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.
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.
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.
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);
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).
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.
... 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.
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.
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.
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.
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.
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().
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.
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.
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().
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.
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.
...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.