This commit completely removes the "Edit -> Fade..." feature,
because...
- The main reason is that "fade" requires us to keep two buffers,
instead of one, for each fadeable undo step, doubling (or worse,
since the extra buffer might have higher precision than the
drawable) the space consumed by these steps. This has notable
impact when editing large images. This overhead is incurred even
when not actually using "fade", and since it seems to be very
rarely used, this is too wasteful.
- "Fade" is broken in 2.10: when comitting a filter, we copy the
cached parts of the result into the apply buffer. However, the
result cache sits after the mode node, while the apply buffer
should contain the result of the filter *before* the mode node,
which can lead to wrong results in the general case.
- The same behavior can be trivially achieved "manually", by
duplicating the layer, editing the duplicate, and changing its
opacity/mode.
- If we really want this feature, now that most filters are GEGL
ops, it makes more sense to just add opacity/mode options to the
filter tool, instead of having this be a separate step.
Add a TRIVIAL layer-mode flag, and corresponding
gimp_layer_mode_is_trivial() function, which indicates if the blend
function of a given layer mode is trivial, i.e., either never
modifies the source pixels (for non-subtractive modes), or always
clears the destination pixels (for subtractive modes).
In the Luminance layer-mode, use the scratch allocator for
allocating temporary buffers, instead of using VLAs.
GimpOperationLayerMode already allocates data on the stack,
calculated as not to overflow the stack on any platform, so having
any of its descendants also allocate big buffers on the stack is
risky.
There was a global 3x3 array of babl fishes used for converting between
blending and compositing pixel representations, these were all hard-coded to
operate within the sRGB babl-space family. This commit updates a per-instance
array during operation prepare instead, that comes preconfigured with fishes
derived from the correct space. Since the same operation instance might get
different space input during its life time we store and compare the cached
fishes with the current format (which is unique depending on space).
This should address the problem seen in issue #2592
Use gimp_tile_handler_validate_validate(), added in the commit
before last, in gimp:buffer-source-validate, in order to pre-render
the necessary region of the buffer, instead of performing the
validation implicitly by iterating over the region. This is both
simpler, and, more importantly, allows us to render the entire
region in a single chunk, instead of tile-by-tile, which can be
considerably more efficient, especially with high thread counts.
This essentially extends the dynamic sizing of rendered projection
chunks to layer groups, which are rendered through
gimp:buffer-source-validate, rather than just the main image
projection.
Add a Babl space parameter to gimp_gegl_color_new() and all utility
functions using it. Pass NULL if the GimpRGB actually is sRGB, pass
the image's layer space if the color was already converted using
gimp_pickable_srgb_to_image_color() or similar.
Fixed by implementing Massimo's two findings:
gimp_operation_cage_transform_process(): if aux_buf is NULL, bail out
after initializing out_buf with identity vectors, fixes the crash.
gimp_cage_tool_create_filter(): set the drawable filter's region to
GIMP_FILTER_REGION_DRAWABLE, fixes offset when there is a selection.
All babl formats now have a space equivalent to a color profile,
determining the format's primaries and TRCs. This commit makes GIMP
aware of this.
libgimp:
- enum GimpPrecision: rename GAMMA values to NON_LINEAR and keep GAMMA
as deprecated aliases, add PERCEPTUAL values so we now have LINEAR,
NON_LINEAR and PERCPTUAL for each encoding, matching the babl
encoding variants RGB, R'G'B' and R~G~B~.
- gimp_color_transform_can_gegl_copy() now returns TRUE if both
profiles can return a babl space, increasing the amount of fast babl
color conversions significantly.
- TODO: no solution yet for getting libgimp drawable proxy buffers in
the right format with space.
plug-ins:
- follow the GimpPrecision change.
- TODO: everything else unchanged and partly broken or sub-optimal,
like setting a new image's color profile too late.
app:
- add enum GimpTRCType { LINEAR, NON_LINEAR, PERCEPTUAL } as
replacement for all "linear" booleans.
- change gimp-babl functions to take babl spaces and GimpTRCType
parameters and support all sorts of new perceptual ~ formats.
- a lot of places changed in the early days of goat invasion didn't
take advantage of gimp-babl utility functions and constructed
formats manually. They all needed revisiting and many now use much
simpler code calling gimp-babl API.
- change gimp_babl_format_get_color_profile() to really extract a
newly allocated color profile from the format, and add
gimp_babl_get_builtin_color_profile() which does the same as
gimp_babl_format_get_color_profile() did before. Visited all callers
to decide whether they are looking for the format's actual profile,
or for one of the builtin profiles, simplifying code that only needs
builtin profiles.
- drawables have a new get_space_api(), get_linear() is now get_trc().
- images now have a "layer space" and an API to get it,
gimp_image_get_layer_format() returns formats in that space.
- an image's layer space is created from the image's color profile,
change gimpimage-color-profile to deal with that correctly
- change many babl_format() calls to babl_format_with_space() and take
the space from passed formats or drawables
- add function gimp_layer_fix_format_space() which replaces the
layer's buffer with one that has the image's layer format, but
doesn't change pixel values
- use gimp_layer_fix_format_space() to make sure layers loaded from
XCF and created by plug-ins have the right space when added to the
image, because it's impossible to always assign the right space upon
layer creation
- "assign color profile" and "discard color profile" now require use
of gimp_layer_fix_format_space() too because the profile is now
embedded in all formats via the space. Add
gimp_image_assign_color_profile() which does all that and call it
instead of a simple gimp_image_set_color_profile(), also from the
PDB set-color-profile functions, which are essentially "assign" and
"discard" calls.
- generally, make sure a new image's color profile is set before
adding layers to it, gimp_image_set_color_profile() is more than
before considered know-what-you-are-doing API.
- take special precaution in all places that call
gimp_drawable_convert_type(), we now must pass a new_profile from
all callers that convert layers within the same image (such as
image_convert_type, image_convert_precision), because the layer's
new space can't be determined from the image's layer format during
the call.
- change all "linear" properties to "trc", in all config objects like
for levels and curves, in the histogram, in the widgets. This results
in some GUI that now has three choices instead of two.
TODO: we might want to reduce that back to two later.
- keep "linear" boolean properties around as compat if needed for file
pasring, but always convert the parsed parsed boolean to
GimpTRCType.
- TODO: the image's "enable color management" switch is currently
broken, will fix that in another commit.
Use gimp_input_data_stream_read_line_always(), instead of
g_input_data_stream_read_line(), in a bunch of places that don't
expect EOF. If we don't do that, the code assumes the GError
parameter is set by the function and returns an error indication,
causing the caller to segfault when it tries to access
error->message. Instead, we now process an empty line when EOF is
reached, which is caught by the normal parsing logic.
Additionally:
- Use gimp_ascii_strto[id]() when loading gradients, generated
brushes, and palettes, to improve error checking for invalid
numeric input.
- Improve gradient-segment endpoint consistency check.
- Allow loading palette files with 0 colors. They can be created
during the session, so we might as well successfully load them.
The default stack size for new threads on MacOS is 512 KiB, making
our 512 KiB limit for stack-allocated buffers in
gimp_operation_layer_mode_real_process() too high. Lower it to
256 KiB.
... instead of gegl_buffer_sample()
GEGL commit 26f13cbfe9aaaa8c176162e54fdbb8af6876538e got rid of the
per-buffer cached samplers, making gegl_buffer_sample() much more
expensive, suitable primarily for one-off samples.
Use a sampler object instead.
... which determines if a layer mode's blend function only affects
the alpha, maintaining the backdrop's color. This is currently
true only for ERASE, SPLIT, and ANTI_ERASE modes.
Add gimp_drawable_gradient_adjust_coords(), which adjusts the
gradient segment coords according to the gradient type, so that, in
cases where the gradient span is unrelated to the segment length,
the gradient cache (in GimpOperationGradient) is big enough not to
produce banding. Use the new function in gimp_drawable_gradient()
and in the gradient tool, instead of duplicating the logic.
Move the shapreburst coordinate-adjustment logic to the new
function, and add appropriate logic for conical gradients.
Remove the code that avoids using the gradient cache for conical
gradients from GimpOperationGradient.
Undo the part of commit fa9a4108c3
that moved cache generation from process() to prepare(). prepare()
is called after each property change, in order to calculate the
op's bounding box for invalidation. Since we only need the cache
for actual processing, generating it in process() avoids that
overhead.
Invalidate the gradient cache while setting relevant properties,
and validate it, reconstructing if necessary, during prepare(),
rather than process(), to avoid the need to use a mutex.
Make sure the cache has at least two elements, corresponding to
the initial and final colors of the gradient, since both colors
might be needed, and to avoid division by zero.
Avoid using a cache if its necessary size is too big, or if the
gradient type is conical, since the necessary cache size for
conical gradients is unrelated to the gradient line length.
Improve index rounding during cache lookup.
Lots of indentation fixes.
First WIP commit, adds:
- enum GimpGradientBlendColorSpace { RGB_PERCEPTUAL, RGB_LINEAR }
- linear blending mode for gradient segments
- tool options GUI for the blend and paint tools which use gradients
In gimp_operation_buffer_source_validate_process(), align the ROI
to the tile grid *before* intersecting it with the validate-
handler's dirty region. This is necessary since, even though
subsequent operations will only read data within the ROI, the
entire tiles containing the ROI will be fetched, resulting in an
area potentially greater than the ROI. We need to validate this
area in advance, or else it will be validated as part of the
subsequent operations, which can lead into the same deadlock we're
trying to prevent.
Fix a CRITICAL when calling gimp_layer_mode_get_format() with an
AUTO composite space and a NULL preferred format, which is valid:
it means the layer mode is composite-space agnostic (as DISSOLVE
is), and that there's no preferred format.
A NULL preferred format can occur during
gimp_operation_layer_mode_prepare() if the layer's mode node is not
yet attached anything through its "input" or "aux" pads, which is
the case during the call to gimp_layer_update_mode_node() while
constructing the layer's node in gimp_layer_get_node().
Our composite modes don't correspond directly to the Porter-Duff
operators after which they're named, and these names aren't too
descriptive anyway.
Rename the composite modes as follows:
Source Over => Union
Source Atop => Clip to Backdrop
Destination Atop => Clip to Layer
Source In => Intersection
Update relevant code, including UI text, enumerator names, function
names, and action names.
PDB function gimp_edit_blend() was based on "gimp:shapeburst" operation
whereas the rest of GIMP (in particular, the Blend tool) used
"gegl:distance-transform" which is much faster.
Setting the operation to "manhattan" metric ensures that it still
renders the same way as in 2.8 while being a lot faster.
There was still a problem regarding as how it renders differently from
the Blend tool, but it turns out that the Blend tool is the one
rendering differently from how it used to in 2.8. We should discuss
adding the "metric" property in the tool options.
This operation is currently broken on multi-thread. So disable
multi-threading, at least temporarily (if not forever since apparently
we can get similar output with "gegl:distance-transform", but much
faster and nicer). See bug 781621.
It seems gegl_buffer_sample() crashes with the first parameter being
NULL so let's just test its value first. This will output a huge
quantity of CRITICALs in this edge case but that's much better than a
crash. :-)
See also bug 793371 where this bug is being processed.
When checking the Cage Transform operation code, I saw some ifdef meant
to be switched to when bug 645810 is fixed. It also says in a comment:
> /* When Gegl bug #645810 will be solved,
> this should be a good optimisation */
Checking said bug, it appears it has been fixed since 2012!
I also fixed a bit the parameters in gegl_buffer_sample() call since it
seems the function signature has changed quite a bit.
which is just a #define to g_assert for now, but can now easily be
turned into something that does some nicer debugging using our new
stack trace infrastructure. This commit also reverts all constructed()
functions to use assert again.
The four remaining "classic" color tools (Brightness-Contrast, Curves,
Levels and Threshold) are in fact just special UIs for otherwise
completely normal filter ops.
Add normal filter actions for them and invoke them like all
other filters, which makes them show up in the filter history
automatically.
The only small hack needed is to special case them in
gimp_gegl_procedure_execute_async() so the right tools are created
instead of the default GimpOperationTool. Also, blacklist the
automatically generated tools actions from action search and the
shortcut editor.
... removed by commit 0f9da165e0, and
improved by this commit.
Our foo-light modes aren't really prepared to handle out-of-range
input. Make sure that in-range input doesn't result in out-of-
range output.
Add a safe_div() function to gimpoperationlayermode-blend.c, and
use it in the relevant blend funcs, instead of plain division.
This function clamps the quotient to some reasonable range, to
avoid infinities, and maps epsilon/... to 0, to avoid NaN. The
latter part results in similar qualitative results to the
corresponding legacy modes, when calculating 0/0.
...and present linear RGB Histograms
This is step one: implement the feature at all (without new defaults
or proper GUI, cough).
Add boolean "linear" properties to GimpOperationPointFilter,
GimpCurvesConfig and GimpLevelsConfig.
In the filter, simply set the input/output formats to linear in
prepare().
In the curves and levels tools, add "Linear" toggles from hell,
like in the histogram dockable, and make sure things work right
wrt changing and resetting the property, switching from levels
to curves, and picking colors.
The result currently changes when switching a non-nop curves/levels
between perceptual and linear, because adjusting the parameters
between the spaces is not implemented yet.
In legacy layer modes that may produce out-of-range output given
in-range input, clamp the result after blending and before
compositing, instead of after compositing, to avoid producing
different results than 2.8 in certain cases.
Add "clamp-input" (which clamps the input values to [0..1])
and "clamp-output" (which clips the final result to [0..1]),
properties, parameters and GUI to:
- GimpLevelsConfig
- GimpOperationLevels
- The levels tool dialog
- The gimp_drawable_levels() PDB API
The old deprecated gimp_levels() PDB API now sets both clamping
options to TRUE which restores the 2.8 behavior.
Also reorder some stuff in GimpLevelsConfig and elsewhere so the
levels parameters are always in the same order.
Override get_invalidated_by_change() with the same logic as
get_required_for_output(), to avoid unnecessary invalidation.
Avoid format conversion when input and aux have the same format.
Add pass-through fast path when the ROI is completely inside/
outside the cropped rectangle.
Use bulk memcpy(), instead of per-pixel test-and-copy, in
process().
This avoids unnecessarily processing regions of the input and/or
aux nodes that will get cropped out. In particular, this avoids
processing cropped-out regions when using the filter tool in split-
view mode.
Keep multi-threading disable on the cage transform operation. It is now
fixed but is a lot much slower than when single-threaded, making it
painful to use on bigger images.
So let's reenable later if we can improve this. See bug 787663.
Use gimp:buffer-source-validate, introduced in the previous commit,
for the source node of GimpDrawables. This avoids threading issues
with layer groups, or any other drawables that may use a validating
buffer, by making sure the buffer is validated before any
succeeding operations, and hence the associated graph is processed
on the same thread as the parent composition.
Restore multithreaded processing in GimpOperationLayerMode.
gimp:buffer-source-validate is a drop-in replacement for
gegl:buffer-source, however, if the attached buffer has a
validating tile-handler, it makes sure the required region is
validated during process(). This avoids a situation in which
validation happens in different worker threads at the same time
during the processing of a succeeding operation; since validation
is protected by the buffer's tile-storage mutex, this can result in
either a deadlock (currently), or an effective fallback to single-
threaded processing.
Update the dprod production of generated enum files to include
abbreviated value descriptions, as per the previous commits.
Add a comment for translators above the abbreviated descriptions,
specifying the full description they abbreviate.
Temporarily disable multithreading for GimpOperationLayerMode, to
avoid the deadlock. The environment variable
GIMP_MULTITHREADED_COMPOSITING can be set to reenable it, for the
sake of debugging.
It was accidentally made applicable to layers by commit
7d345071c7. Only the non-legacy
color-erase mode shoule be applicable to layers (since 2.8 didn't
allow it as a layer mode), while the legacy mode is only available
for painting, and in the fade dialog.
When rendering a gradient with a repeat mode of NONE, don't sample
the gradient at 0.0 and 1.0, for pixels that lie to the left and to
the right of the gradient, respectively. Instead, use the left
color of the leftmost segment directly, and, likewise, the right
color of the rightmost segment. This always gives us the right
color for such pixels, even when there are gradient stops, that may
use different colors, at 0.0 and 1.0,
Remember the gradient segment at which the most-recent sample lies,
and pass it to gimp_gradient_get_color_at() as a seed for segement
lookup on the next sample. This improves the performance
marginally.
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.
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.
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.
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.
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);
Ell