Now that we typically have premultiplied data stored in Cogl
textures, when fetching a texture into local data for temporary
storage, use a premultiplied format to avoid an unpremultiply/
premultiply roundtrip.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Many operations, like mixing two textures together or alpha-blending
onto a destination with alpha, are done most logically if texture data
is in premultiplied form. We also have many sources of premultiplied
texture data, like X pixmaps, FBOs, cairo surfaces. Rather than trying
to work with two different types of texture data, simplify things by
always premultiplying texture data before uploading to GL.
Because the default blend function is changed to accommodate this,
uses of pure-color CoglMaterial need to be adapted to add
premultiplication.
gl/cogl-texture.c gles/cogl-texture.c: Always premultiply
non-premultiplied texture data before uploading to GL.
cogl-material.c cogl-material.h: Switch the default blend functions
to ONE, ONE_MINUS_SRC_ALPHA so they work correctly with premultiplied
data.
cogl.c: Make cogl_set_source_color() premultiply the color.
cogl.h.in color-material.h: Add some documentation about
premultiplication and its interaction with color values.
cogl-pango-render.c clutter-texture.c tests/interactive/test-cogl-offscreen.c:
Use premultiplied colors.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
RGBA data in X pixmaps and in FBOs is already premultiplied; use
the right format when creating cogl textures.
http://bugzilla.openedhand.com/show_bug.cgi?id=1406
Signed-off-by: Robert Bragg <robert@linux.intel.com>
The load-finished signal has a GError* argument which is meant to
signify whether the loading was successful. However many of the
places in ClutterTexture that emit this signal directly pass their
'error' variable which is a GError** and will be NULL or not
completely independently of whether there was an error. If the
argument was dereferenced it would probably crash.
The test-texture-async interactive test case should also verify
that the ::load-finished signal is correctly emitted.
Fixes bug:
http://bugzilla.openedhand.com/show_bug.cgi?id=1622
The texture filters are now a property of the material layer rather
than the texture object. Whenever a texture is painted with a material
it sets the filters on all of the GL textures in the Cogl texture. The
filter is cached so that it won't be changed unnecessarily.
The automatic mipmap generation has changed so that the mipmaps are
only generated when the texture is painted instead of every time the
data changes. Changing the texture sets a flag to mark that the
mipmaps are dirty. This works better if the FBO extension is available
because we can use glGenerateMipmap. If the extension is not available
it will temporarily enable automatic mipmap generation and reupload
the first pixel of each slice. This requires tracking the data for the
first pixel.
The COGL_TEXTURE_AUTO_MIPMAP flag has been replaced with
COGL_TEXTURE_NO_AUTO_MIPMAP so that it will default to
auto-mipmapping. The mipmap generation is now effectively free if you
are not using a mipmap filter mode so you would only want to disable
it if you had some special reason to generate your own mipmaps.
ClutterTexture no longer has to store its own copy of the filter
mode. Instead it stores it in the material and the property is
directly set and read from that. This fixes problems with the filters
getting out of sync when a cogl handle is set on the texture
directly. It also avoids the mess of having to rerealize the texture
if the filter quality changes to HIGH because Cogl will take of
generating the mipmaps if needed.
Instead of passing a boolean value, the ::allocate virtual function
should use a bitmask and flags. This gives us room for expansion
without breaking API/ABI, and allows to encode more information to
the allocation process instead of just changes of absolute origin.
The documentation for ClutterTexture's set_from_rgb_data() and
set_from_yuv_data() says:
Note: This function is likely to change in future versions.
This is not true, since they'll remain for the whole 1.x API cycle.
The setup_viewport() function should only be used by Clutter and
not by application code.
It can be emulated by changing the Stage size and perspective and
requeueing a redraw after calling clutter_stage_ensure_viewport().
The CoglTexture constructors expose the "max-waste" argument for
controlling the maximum amount of wasted areas for slicing or,
if set to -1, disables slicing.
Slicing is really relevant only for large images that are never
repeated, so it's a useful feature only in controlled use cases.
Specifying the amount of wasted area is, on the other hand, just
a way to mess up this feature; 99% the times, you either pull this
number out of thin air, hoping it's right, or you try to do the
right thing and you choose the wrong number anyway.
Instead, we can use the CoglTextureFlags to control whether the
texture should not be sliced (useful for Clutter-GST and for the
texture-from-pixmap actors) and provide a reasonable value for
enabling the slicing ourself. At some point, we might even
provide a way to change the default at compile time or at run time,
for particular platforms.
Since max_waste is gone, the :tile-waste property of ClutterTexture
becomes read-only, and it proxies the cogl_texture_get_max_waste()
function.
Inside Clutter, the only cases where the max_waste argument was
not set to -1 are in the Pango glyph cache (which is a POT texture
anyway) and inside the test cases where we want to force slicing;
for the latter we can create larger textures that will be bigger than
the threshold we set.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Signed-off-by: Neil Roberts <neil@linux.intel.com>
Now that everything is float, the marsharlling function of the
size-change signal should reflect that fact.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
Most of the operations involving the texture's allocated area require
floats -- either for computations or for setting the geometry into
COGL. So it doesn't make any sense to use get_allocation_coords() and
cast everything to floats.
ClutterTexture has many properties that can only be accessed using
the GObject API. This is fairly inefficient and makes binding the
class overly complicated.
The Texture class should have accessor methods for all its properties,
properly documented.
The stencil buffer is always cleared the first time a clip is used
that needs it and the stencil test is disabled otherwise so there is
no need to clear before a paint.
COGLenum, COGLint and COGLuint which were simply typedefs for GL{enum,int,uint}
have been removed from the API and replaced with specialised enum typedefs, int
and unsigned int. These were causing problems for generating bindings and also
considered poor style.
The cogl texture filter defines CGL_NEAREST and CGL_LINEAR etc are now replaced
by a namespaced typedef 'CoglTextureFilter' so they should be replaced with
COGL_TEXTURE_FILTER_NEAREST and COGL_TEXTURE_FILTER_LINEAR etc.
The shader type defines CGL_VERTEX_SHADER and CGL_FRAGMENT_SHADER are handled by
a CoglShaderType typedef and should be replaced with COGL_SHADER_TYPE_VERTEX and
COGL_SHADER_TYPE_FRAGMENT.
cogl_shader_get_parameteriv has been replaced by cogl_shader_get_type and
cogl_shader_is_compiled. More getters can be added later if desired.
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
There were several functions I believe no one is currently using that were
only implemented in the GL backend (cogl_offscreen_blit_region and
cogl_offscreen_blit) that have simply been removed so we have a chance to
think about design later with a real use case.
There was one nonsense function (cogl_offscreen_new_multisample) that
sounded exciting but in all cases it just returned COGL_INVALID_HANDLE
(though at least for GL it checked for multisampling support first!?)
it has also been removed.
The MASK draw buffer type has been removed. If we want to expose color
masking later then I think it at least would be nicer to have the mask be a
property that can be set on any draw buffer.
The cogl_draw_buffer and cogl_{push,pop}_draw_buffer function prototypes
have been moved up into cogl.h since they are for managing global Cogl state
and not for modifying or creating the actual offscreen buffers.
This also documents the API so for example desiphering the semantics of
cogl_offscreen_new_to_texture() should be a bit easier now.
Redundant clearing of depth and stencil buffers every render can be very
expensive, so cogl now gives control over which auxiliary buffers are
cleared.
Note: For now clutter continues to clear the color, depth and stencil buffer
each paint.
When replacing the fbo_handle with a new handle it first unrefs the
old handle. This was previously a call to cogl_offscreen_unref which
silently ignored attempts to unref COGL_INVALID_HANDLE. However the
new cogl_handle_unref does check for this so we should make sure the
handle is valid to avoid the warning.
Bug 1565 - test-fbo case failed in many platform
clutter_texture_new_from_actor was broken because it created the FBO
texture but then never attached it to the material so it was never
used for rendering. The old behaviour in Clutter 0.8 was to assign the
texture directly to priv->tex. In 0.9 priv->tex is replaced with
priv->material which has a reference to the tex in layer 0. So putting
the FBO texture directly in layer 0 more closely matches the original
behaviour.
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
The cogl_is_* functions were showing up quite high on profiles due to
iterating through arrays of cogl handles.
This does away with all the handle arrays and implements a simple struct
inheritance scheme. All cogl objects now add a CoglHandleObject _parent;
member to their main structures. The base object includes 2 members a.t.m; a
ref_count, and a klass pointer. The klass in turn gives you a type and
virtual function for freeing objects of that type.
Each handle type has a _cogl_##handle_type##_get_type () function
automatically defined which returns a GQuark of the handle type, so now
implementing the cogl_is_* funcs is just a case of comparing with
obj->klass->type.
Another outcome of the re-work is that cogl_handle_{ref,unref} are also much
more efficient, and no longer need extending for each handle type added to
cogl. The cogl_##handle_type##_{ref,unref} functions are now deprecated and
are no longer used internally to Clutter or Cogl. Potentially we can remove
them completely before 1.0.
Only have load-data-async and load-async properties, both are construct
only and the latter adds the former load-size-async behavior on top of
load-data-async.
The fog and perspective API is currently split in two parts:
- the floating point version, using values
- the fixed point version, using structures
The relative properties are using the structure types, since they
are meant to set multiple values at the same time. Instead of
using bare values, the whole API should be coalesced into two
simple calls using structures to match the GObject properties.
Thus:
clutter_stage_set_fog (ClutterStage*, const ClutterFog*)
clutter_stage_get_fog (ClutterStage*, ClutterFog*)
clutter_stage_set_perspective (ClutterStage*, const ClutterPerspective*)
clutter_stage_get_perspective (ClutterStage*, ClutterPerspective*)
Which supercedes the fixed point and floating point variants.
More importantly, both ClutterFog and ClutterPerspective should
using floating point values, since that's what get passed to
COGL anyway.
ClutterFog should also drop the "density" member, since ClutterStage
only allows linear fog; non-linear fog distribution can be achieved
using a signal handler and calling cogl_set_fog() directly; this keeps
the API compact yet extensible.
Finally, there is no ClutterStage:fog so it should be added.
Instead of creating a separate thread for each texture load, the loads
are now queued in a thread pool which will use at most 3 textures at a
time. This avoids the potentially large overhead of creating and
destroying a thread for every image while still avoiding throttling
the CPU if a large number of textures are loaded at the same time.
If the load is cancelled the old code needed to join the thread which
effectively meant it had to wait for the load to finish. With this
model the threads are owned by the thread pool so there is no need to
wait for them to finish. Instead, each thread is given its own thread
data which we can use to mark the thread as aborted. Once the load is
finished the thread will check whether it is aborted before setting
the texture data. That way the ClutterTexture can just disown threads
that it needs to cancel.
This causes clutter to skip all the GL work of state changes and
texturing if the opacity was 0. This is done in ClutterTexture and not
ClutterActor to ensure that pre and post paint signals work correctly.
Other expensive actors should be doing the same thing.
When setting the COGL texture handle for a ClutterTexture the
texture will be set as the first layer of the material used
by the ClutterTexture. The documentation should clarify this
point.
The :texture and :material properties of ClutterTexture use a
Clutter-provided GType shielding from CoglHandle. Since CoglHandle
now has a GType we can use COGL_TYPE_HANDLE instead.
This commit also removes the conditional compilation of the
:material property, as it makes little sense now that the
Materials API has landed.
cogl_paint_init was a bit too miscellaneous; it mainly cleared the color, depth
and stencil buffers but arbitrarily it also disabled fogging and lighting.
It no longer disables lighting, since we know Cogl never enables lighting and
disabling of fog is now handled with a seperate function.
Since I noticed cogl_set_fog was taking a density argument documented as
"Ignored" I've also added a mode argument to cogl_set_fog which exposes the
exponential fog modes which can make use of the density.
Bug #1453 - Asynchronous texture loading can starve cpu.
Add a mutex that is held in the loader threads during the image
decoding. We were spawning and starting a thread for each asynchronously
loaded texture. This can cause cpu / memory starvation when many pixbuf
loaders allocate their temporary memory at the same time.
Also added -fno-strict-aliasing to MAINTAINER_CFLAGS in configure.ac
to avoid incorrect warnings caused by the static mutex code.
To convert from a TextureQuality to a COGL filter enumeration
we need to perform two function calls. Since we need both filters
when calling cogl_texture_set_filters() we can safely consolidate
the two conversion functions into one. This also allows using a
shorter function name, thus reducing the lenght of the lines
involved and, more importantly preventing Emmanuele from crying.
Otherwise if the dipose method is called twice the material will
already be destroyed when texture_free_gl_resources is called so it
will issue a warning about an invalid handle.
The 'no_slice' property means the texture should never be sliced. We
want cogl to create a texture with any amount of waste so we pass
max_waste as -1. However this got broken in commit 168d55 so that the
meaning got negated (no_slice enabled slicing).
