Gimp/app/operations/layer-modes/gimpoperationlayermode-blend.c

1183 lines
29 KiB
C
Raw Normal View History

app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
/* GIMP - The GNU Image Manipulation Program
* Copyright (C) 1995 Spencer Kimball and Peter Mattis
*
* gimpoperationlayermode-blend.c
* Copyright (C) 2017 Michael Natterer <mitch@gimp.org>
* 2017 Øyvind Kolås <pippin@gimp.org>
* 2017 Ell
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <https://www.gnu.org/licenses/>.
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
*/
#include "config.h"
#include <gegl-plugin.h>
#include <cairo.h>
#include <gdk-pixbuf/gdk-pixbuf.h>
#include "libgimpcolor/gimpcolor.h"
#include "libgimpbase/gimpbase.h"
#include "libgimpmath/gimpmath.h"
#include "../operations-types.h"
#include "gimpoperationlayermode-blend.h"
#define EPSILON 1e-6f
#define SAFE_DIV_MIN EPSILON
#define SAFE_DIV_MAX (1.0f / SAFE_DIV_MIN)
/* local function prototypes */
static inline gfloat safe_div (gfloat a,
gfloat b);
/* private functions */
/* returns a / b, clamped to [-SAFE_DIV_MAX, SAFE_DIV_MAX].
* if -SAFE_DIV_MIN <= a <= SAFE_DIV_MIN, returns 0.
*/
static inline gfloat
safe_div (gfloat a,
gfloat b)
{
gfloat result = 0.0f;
if (fabsf (a) > SAFE_DIV_MIN)
{
result = a / b;
result = CLAMP (result, -SAFE_DIV_MAX, SAFE_DIV_MAX);
}
return result;
}
/* public functions */
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
/* non-subtractive blending functions. these functions must set comp[ALPHA]
* to the same value as layer[ALPHA]. when in[ALPHA] or layer[ALPHA] are
* zero, the value of comp[RED..BLUE] is unconstrained (in particular, it may
* be NaN).
*/
void /* aka linear_dodge */
gimp_operation_layer_mode_blend_addition (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] + layer[c];
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_burn (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = 1.0f - safe_div (1.0f - in[c], layer[c]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_darken_only (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = MIN (in[c], layer[c]);
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_difference (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = fabsf (in[c] - layer[c]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_divide (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = safe_div (in[c], layer[c]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_dodge (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = safe_div (in[c], 1.0f - layer[c]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_exclusion (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = 0.5f - 2.0f * (in[c] - 0.5f) * (layer[c] - 0.5f);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_grain_extract (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] - layer[c] + 0.5f;
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_grain_merge (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] + layer[c] - 0.5f;
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_hard_mix (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] + layer[c] < 1.0f ? 0.0f : 1.0f;
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_hardlight (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
{
gfloat val;
if (layer[c] > 0.5f)
{
val = (1.0f - in[c]) * (1.0f - (layer[c] - 0.5f) * 2.0f);
val = MIN (1.0f - val, 1.0f);
}
else
{
val = in[c] * (layer[c] * 2.0f);
val = MIN (val, 1.0f);
}
comp[c] = val;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_hsl_color (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat dest_min, dest_max, dest_l;
gfloat src_min, src_max, src_l;
dest_min = MIN (in[0], in[1]);
dest_min = MIN (dest_min, in[2]);
dest_max = MAX (in[0], in[1]);
dest_max = MAX (dest_max, in[2]);
dest_l = (dest_min + dest_max) / 2.0f;
src_min = MIN (layer[0], layer[1]);
src_min = MIN (src_min, layer[2]);
src_max = MAX (layer[0], layer[1]);
src_max = MAX (src_max, layer[2]);
src_l = (src_min + src_max) / 2.0f;
if (fabs (src_l) > EPSILON && fabs (1.0 - src_l) > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gboolean dest_high;
gboolean src_high;
gfloat ratio;
gfloat offset;
gint c;
dest_high = dest_l > 0.5f;
src_high = src_l > 0.5f;
dest_l = MIN (dest_l, 1.0f - dest_l);
src_l = MIN (src_l, 1.0f - src_l);
ratio = dest_l / src_l;
offset = 0.0f;
if (dest_high) offset += 1.0f - 2.0f * dest_l;
if (src_high) offset += 2.0f * dest_l - ratio;
for (c = 0; c < 3; c++)
comp[c] = layer[c] * ratio + offset;
}
else
{
comp[RED] = dest_l;
comp[GREEN] = dest_l;
comp[BLUE] = dest_l;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_hsv_hue (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat src_min, src_max, src_delta;
gfloat dest_min, dest_max, dest_delta, dest_s;
src_min = MIN (layer[0], layer[1]);
src_min = MIN (src_min, layer[2]);
src_max = MAX (layer[0], layer[1]);
src_max = MAX (src_max, layer[2]);
src_delta = src_max - src_min;
if (src_delta > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat ratio;
gfloat offset;
gint c;
dest_min = MIN (in[0], in[1]);
dest_min = MIN (dest_min, in[2]);
dest_max = MAX (in[0], in[1]);
dest_max = MAX (dest_max, in[2]);
dest_delta = dest_max - dest_min;
dest_s = dest_max ? dest_delta / dest_max : 0.0f;
ratio = dest_s * dest_max / src_delta;
offset = dest_max - src_max * ratio;
for (c = 0; c < 3; c++)
comp[c] = layer[c] * ratio + offset;
}
else
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c];
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_hsv_saturation (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat src_min, src_max, src_delta, src_s;
gfloat dest_min, dest_max, dest_delta;
dest_min = MIN (in[0], in[1]);
dest_min = MIN (dest_min, in[2]);
dest_max = MAX (in[0], in[1]);
dest_max = MAX (dest_max, in[2]);
dest_delta = dest_max - dest_min;
if (dest_delta > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat ratio;
gfloat offset;
gint c;
src_min = MIN (layer[0], layer[1]);
src_min = MIN (src_min, layer[2]);
src_max = MAX (layer[0], layer[1]);
src_max = MAX (src_max, layer[2]);
src_delta = src_max - src_min;
src_s = src_max ? src_delta / src_max : 0.0f;
ratio = src_s * dest_max / dest_delta;
offset = (1.0f - ratio) * dest_max;
for (c = 0; c < 3; c++)
comp[c] = in[c] * ratio + offset;
}
else
{
comp[RED] = dest_max;
comp[GREEN] = dest_max;
comp[BLUE] = dest_max;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_hsv_value (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat dest_v;
gfloat src_v;
dest_v = MAX (in[0], in[1]);
dest_v = MAX (dest_v, in[2]);
src_v = MAX (layer[0], layer[1]);
src_v = MAX (src_v, layer[2]);
if (fabs (dest_v) > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat ratio = src_v / dest_v;
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] * ratio;
}
else
{
comp[RED] = src_v;
comp[GREEN] = src_v;
comp[BLUE] = src_v;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_lch_chroma (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat A1 = in[1];
gfloat B1 = in[2];
gfloat c1 = hypotf (A1, B1);
if (c1 > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat A2 = layer[1];
gfloat B2 = layer[2];
gfloat c2 = hypotf (A2, B2);
gfloat A = c2 * A1 / c1;
gfloat B = c2 * B1 / c1;
comp[0] = in[0];
comp[1] = A;
comp[2] = B;
}
else
{
comp[0] = in[0];
comp[1] = in[1];
comp[2] = in[2];
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_lch_color (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
comp[0] = in[0];
comp[1] = layer[1];
comp[2] = layer[2];
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_lch_hue (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat A2 = layer[1];
gfloat B2 = layer[2];
gfloat c2 = hypotf (A2, B2);
if (c2 > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat A1 = in[1];
gfloat B1 = in[2];
gfloat c1 = hypotf (A1, B1);
gfloat A = c1 * A2 / c2;
gfloat B = c1 * B2 / c2;
comp[0] = in[0];
comp[1] = A;
comp[2] = B;
}
else
{
comp[0] = in[0];
comp[1] = in[1];
comp[2] = in[2];
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_lch_lightness (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
comp[0] = layer[0];
comp[1] = in[1];
comp[2] = in[2];
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_lighten_only (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = MAX (in[c], layer[c]);
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_linear_burn (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] + layer[c] - 1.0f;
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
/* added according to:
http://www.deepskycolors.com/archivo/2010/04/21/formulas-for-Photoshop-blending-modes.html */
void
gimp_operation_layer_mode_blend_linear_light (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
{
gfloat val;
if (layer[c] <= 0.5f)
val = in[c] + 2.0f * layer[c] - 1.0f;
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
else
val = in[c] + 2.0f * (layer[c] - 0.5f);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
comp[c] = val;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_luma_darken_only (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat dest_luminance;
gfloat src_luminance;
gint c;
dest_luminance = GIMP_RGB_LUMINANCE (in[0], in[1], in[2]);
src_luminance = GIMP_RGB_LUMINANCE (layer[0], layer[1], layer[2]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
if (dest_luminance <= src_luminance)
{
for (c = 0; c < 3; c++)
comp[c] = in[c];
}
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
else
{
for (c = 0; c < 3; c++)
comp[c] = layer[c];
}
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_luma_lighten_only (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gfloat dest_luminance;
gfloat src_luminance;
gint c;
dest_luminance = GIMP_RGB_LUMINANCE (in[0], in[1], in[2]);
src_luminance = GIMP_RGB_LUMINANCE (layer[0], layer[1], layer[2]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
if (dest_luminance >= src_luminance)
{
for (c = 0; c < 3; c++)
comp[c] = in[c];
}
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
else
{
for (c = 0; c < 3; c++)
comp[c] = layer[c];
}
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_luminance (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
static const Babl *fish;
gfloat *scratch;
gfloat *in_Y;
gfloat *layer_Y;
if (! fish)
fish = babl_fish ("RGBA float", "Y float");
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
scratch = gegl_scratch_new (gfloat, 2 * samples);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
in_Y = scratch;
layer_Y = scratch + samples;
babl_process (fish, in, in_Y, samples);
babl_process (fish, layer, layer_Y, samples);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
while (samples--)
{
if (layer[ALPHA] != 0.0f && in[ALPHA] != 0.0f)
{
gfloat ratio = safe_div (layer_Y[0], in_Y[0]);
gint c;
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
for (c = 0; c < 3; c ++)
comp[c] = in[c] * ratio;
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
in += 4;
layer += 4;
in_Y ++;
layer_Y ++;
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
gegl_scratch_free (scratch);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
}
void
gimp_operation_layer_mode_blend_multiply (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] * layer[c];
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_overlay (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
{
gfloat val;
if (in[c] < 0.5f)
val = 2.0f * in[c] * layer[c];
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
else
val = 1.0f - 2.0f * (1.0f - layer[c]) * (1.0f - in[c]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
comp[c] = val;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
/* added according to:
http://www.deepskycolors.com/archivo/2010/04/21/formulas-for-Photoshop-blending-modes.html */
void
gimp_operation_layer_mode_blend_pin_light (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
{
gfloat val;
if (layer[c] > 0.5f)
val = MAX(in[c], 2.0f * (layer[c] - 0.5f));
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
else
val = MIN(in[c], 2.0f * layer[c]);
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
comp[c] = val;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_screen (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = 1.0f - (1.0f - in[c]) * (1.0f - layer[c]);
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_softlight (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
{
gfloat multiply = in[c] * layer[c];
gfloat screen = 1.0f - (1.0f - in[c]) * (1.0f - layer[c]);
gfloat val = (1.0f - in[c]) * multiply + in[c] * screen;
comp[c] = val;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
void
gimp_operation_layer_mode_blend_subtract (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
comp[c] = in[c] - layer[c];
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
/* added according to:
http://www.simplefilter.de/en/basics/mixmods.html */
void
gimp_operation_layer_mode_blend_vivid_light (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
gint c;
for (c = 0; c < 3; c++)
{
gfloat val;
if (layer[c] <= 0.5f)
{
val = 1.0f - safe_div (1.0f - in[c], 2.0f * layer[c]);
val = MAX (val, 0.0f);
}
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
else
{
val = safe_div (in[c], 2.0f * (1.0f - layer[c]));
val = MIN (val, 1.0f);
}
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
comp[c] = val;
}
}
comp[ALPHA] = layer[ALPHA];
comp += 4;
layer += 4;
in += 4;
}
}
/* subtractive blending functions. these functions must set comp[ALPHA] to
* the modified alpha of the overlapping content, as a fraction of the
* original overlapping content (i.e., an alpha of 1.0 specifies that no
* content is subtracted.) when in[ALPHA] or layer[ALPHA] are zero, the value
* of comp[RED..BLUE] is unconstrained (in particular, it may be NaN).
*/
void
gimp_operation_layer_mode_blend_color_erase (const gfloat *in,
const gfloat *layer,
gfloat *comp,
gint samples)
{
while (samples--)
{
if (in[ALPHA] != 0.0f && layer[ALPHA] != 0.0f)
{
const gfloat *color = in;
const gfloat *bgcolor = layer;
gfloat alpha;
gint c;
alpha = 0.0f;
for (c = 0; c < 3; c++)
{
gfloat col = CLAMP (color[c], 0.0f, 1.0f);
gfloat bgcol = CLAMP (bgcolor[c], 0.0f, 1.0f);
if (fabs (col - bgcol) > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat a;
if (col > bgcol)
a = (col - bgcol) / (1.0f - bgcol);
else
a = (bgcol - col) / bgcol;
alpha = MAX (alpha, a);
}
}
if (alpha > EPSILON)
app: layer mode code shuffling Commit 3635cf04ab69bafb76d7583da804f580f2d5fbf3 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.
2017-08-17 07:26:49 -07:00
{
gfloat alpha_inv = 1.0f / alpha;
for (c = 0; c < 3; c++)
comp[c] = (color[c] - bgcolor[c]) * alpha_inv + bgcolor[c];
}
else
{
comp[RED] = comp[GREEN] = comp[BLUE] = 0.0f;
}
comp[ALPHA] = alpha;
}
else
comp[ALPHA] = 0.0f;
comp += 4;
layer += 4;
in += 4;
}
}