mutter/clutter/clutter/clutter-offscreen-effect.c

738 lines
25 KiB
C
Raw Normal View History

effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
/*
* Clutter.
*
* An OpenGL based 'interactive canvas' library.
*
* Copyright (C) 2010 Intel Corporation.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library 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
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see <http://www.gnu.org/licenses/>.
*
* Authors:
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
* Emmanuele Bassi <ebassi@linux.intel.com>
* Robert Bragg <robert@linux.intel.com>
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
*/
/**
* SECTION:clutter-offscreen-effect
* @short_description: Base class for effects using offscreen buffers
* @see_also: #ClutterBlurEffect, #ClutterEffect
*
* #ClutterOffscreenEffect is an abstract class that can be used by
* #ClutterEffect sub-classes requiring access to an offscreen buffer.
*
* Some effects, like the fragment shader based effects, can only use GL
* textures, and in order to apply those effects to any kind of actor they
* require that all drawing operations are applied to an offscreen framebuffer
* that gets redirected to a texture.
*
* #ClutterOffscreenEffect provides all the heavy-lifting for creating the
* offscreen framebuffer, the redirection and the final paint of the texture on
* the desired stage.
*
* #ClutterOffscreenEffect is available since Clutter 1.4
*
* ## Implementing a ClutterOffscreenEffect
*
* Creating a sub-class of #ClutterOffscreenEffect requires, in case
* of overriding the #ClutterEffect virtual functions, to chain up to the
* #ClutterOffscreenEffect's implementation.
*
* On top of the #ClutterEffect's virtual functions,
* #ClutterOffscreenEffect also provides a #ClutterOffscreenEffectClass.paint_target()
* function, which encapsulates the effective painting of the texture that
* contains the result of the offscreen redirection.
*
* The size of the target material is defined to be as big as the
* transformed size of the #ClutterActor using the offscreen effect.
* Sub-classes of #ClutterOffscreenEffect can change the texture creation
* code to provide bigger textures by overriding the
* #ClutterOffscreenEffectClass.create_texture() virtual function; no chain up
* to the #ClutterOffscreenEffect implementation is required in this
* case.
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
*/
#include "clutter-build-config.h"
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
#include "clutter-offscreen-effect.h"
#include <math.h>
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
#include "cogl/cogl.h"
#include "clutter-actor-private.h"
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
#include "clutter-debug.h"
#include "clutter-private.h"
#include "clutter-stage-private.h"
#include "clutter-paint-context-private.h"
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
#include "clutter-paint-volume-private.h"
#include "clutter-actor-box-private.h"
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
struct _ClutterOffscreenEffectPrivate
{
CoglHandle offscreen;
CoglPipeline *pipeline;
CoglHandle texture;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
ClutterActor *actor;
ClutterActor *stage;
graphene_point3d_t position;
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
int fbo_offset_x;
int fbo_offset_y;
/* This is the calculated size of the fbo before being passed
through create_texture(). This needs to be tracked separately so
that we can detect when a different size is calculated and
regenerate the fbo */
int target_width;
int target_height;
gint old_opacity_override;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
};
G_DEFINE_ABSTRACT_TYPE_WITH_PRIVATE (ClutterOffscreenEffect,
clutter_offscreen_effect,
CLUTTER_TYPE_EFFECT)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
static void
clutter_offscreen_effect_set_actor (ClutterActorMeta *meta,
ClutterActor *actor)
{
ClutterOffscreenEffect *self = CLUTTER_OFFSCREEN_EFFECT (meta);
ClutterOffscreenEffectPrivate *priv = self->priv;
ClutterActorMetaClass *meta_class;
meta_class = CLUTTER_ACTOR_META_CLASS (clutter_offscreen_effect_parent_class);
meta_class->set_actor (meta, actor);
/* clear out the previous state */
g_clear_pointer (&priv->offscreen, cogl_object_unref);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
/* we keep a back pointer here, to avoid going through the ActorMeta */
priv->actor = clutter_actor_meta_get_actor (meta);
}
static CoglHandle
clutter_offscreen_effect_real_create_texture (ClutterOffscreenEffect *effect,
gfloat width,
gfloat height)
{
return cogl_texture_new_with_size (MAX (width, 1), MAX (height, 1),
COGL_TEXTURE_NO_SLICING,
COGL_PIXEL_FORMAT_RGBA_8888_PRE);
}
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
static void
ensure_pipeline_filter_for_scale (ClutterOffscreenEffect *self,
float resource_scale)
{
CoglPipelineFilter filter;
if (!self->priv->pipeline)
return;
/* If no fractional scaling is set, we're always going to render the texture
at a 1:1 texel:pixel ratio so, in such case we can use 'nearest' filtering
to decrease the effects of rounding errors in the geometry calculation;
if instead we we're using a global fractional scaling we need to make sure
that we're using the default linear effect, not to create artifacts when
scaling down the texture */
if (fmodf (resource_scale, 1.0f) == 0)
filter = COGL_PIPELINE_FILTER_NEAREST;
else
filter = COGL_PIPELINE_FILTER_LINEAR;
cogl_pipeline_set_layer_filters (self->priv->pipeline, 0 /* layer_index */,
filter, filter);
}
static gboolean
update_fbo (ClutterEffect *effect,
int target_width,
int target_height,
float resource_scale)
{
ClutterOffscreenEffect *self = CLUTTER_OFFSCREEN_EFFECT (effect);
ClutterOffscreenEffectPrivate *priv = self->priv;
priv->stage = clutter_actor_get_stage (priv->actor);
if (priv->stage == NULL)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
{
CLUTTER_NOTE (MISC, "The actor '%s' is not part of a stage",
clutter_actor_get_name (priv->actor) == NULL
? G_OBJECT_TYPE_NAME (priv->actor)
: clutter_actor_get_name (priv->actor));
return FALSE;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
if (priv->target_width == target_width &&
priv->target_height == target_height &&
priv->offscreen != NULL)
{
ensure_pipeline_filter_for_scale (self, resource_scale);
return TRUE;
}
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
if (priv->pipeline == NULL)
{
CoglContext *ctx =
clutter_backend_get_cogl_context (clutter_get_default_backend ());
priv->pipeline = cogl_pipeline_new (ctx);
ensure_pipeline_filter_for_scale (self, resource_scale);
}
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
g_clear_pointer (&priv->texture, cogl_object_unref);
g_clear_pointer (&priv->offscreen, cogl_object_unref);
priv->texture =
clutter_offscreen_effect_create_texture (self, target_width, target_height);
if (priv->texture == NULL)
return FALSE;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
cogl_pipeline_set_layer_texture (priv->pipeline, 0, priv->texture);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
priv->target_width = target_width;
priv->target_height = target_height;
priv->offscreen = cogl_offscreen_new_to_texture (priv->texture);
if (priv->offscreen == NULL)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
{
g_warning ("%s: Unable to create an Offscreen buffer", G_STRLOC);
cogl_object_unref (priv->pipeline);
priv->pipeline = NULL;
priv->target_width = 0;
priv->target_height = 0;
return FALSE;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
return TRUE;
}
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
static gboolean
clutter_offscreen_effect_pre_paint (ClutterEffect *effect,
ClutterPaintContext *paint_context)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
{
ClutterOffscreenEffect *self = CLUTTER_OFFSCREEN_EFFECT (effect);
ClutterOffscreenEffectPrivate *priv = self->priv;
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
ClutterActorBox raw_box, box;
ClutterActor *stage;
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
CoglMatrix projection, old_modelview, modelview;
const ClutterPaintVolume *volume;
CoglColor transparent;
gfloat stage_width, stage_height;
gfloat target_width = -1, target_height = -1;
CoglFramebuffer *framebuffer;
gfloat resource_scale;
gfloat ceiled_resource_scale;
graphene_point3d_t local_offset;
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
gfloat old_viewport[4];
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
local_offset = GRAPHENE_POINT3D_INIT (0.0f, 0.0f, 0.0f);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
if (!clutter_actor_meta_get_enabled (CLUTTER_ACTOR_META (effect)))
return FALSE;
if (priv->actor == NULL)
return FALSE;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
stage = _clutter_actor_get_stage_internal (priv->actor);
clutter_actor_get_size (stage, &stage_width, &stage_height);
if (_clutter_actor_get_real_resource_scale (priv->actor, &resource_scale))
{
ceiled_resource_scale = ceilf (resource_scale);
stage_width *= ceiled_resource_scale;
stage_height *= ceiled_resource_scale;
}
else
{
/* We are sure we have a resource scale set to a good value at paint */
g_assert_not_reached ();
}
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* Get the minimal bounding box for what we want to paint, relative to the
* parent of priv->actor. Note that we may actually be painting a clone of
* priv->actor so we need to be careful to avoid querying the transformation
* of priv->actor (like clutter_actor_get_paint_box would). Just stay in
* local coordinates for now...
*/
volume = clutter_actor_get_paint_volume (priv->actor);
if (volume)
{
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
ClutterPaintVolume mutable_volume;
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
_clutter_paint_volume_copy_static (volume, &mutable_volume);
_clutter_paint_volume_get_bounding_box (&mutable_volume, &raw_box);
clutter_paint_volume_free (&mutable_volume);
}
paint volumes: another pass at the design This is a fairly extensive second pass at exposing paint volumes for actors. The API has changed to allow clutter_actor_get_paint_volume to fail since there are times - such as when an actor isn't a descendent of the stage - when the volume can't be determined. Another example is when something has connected to the "paint" signal of the actor and we simply have no way of knowing what might be drawn in that handler. The API has also be changed to return a const ClutterPaintVolume pointer (transfer none) so we can avoid having to dynamically allocate the volumes in the most common/performance critical code paths. Profiling was showing the slice allocation of volumes taking about 1% of an apps time, for some fairly basic tests. Most volumes can now simply be allocated on the stack; for clutter_actor_get_paint_volume we return a pointer to &priv->paint_volume and if we need a more dynamic allocation there is now a _clutter_stage_paint_volume_stack_allocate() mechanism which lets us allocate data which expires at the start of the next frame. The API has been extended to make it easier to implement get_paint_volume for containers by using clutter_actor_get_transformed_paint_volume and clutter_paint_volume_union. The first allows you to query the paint volume of a child but transformed into parent actor coordinates. The second lets you combine volumes together so you can union all the volumes for a container's children and report that as the container's own volume. The representation of paint volumes has been updated to consider that 2D actors are the most common. The effect apis, clutter-texture and clutter-group have been update accordingly.
2010-09-07 17:04:19 +00:00
else
{
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
clutter_actor_get_allocation_box (priv->actor, &raw_box);
paint volumes: another pass at the design This is a fairly extensive second pass at exposing paint volumes for actors. The API has changed to allow clutter_actor_get_paint_volume to fail since there are times - such as when an actor isn't a descendent of the stage - when the volume can't be determined. Another example is when something has connected to the "paint" signal of the actor and we simply have no way of knowing what might be drawn in that handler. The API has also be changed to return a const ClutterPaintVolume pointer (transfer none) so we can avoid having to dynamically allocate the volumes in the most common/performance critical code paths. Profiling was showing the slice allocation of volumes taking about 1% of an apps time, for some fairly basic tests. Most volumes can now simply be allocated on the stack; for clutter_actor_get_paint_volume we return a pointer to &priv->paint_volume and if we need a more dynamic allocation there is now a _clutter_stage_paint_volume_stack_allocate() mechanism which lets us allocate data which expires at the start of the next frame. The API has been extended to make it easier to implement get_paint_volume for containers by using clutter_actor_get_transformed_paint_volume and clutter_paint_volume_union. The first allows you to query the paint volume of a child but transformed into parent actor coordinates. The second lets you combine volumes together so you can union all the volumes for a container's children and report that as the container's own volume. The representation of paint volumes has been updated to consider that 2D actors are the most common. The effect apis, clutter-texture and clutter-group have been update accordingly.
2010-09-07 17:04:19 +00:00
}
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
box = raw_box;
_clutter_actor_box_enlarge_for_effects (&box);
priv->fbo_offset_x = box.x1 - raw_box.x1;
priv->fbo_offset_y = box.y1 - raw_box.y1;
clutter_actor_box_scale (&box, ceiled_resource_scale);
clutter_actor_box_get_size (&box, &target_width, &target_height);
target_width = ceilf (target_width);
target_height = ceilf (target_height);
paint volumes: another pass at the design This is a fairly extensive second pass at exposing paint volumes for actors. The API has changed to allow clutter_actor_get_paint_volume to fail since there are times - such as when an actor isn't a descendent of the stage - when the volume can't be determined. Another example is when something has connected to the "paint" signal of the actor and we simply have no way of knowing what might be drawn in that handler. The API has also be changed to return a const ClutterPaintVolume pointer (transfer none) so we can avoid having to dynamically allocate the volumes in the most common/performance critical code paths. Profiling was showing the slice allocation of volumes taking about 1% of an apps time, for some fairly basic tests. Most volumes can now simply be allocated on the stack; for clutter_actor_get_paint_volume we return a pointer to &priv->paint_volume and if we need a more dynamic allocation there is now a _clutter_stage_paint_volume_stack_allocate() mechanism which lets us allocate data which expires at the start of the next frame. The API has been extended to make it easier to implement get_paint_volume for containers by using clutter_actor_get_transformed_paint_volume and clutter_paint_volume_union. The first allows you to query the paint volume of a child but transformed into parent actor coordinates. The second lets you combine volumes together so you can union all the volumes for a container's children and report that as the container's own volume. The representation of paint volumes has been updated to consider that 2D actors are the most common. The effect apis, clutter-texture and clutter-group have been update accordingly.
2010-09-07 17:04:19 +00:00
/* First assert that the framebuffer is the right size... */
if (!update_fbo (effect, target_width, target_height, resource_scale))
return FALSE;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
framebuffer = clutter_paint_context_get_framebuffer (paint_context);
cogl_framebuffer_get_modelview_matrix (framebuffer, &old_modelview);
clutter_paint_context_push_framebuffer (paint_context, priv->offscreen);
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* We don't want the FBO contents to be transformed. That could waste memory
* (e.g. during zoom), or result in something that's not rectangular (clipped
* incorrectly). So drop the modelview matrix of the current paint chain.
* This is fine since paint_texture runs with the same modelview matrix,
* so it will come out correctly whenever that is used to put the FBO
* contents on screen...
*/
clutter_actor_get_transform (priv->stage, &modelview);
cogl_framebuffer_set_modelview_matrix (priv->offscreen, &modelview);
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* Save the original viewport for calculating priv->position */
_clutter_stage_get_viewport (CLUTTER_STAGE (priv->stage),
&old_viewport[0],
&old_viewport[1],
&old_viewport[2],
&old_viewport[3]);
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* Set up the viewport so that it has the same size as the stage (avoid
* distortion), but translated to account for the FBO offset...
*/
cogl_framebuffer_set_viewport (priv->offscreen,
-priv->fbo_offset_x,
-priv->fbo_offset_y,
stage_width,
stage_height);
/* Copy the stage's projection matrix across to the framebuffer */
_clutter_stage_get_projection_matrix (CLUTTER_STAGE (priv->stage),
&projection);
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* Now save the global position of the effect (not just of the actor).
* It doesn't appear anyone actually uses this yet, but get_target_rect is
* documented as returning it. So we should...
*/
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
_clutter_util_fully_transform_vertices (&old_modelview,
&projection,
old_viewport,
&local_offset,
&priv->position,
1);
cogl_framebuffer_set_projection_matrix (priv->offscreen, &projection);
cogl_color_init_from_4ub (&transparent, 0, 0, 0, 0);
cogl_framebuffer_clear (priv->offscreen,
COGL_BUFFER_BIT_COLOR |
COGL_BUFFER_BIT_DEPTH,
&transparent);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
cogl_framebuffer_push_matrix (priv->offscreen);
/* Override the actor's opacity to fully opaque - we paint the offscreen
* texture with the actor's paint opacity, so we need to do this to avoid
* multiplying the opacity twice.
*/
priv->old_opacity_override =
clutter_actor_get_opacity_override (priv->actor);
clutter_actor_set_opacity_override (priv->actor, 0xff);
return TRUE;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
static void
clutter_offscreen_effect_real_paint_target (ClutterOffscreenEffect *effect,
ClutterPaintContext *paint_context)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
{
ClutterOffscreenEffectPrivate *priv = effect->priv;
CoglFramebuffer *framebuffer =
clutter_paint_context_get_framebuffer (paint_context);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
guint8 paint_opacity;
paint_opacity = clutter_actor_get_paint_opacity (priv->actor);
cogl_pipeline_set_color4ub (priv->pipeline,
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
paint_opacity,
paint_opacity,
paint_opacity,
paint_opacity);
paint volumes: another pass at the design This is a fairly extensive second pass at exposing paint volumes for actors. The API has changed to allow clutter_actor_get_paint_volume to fail since there are times - such as when an actor isn't a descendent of the stage - when the volume can't be determined. Another example is when something has connected to the "paint" signal of the actor and we simply have no way of knowing what might be drawn in that handler. The API has also be changed to return a const ClutterPaintVolume pointer (transfer none) so we can avoid having to dynamically allocate the volumes in the most common/performance critical code paths. Profiling was showing the slice allocation of volumes taking about 1% of an apps time, for some fairly basic tests. Most volumes can now simply be allocated on the stack; for clutter_actor_get_paint_volume we return a pointer to &priv->paint_volume and if we need a more dynamic allocation there is now a _clutter_stage_paint_volume_stack_allocate() mechanism which lets us allocate data which expires at the start of the next frame. The API has been extended to make it easier to implement get_paint_volume for containers by using clutter_actor_get_transformed_paint_volume and clutter_paint_volume_union. The first allows you to query the paint volume of a child but transformed into parent actor coordinates. The second lets you combine volumes together so you can union all the volumes for a container's children and report that as the container's own volume. The representation of paint volumes has been updated to consider that 2D actors are the most common. The effect apis, clutter-texture and clutter-group have been update accordingly.
2010-09-07 17:04:19 +00:00
/* At this point we are in stage coordinates translated so if
* we draw our texture using a textured quad the size of the paint
* box then we will overlay where the actor would have drawn if it
* hadn't been redirected offscreen.
*/
cogl_framebuffer_draw_textured_rectangle (framebuffer,
priv->pipeline,
0, 0,
cogl_texture_get_width (priv->texture),
cogl_texture_get_height (priv->texture),
0.0, 0.0,
1.0, 1.0);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
static void
clutter_offscreen_effect_paint_texture (ClutterOffscreenEffect *effect,
ClutterPaintContext *paint_context)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
{
ClutterOffscreenEffectPrivate *priv = effect->priv;
CoglFramebuffer *framebuffer =
clutter_paint_context_get_framebuffer (paint_context);
CoglMatrix modelview;
float resource_scale;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
cogl_framebuffer_push_matrix (framebuffer);
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* The current modelview matrix is *almost* perfect already. It's only
* missing a correction for the expanded FBO and offset rendering within...
*/
cogl_framebuffer_get_modelview_matrix (framebuffer, &modelview);
if (clutter_actor_get_resource_scale (priv->actor, &resource_scale) &&
resource_scale != 1.0f)
{
float paint_scale = 1.0f / resource_scale;
cogl_matrix_scale (&modelview, paint_scale, paint_scale, 1);
}
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
cogl_matrix_translate (&modelview,
priv->fbo_offset_x,
priv->fbo_offset_y,
0.0f);
cogl_framebuffer_set_modelview_matrix (framebuffer, &modelview);
/* paint the target material; this is virtualized for
* sub-classes that require special hand-holding
*/
clutter_offscreen_effect_paint_target (effect, paint_context);
cogl_framebuffer_pop_matrix (framebuffer);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
static void
clutter_offscreen_effect_post_paint (ClutterEffect *effect,
ClutterPaintContext *paint_context)
{
ClutterOffscreenEffect *self = CLUTTER_OFFSCREEN_EFFECT (effect);
ClutterOffscreenEffectPrivate *priv = self->priv;
CoglFramebuffer *framebuffer;
g_warn_if_fail (priv->offscreen);
g_warn_if_fail (priv->pipeline);
g_warn_if_fail (priv->actor);
/* Restore the previous opacity override */
if (priv->actor)
{
clutter_actor_set_opacity_override (priv->actor,
priv->old_opacity_override);
}
framebuffer = clutter_paint_context_get_framebuffer (paint_context);
cogl_framebuffer_pop_matrix (framebuffer);
clutter_paint_context_pop_framebuffer (paint_context);
clutter_offscreen_effect_paint_texture (self, paint_context);
}
static void
clutter_offscreen_effect_paint (ClutterEffect *effect,
ClutterPaintContext *paint_context,
ClutterEffectPaintFlags flags)
{
ClutterOffscreenEffect *self = CLUTTER_OFFSCREEN_EFFECT (effect);
ClutterOffscreenEffectPrivate *priv = self->priv;
if (flags & CLUTTER_EFFECT_PAINT_BYPASS_EFFECT)
{
clutter_actor_continue_paint (priv->actor, paint_context);
cogl_clear_object (&priv->offscreen);
return;
}
clutter: Fix offscreen-effect painting of clones `ClutterOffscreenEffect` had been getting the wrong bounding box in the case of clones and descendents of clones, causing visibly incorrect clipping. This was due to `clutter_actor_get_paint_box` only ever being given the source actor during a paint (which is correct) and not the clone. Even if we weren't painting a clone but an offscreened descendent of a clone (like in gnome-shell's desktop zoom), we would get the wrong result. Fortunately we don't need to know the actual clone/actor being painted so don't need to call the problematic `clutter_actor_get_paint_box` at all. The solution is to only keep untransformed rendering in the FBO and leave the correct transformation for later. The correct clone/actor's transformation is already set for us as the current cogl modelview matrix by `clutter_actor_paint`. Bonus optimization: This all means we don't need to keep `last_matrix_drawn` or force a full repaint every time some part of the transformation changes. Because the FBO contents are no longer affected by transformations. As it should be. In other words, offscreen-effected actors can now move around on screen without themselves being repainted. Special thanks to Mai Lavelle for identifying the cause of the problem. Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=789050, https://bugzilla.gnome.org/show_bug.cgi?id=659523#c9, https://gitlab.gnome.org/GNOME/mutter/issues/196, https://gitlab.gnome.org/GNOME/mutter/issues/282, https://gitlab.gnome.org/GNOME/gnome-shell/issues/387, https://launchpad.net/bugs/1767648, https://launchpad.net/bugs/1779615
2018-05-24 09:51:22 +00:00
/* If we've already got a cached image and the actor hasn't been redrawn
* then we can just use the cached image in the FBO.
*/
if (priv->offscreen == NULL || (flags & CLUTTER_EFFECT_PAINT_ACTOR_DIRTY))
{
ClutterEffectClass *effect_class = CLUTTER_EFFECT_GET_CLASS (effect);
gboolean pre_paint_succeeded;
pre_paint_succeeded = effect_class->pre_paint (effect, paint_context);
clutter_actor_continue_paint (priv->actor, paint_context);
if (pre_paint_succeeded)
effect_class->post_paint (effect, paint_context);
else
g_clear_pointer (&priv->offscreen, cogl_object_unref);
}
else
clutter_offscreen_effect_paint_texture (self, paint_context);
}
static void
clutter_offscreen_effect_set_enabled (ClutterActorMeta *meta,
gboolean is_enabled)
{
ClutterActorMetaClass *parent_class =
CLUTTER_ACTOR_META_CLASS (clutter_offscreen_effect_parent_class);
ClutterOffscreenEffect *offscreen_effect = CLUTTER_OFFSCREEN_EFFECT (meta);
ClutterOffscreenEffectPrivate *priv = offscreen_effect->priv;
g_clear_pointer (&priv->offscreen, cogl_object_unref);
parent_class->set_enabled (meta, is_enabled);
}
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
static void
clutter_offscreen_effect_finalize (GObject *gobject)
{
ClutterOffscreenEffect *self = CLUTTER_OFFSCREEN_EFFECT (gobject);
ClutterOffscreenEffectPrivate *priv = self->priv;
g_clear_pointer (&priv->offscreen, cogl_object_unref);
g_clear_pointer (&priv->texture, cogl_object_unref);
g_clear_pointer (&priv->pipeline, cogl_object_unref);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
G_OBJECT_CLASS (clutter_offscreen_effect_parent_class)->finalize (gobject);
}
static void
clutter_offscreen_effect_class_init (ClutterOffscreenEffectClass *klass)
{
ClutterActorMetaClass *meta_class = CLUTTER_ACTOR_META_CLASS (klass);
ClutterEffectClass *effect_class = CLUTTER_EFFECT_CLASS (klass);
GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
klass->create_texture = clutter_offscreen_effect_real_create_texture;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
klass->paint_target = clutter_offscreen_effect_real_paint_target;
meta_class->set_actor = clutter_offscreen_effect_set_actor;
meta_class->set_enabled = clutter_offscreen_effect_set_enabled;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
effect_class->pre_paint = clutter_offscreen_effect_pre_paint;
effect_class->post_paint = clutter_offscreen_effect_post_paint;
effect_class->paint = clutter_offscreen_effect_paint;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
gobject_class->finalize = clutter_offscreen_effect_finalize;
}
static void
clutter_offscreen_effect_init (ClutterOffscreenEffect *self)
{
self->priv = clutter_offscreen_effect_get_instance_private (self);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
/**
* clutter_offscreen_effect_get_texture:
* @effect: a #ClutterOffscreenEffect
*
* Retrieves the texture used as a render target for the offscreen
* buffer created by @effect
*
* You should only use the returned texture when painting. The texture
* may change after ClutterEffect::pre_paint is called so the effect
* implementation should update any references to the texture after
* chaining-up to the parent's pre_paint implementation. This can be
* used instead of clutter_offscreen_effect_get_target() when the
* effect subclass wants to paint using its own material.
*
* Return value: (transfer none): a #CoglHandle or %NULL. The
* returned texture is owned by Clutter and it should not be
* modified or freed
*
* Since: 1.10
*/
CoglHandle
clutter_offscreen_effect_get_texture (ClutterOffscreenEffect *effect)
{
g_return_val_if_fail (CLUTTER_IS_OFFSCREEN_EFFECT (effect),
NULL);
return effect->priv->texture;
}
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
/**
* clutter_offscreen_effect_get_target: (skip)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
* @effect: a #ClutterOffscreenEffect
*
* Retrieves the material used as a render target for the offscreen
* buffer created by @effect
*
* You should only use the returned #CoglMaterial when painting. The
* returned material might change between different frames.
*
* Return value: (transfer none): a #CoglMaterial or %NULL. The
* returned material is owned by Clutter and it should not be
* modified or freed
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
*
* Since: 1.4
*/
CoglMaterial *
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
clutter_offscreen_effect_get_target (ClutterOffscreenEffect *effect)
{
g_return_val_if_fail (CLUTTER_IS_OFFSCREEN_EFFECT (effect),
NULL);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
return (CoglMaterial *)effect->priv->pipeline;
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
/**
* clutter_offscreen_effect_paint_target:
* @effect: a #ClutterOffscreenEffect
* @paint_context: a #ClutterPaintContext
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
*
* Calls the paint_target() virtual function of the @effect
*
* Since: 1.4
*/
void
clutter_offscreen_effect_paint_target (ClutterOffscreenEffect *effect,
ClutterPaintContext *paint_context)
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
{
g_return_if_fail (CLUTTER_IS_OFFSCREEN_EFFECT (effect));
CLUTTER_OFFSCREEN_EFFECT_GET_CLASS (effect)->paint_target (effect,
paint_context);
effect: Add OffscreenEffect The OffscreenEffect class is meant to be used to implement Effect sub-classes that create an offscreen framebuffer and redirect the actor's paint sequence there. The OffscreenEffect is useful for effects using fragment shaders. Any shader-based effect being applied to an actor through an offscreen buffer should be used before painting the resulting target material and not for every actor. This means that doing: pre_paint: cogl_program_use(program) set up offscreen buffer paint: [ actors ] → offscreen buffer → target material post_paint: paint target material cogl_program_use(null) Is not correct. Unfortunately, we cannot really do: post_paint: cogl_program_use(program) paint target material cogl_program_use(null) Because the OffscreenEffect::post_paint() implementation also pops the offscreen buffer and re-instates the previous framebuffer: post_paint: cogl_program_use(program) change frame buffer ← ouch! paint target material cogl_program_use(null) One way to fix it is to allow using the shader right before painting the target material - which means adding a new virtual inside the OffscreenEffect class vtable in additions to the ones defined by the parent Effect class. The newly-added paint_target() virtual allows the correct sequence of actions by adding an entry point for sub-classes to wrap the "paint target material" operation with custom code, in order to implement the case above correctly as: post_paint: change frame buffer cogl_program_use(program) paint target material cogl_program_use(null) The added upside is that sub-classes of OffscreenEffect involving shaders really just need to override the prepare() and paint_target() virtuals, since the pre_paint() and post_paint() do all that's needed.
2010-05-17 10:39:27 +00:00
}
/**
* clutter_offscreen_effect_create_texture:
* @effect: a #ClutterOffscreenEffect
* @width: the minimum width of the target texture
* @height: the minimum height of the target texture
*
* Calls the create_texture() virtual function of the @effect
*
* Return value: (transfer full): a handle to a Cogl texture, or
* %NULL. The returned handle has its reference
* count increased.
*
* Since: 1.4
*/
CoglHandle
clutter_offscreen_effect_create_texture (ClutterOffscreenEffect *effect,
gfloat width,
gfloat height)
{
g_return_val_if_fail (CLUTTER_IS_OFFSCREEN_EFFECT (effect),
NULL);
return CLUTTER_OFFSCREEN_EFFECT_GET_CLASS (effect)->create_texture (effect,
width,
height);
}
/**
* clutter_offscreen_effect_get_target_size:
* @effect: a #ClutterOffscreenEffect
* @width: (out): return location for the target width, or %NULL
* @height: (out): return location for the target height, or %NULL
*
* Retrieves the size of the offscreen buffer used by @effect to
* paint the actor to which it has been applied.
*
* This function should only be called by #ClutterOffscreenEffect
* implementations, from within the #ClutterOffscreenEffectClass.paint_target()
* virtual function.
*
* Return value: %TRUE if the offscreen buffer has a valid size,
* and %FALSE otherwise
*
* Since: 1.8
*
* Deprecated: 1.14: Use clutter_offscreen_effect_get_target_rect() instead
*/
gboolean
clutter_offscreen_effect_get_target_size (ClutterOffscreenEffect *effect,
gfloat *width,
gfloat *height)
{
ClutterOffscreenEffectPrivate *priv;
g_return_val_if_fail (CLUTTER_IS_OFFSCREEN_EFFECT (effect), FALSE);
priv = effect->priv;
if (priv->texture == NULL)
return FALSE;
if (width)
*width = cogl_texture_get_width (priv->texture);
if (height)
*height = cogl_texture_get_height (priv->texture);
return TRUE;
}
/**
* clutter_offscreen_effect_get_target_rect:
* @effect: a #ClutterOffscreenEffect
* @rect: (out caller-allocates): return location for the target area
*
* Retrieves the origin and size of the offscreen buffer used by @effect to
* paint the actor to which it has been applied.
*
* This function should only be called by #ClutterOffscreenEffect
* implementations, from within the #ClutterOffscreenEffectClass.paint_target()
* virtual function.
*
* Return value: %TRUE if the offscreen buffer has a valid rectangle,
* and %FALSE otherwise
*
* Since: 1.14
*/
gboolean
clutter_offscreen_effect_get_target_rect (ClutterOffscreenEffect *effect,
graphene_rect_t *rect)
{
ClutterOffscreenEffectPrivate *priv;
g_return_val_if_fail (CLUTTER_IS_OFFSCREEN_EFFECT (effect), FALSE);
g_return_val_if_fail (rect != NULL, FALSE);
priv = effect->priv;
if (priv->texture == NULL)
return FALSE;
graphene_rect_init (rect,
priv->position.x,
priv->position.y,
cogl_texture_get_width (priv->texture),
cogl_texture_get_height (priv->texture));
return TRUE;
}