no need to keep XCF 6. We just add a warning telling
that metadata won't be visible in older GIMP when saving a file with
metadata and compatibility mode.
Allow to set profiles on grayscale images. Change profile validation
to check for image type and profile type. Actually the patch simply
makes some pieces of code less restrictive. Change user-visible
strings in the profile dialogs accordingly. Change PDB docs
accordingly.
Support creating linear/sRGB-gamma variants of gray profiles and
rename gimp_color_profile_new_linear_rgb_from_color_profile() to
gimp_color_profile_new_linear_gamma_from_color_profile() because it's
not RGB-specific any longer.
Remove the parameters from gimp_prefs_box_add_page() and instead
hardcode them once in GimpPrefsBox. This way they can also easily be
made configurable in the future.
Add gimp_stock_set_icon_theme() which can be called at any time, also
before gimp_stock_init(), in which case we avoid setting the configured
icon theme twice on startup. Call it from libgimp/gimpui.c and
from app/gui/icon-themes.c so the app and plug-ins use the same
icon theme.
