Reflect camera support in Blackberry-specific documentation page.
Change-Id: I5267a32b99aa88cd73f33fe045d67579442f9932
Reviewed-by: Bernd Weimer <bweimer@rim.com>
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Fix the selection of the video viewfinder resolution depending
on the features provided by the camera. If the viewfinder is responsible
for encoding the video frames, the resolution of the viewfinder
and the video output must match. Otherwise only the ratio must match
and we can use a lower resolution for the viewfinder to improve
the performance of the NV12->RGB frame conversion.
Change-Id: I66ace84841a69be84f83bf9c5240cd23d96fbb76
Reviewed-by: Bernd Weimer <bweimer@rim.com>
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
This change modifies setVideoSurface() in the snippet code
to be a bit more robust.
Task-number: QTBUG-29383
Change-Id: I2c691f19fe8f0912c1287597045da3a5579d2108
Reviewed-by: Yoann Lopes <yoann.lopes@digia.com>
warning: cast to pointer from integer of ifferent size.
Change-Id: I4294fe7b67c56d1eca59f81c3af26919ad5089a8
Reviewed-by: Yoann Lopes <yoann.lopes@digia.com>
Parameters are DWORD_PTR instead of DWORD, which did not matter
on 32bit but caused MinGW64-compiler warnings and potential
crashes on 64bit.
Change-Id: I840ae3ea7ea532746f73f20b3233cae842931d45
Reviewed-by: Yoann Lopes <yoann.lopes@digia.com>
Before audio is played we always seek to position 0. Unfortunately, due
to a recent change in mmrenderer, playback stopped working for
"non-seekable" media. There is a check now, whether the media is
seekable or not.
Change-Id: Ieafd8d1364f7ce0194f4fa17d3efe894aa1b289b
Reviewed-by: Fabian Bumberger <fbumberger@rim.com>
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Ensure that the taken photos are always upright.
As long as the QCamera API does not provide information about
the native orientation of the camera, we have to do the rotation
inside the BBCamera plugin.
Change-Id: Iebb04a5417274c2baee5e944dadf10fc706484d9
Reviewed-by: Bernd Weimer <bweimer@rim.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Listen to the CAMERA_STATUS_POWERUP signal, which is sent
after standby, and trigger a restart of the viewfinder
in this case.
Change-Id: I426f23657844bb270ddfd0e4bfb35880040aa583
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Make sure that the texture is upright when the device is upright.
When rotating the device, it's the applications responsibility to
rotate the texture of the viewfinder (e.g. through the
QDeclarativeVideoOutput::orientation property).
Change-Id: I6e453ce9bdde5197104533f844b1079b3c3db908
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reimplement all camera related controls as they are supported by the
BB10 camera API.
Change-Id: Ie7d2258be22c1c1821489bd7074b80cff4242afa
Reviewed-by: Bernd Weimer <bweimer@rim.com>
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
qcamera.cpp:183: Cannot invoke tr() like this
Change-Id: Ia8f29c02bbc4b91c6542d9ed44136094e1e8a2e1
Reviewed-by: Jan Arve Sæther <jan-arve.saether@digia.com>
Move the mediaplayer related code in its own subdirectory.
This is a preparation for adding camera support in the future.
Change-Id: I880ff90e154165c2a3a31fd2b51e1dd224fea8fd
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Replace 'headerfile' with 'qmlclass' in qdocconf file.
As a positive side-effect, QDoc will also add files for
'All members' for QML Types into .qhp as well.
Task-number: QTBUG-29226
Change-Id: I892f386cbd22bfbf7a04be5aecc800896e1090ec
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
Reviewed-by: Geir Vattekar <geir.vattekar@digia.com>
Classes with virtuals are used polymorphically (why have virtuals
otherwise?), so they need virtual destructors to be deleted
properly. Unless they are never deleted using objects of those two
classes -- but why have the interface in the first place if that's the
case?
This is binary incompatible change. It was agreed upon in the mailing
list.
Change-Id: I697e4bd53251452a0e6b0c09edd08c4835f90cbd
Discussed-on: http://lists.qt-project.org/pipermail/development/2012-December/008908.html
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Having a code snippet use a slot (stateChanged) with a name
identical to a signal in QAudioInput / QAudioOutput classes causes
problems (incorrect links) in class documentation that refer to
the snippet code.
This change renames the slots according to standard Qt convention.
Same change is applied to example applications as well.
Task-number: QTBUG-26688
Change-Id: I0c6181240237d0c642b73fe18f3ddb5137b7f207
Reviewed-by: Dmytro Poplavskiy <dmytro.poplavskiy@gmail.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
It also applies to QGraphicsVideoItem when used on a GL viewport.
We now have a new video sink that is based on Microsoft's EVR sink, we just
replace the default Presenter with our own. Frames are rendered into D3D
surfaces using DXVA, then copied into a shared D3D/EGL surface and finally
bound to a GL texture to be used by the video surface.
The shared D3D/EGL surface is a feature provided by ANGLE and therefore Qt
must be compiled with ANGLE for this new video sink to be compiled and
used.
Change-Id: I0b7b9968eed5488f9ef1a2dcca5213bd0af232ab
Reviewed-by: Yoann Lopes <yoann.lopes@digia.com>
When using our custom MediaSink with RGB formats, Media Foundation fails
sometimes to resolve the topology. Inserting ourselves a ColorConverter
transform in the topology resolves the problem.
The ColorConverter transform cannot handle dynamic frame size changes
(this can happen with H264 videos for example) so we also need to insert a
Resizer transform to handle transparently frame size changes.
Change-Id: Id7f37a0af65f142fbe6d420ad7b2c1ac2156c21b
Reviewed-by: Yoann Lopes <yoann.lopes@digia.com>
- Fixed \snippet paths
- Minor language edits
- Removed unnecessary multiple page commands in a single comment
block. For example, \fn and \qmlsignal in a single comment block
is not allowed. Such instances must be documented in separate
comment blocks.
Change-Id: I65f4518499e2600c4e1807356d4116c575e48c19
Reviewed-by: Andy Nichols <andy.nichols@digia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
(cherry picked from commit 51334d2995)
QVideoSurfaceGLPainter should not assume that the GL texture provided
by the QVideoFrame is fully initialized for drawing.
Change-Id: I752ecf69a3cf1faf0a7cf8f84b372c3b347824be
Reviewed-by: Gunnar Sletta <gunnar.sletta@digia.com>
- Fixed \snippet paths
- Minor language edits
- Removed unnecessary multiple page commands in a single comment
block. For example, \fn and \qmlsignal in a single comment block
is not allowed. Such instances must be documented in separate
comment blocks.
Change-Id: I65f4518499e2600c4e1807356d4116c575e48c19
Reviewed-by: Andy Nichols <andy.nichols@digia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
The Video item inline example was wrong, changing read-only properties.
Task-number: QTBUG-28741
Change-Id: Ia1f6c1110a301a96743e7ebfec51731fcaf6a862
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Dmytro Poplavskiy <dmytro.poplavskiy@gmail.com>
YUV and other compressed frame formats are always TopToBottom. Only for uncompressed RGB formats the sign of their height is used to indicate the scanline direction.
Change-Id: I6474783583672861eeeb538b79d14ee3d6995f59
Reviewed-by: Dmytro Poplavskiy <dmytro.poplavskiy@gmail.com>
Top & bottom of the texture coordinates are already flipped for BottomToTop scanline directions.
So there is no need to flip top & bottom of the vertex coordinates too.
Change-Id: I221b94afb65c194bdedb787414e6a436b8095fb2
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reviewed-by: Dmytro Poplavskiy <dmytro.poplavskiy@gmail.com>
CGRect == NSRect only in 64-bit mode.
Don't add -framework QuickTime based on the host arch,
use QT_ARCH instead.
Change-Id: I829bc1bcfe37f70a07e8c55a5bc7dee90aaa981a
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@digia.com>
For the VideoRendererControl, also stop the video surface at the same
time.
This fixes a crash when changing video.
Change-Id: I49484f8b277c345dafb3e5947cf5d23df15546f3
Reviewed-by: Jason Barron <jason@cutehacks.com>
Fixed the way the custom MF Transform (getting the frames) works:
- Recreate it whenever we load a new media
- During media type negotiation between nodes, the MFT should support
the same types as the video sink supports
- Allow input and output types to be changed as many times as needed,
otherwise the topology cannot be resolved in some cases
Change-Id: I7ca77e1a3dee83643f1a97f2e6ada9c5c0e88309
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
