I just added a field version_ to read it at the beginning and added extra code to check it and read the extra field if needed and read the good filename"
been removed in the most recent versions of libavcodec/ffmpeg. You're
already using avcodec_open2() elsewhere, but one appears to have been
missed. The change is trivial:
[skynet](0) $ svn diff
Index: src/osgPlugins/ffmpeg/FFmpegDecoderAudio.cpp
===================================================================
--- src/osgPlugins/ffmpeg/FFmpegDecoderAudio.cpp (revision 13355)
+++ src/osgPlugins/ffmpeg/FFmpegDecoderAudio.cpp (working copy)
@@ -109,7 +109,7 @@
// m_context->flags |= CODEC_FLAG_TRUNCATED;
// Open codec
- if (avcodec_open(m_context, p_codec) < 0)
+ if (avcodec_open2(m_context, p_codec, NULL) < 0)
throw std::runtime_error("avcodec_open() failed");
}
[skynet](0) $
I've applied similar changes this year to many other packages, including
cheese, blender, linphone, ad nauseam. It's been tested by verifying that
with my patch, OSG builds against the newest libavcodec and ffmpeg, whereas
otherwise it does not.
The modified src/osgPlugins/ffmpeg/FFmpegDecoderAudio.cpp is attached. It
originates in svn trunk revision 13355. Thanks!"
// A custom class
namespace CustomDomain {
class MyGroup : public osg::Group
{
public:
META_Node( CustomDomain, MyGroup );
void setMyName( const std::string& n );
const std::string& getMyName() const;
void setMyID( int id );
int getMyID() const;
...
};
}
// The serialization wrapper using a custom domain name
REGISTER_CUSTOM_OBJECT_WRAPPER( MyDomain,
CustomDomain_MyGroup,
new CustomDomain::MyGroup,
CustomDomain::MyGroup,
"osg::Object osg::Node osg::Group CustomDomain::MyGroup" )
{
ADD_STRING_SERIALIZER( MyName, std::string() );
{
UPDATE_TO_VERSION_SCOPED( 1 ); // Updated for a new domain version
ADD_INT_SERIALIZER( MyID, 0 );
}
}
Save the class instance as follows:
osgDB::writeNodeFile( *myGroup, "serializer_test.osgt", new osgDB::Options("CustomDomains=MyDomain:1") );
The output file will include the domain version definition and all the class data, and can be read back. We can also force setting the domain version by the CustomDomains option while reading the saved files. If we save the class instance without any options, MyID will be ignored because the default domain version is 0.
This may help third-party libraries like osgEarth to maintain their own serializers without regarding to the OSG soversion changes.
Another feature added is a more robust binary format, which in fact adds a size-offset at each block's beginning. When there are problems or unsupported data types while reading, we can now directly jump to the block end indicated by the offset value. So a .osgb file will automatically ignore bad data and read remains as normal (at present it will fail at all). This feature will not break the backward compatibility, and can be disabled by setting "RobustBinaryFormat=false" while writing out.
Hope these changes can work smoothly with present and future community projects. Maybe we should also consider have an osgserializer example to test and demonstrate all things we can do now."
a few modifications to these files:
- applications/present3D/Cluster.cpp
- src/osgDB/FileUtils.cpp
The changes are needed to fix a few platform specific things such as the
absence of stat64, int/socklen_t differences, etc... and are in the same
line as Linux, Apple and FreeBSD specific checks.
I have attached the modified files; the original patches to 3.0.1 are
also visible here if you find it more convenient:
http://dl.wolfpond.org/dports/graphics.osg/dragonfly/patch-applications_present3D_Cluster.cpphttp://dl.wolfpond.org/dports/graphics.osg/dragonfly/patch-src_osgDB_FileUtils.cpp
With these changes, osg 3.0.1 is able to be built and packaged on DragonFly
whereas it previously failed to compile."
This is because Qt auto repeats keyboard events, so multiple calls are made to GLWidget::keyPressEvent and GLWidget::keyReleaseEvent by Qt, and subsequently translated to OSG events.
The way to solve this is ignoring key released auto repeated events (see http://qt-project.org/doc/qt-4.8/qkeyevent.html#isAutoRepeat), so multiple KEYDOWN events are fired, but only one KEYUP.
I attach a modified osgQt/GraphicsWindowQt.cpp with this change."
New methods osg::Geometry::containsDeprecatedData() and osg::Geometry::fixDeprecatedData() provide a means for converting geometries that still use the array indices and BIND_PER_PRIMITIVE across to complient
versions.
Cleaned up the rest of the OSG where use of array indices and BIND_PER_PRIMITIVE were accessed or used.
It's
return osg::Image::computeImageSizeInBytes(width, height, depth, packing, slice_packing, image_packing);
but I think it should be
return osg::Image::computeImageSizeInBytes(width, height, depth, pixelFormat, pixelType, packing, slice_packing, image_packing);"
GeometryNew is only temporary and will be renamed to Geometry on the completion of refactoring work and feedback from community.
Ported osggeometry across to use GeometryNew.
The API has changed quite a bit, so lots of changes had to be made in the osg readerwriter. The preious version of the FBX SDK (2013.3) already deprecated a lot of the names and functions. The code I submit now still compiles against 2013.3 (possibly needs a #define FBX_NEW_API). Not sure if that's useful, but it might ease the transition."
a change in version 5.0.0 required a call to the fuction "DGifOpen" to pass an int ref for an error code.
My fix ignores the error, just fixes the compile."
The plugin does not support images without alpha channel for opacity
This seems to indicate that the check for alpha should be against the opacity map. I've attached the updated file.
"
The following change to State.cpp fixes the case where textures in a stateset along with a compute shader program did not make those textures available to the shader, since it got evaluated before the texture bind(s)."
Note from Robert, Michael only change State::apply(StateSet*) so I also applied the same reversal in apply of texutre modes/state into the State::apply() method for consistency.
I looked into the "State::applyShaderComposition" method and noticed that it will print debug information to the osg::INFO stream. I commented that line out and the performance drop was gone.
I'm not sure if the printing was accidentally left in or is meant to be enabled. I've modified the function to only call the print function if INFO logging is enabled. I've attached the change."
- Added missing packing value on S3TC images. Images are coded with 4x4 blocs, whatever the image size. So there is an horizontal packing of 4 pixels (2 bytes in DXT1, 4 bytes in DXT2-5).
- Added crash guard against writing corrupted S3TC images.
Notes:
- What is missing is a support of "lines packing" in osg::Image (see code comments).
- S3TC-DXTC vertical flipping crashes (access violation) with some unusual dimensions (see code). I could not implement missing cases, so I added guards to avoid crashing."
Failure to perform a vertical flip on S3TC-DXTC now simply leaves the original image instead of corrupting it.
Image.cpp was sometimes performing a "normal" (= for uncompressed images) vertical flip on S3TC-DXTC images, producing weird results.
Actually, code was trying a "DXTC vertical flip" and relied on the result to call a "normal vertical flip". But when the "DXTC v-flip" encounters an error, this is is not necessarily because the image is not S3TC (ex: unhandled image dimensions)!
So now the code simply does "if dxtc, then flip_dxtc; else flip_normal;".
Note from Robert Osfield, moved the isDXT function into the dxt_tool file and namespace.
The solution for to refactor the way that events are checked so I add a bool return type to checkEvents() method across osgViewer::GraphcisWindow, osgGA::Devive and osgViewer::Viewer/CompositeViewer classes
Instead of
INCLUDE_DIRECTORIES(${LIBVNCCLIENT_INCLUDE_DIR})
the CMake variable
INCLUDE_DIRECTORIES(${LIBVNCSERVER_INCLUDE_DIR})
should be used.
Attached is a fix for src/osgPlugins/vnc/CMakeLists.txt"
TextureBuffer objects may use osg::Texture::bindToImageUnit(), so GLSL shaders are able to use not only texelFetch() function , but also functions defined in GL_ARB_shader_image_load_store extension : imageLoad(), imageStore(), imageAtomicAdd() etc."
second email: "After a while I found that osg::Texture::applyTexParameters() used with TextureBuffer may cause some OpenGL errors ( applying texture filters and wraps to TextureBuffer makes no sense ) so I fixed it."
I hope the change seems reasonable. I searched the entire code base for code that accesses the StateSetStack, and noticed RenderBin and RenderStage use it. However, I don't think that code would ever be called from within the State::apply() function, so I'm fairly confident the change should be safe."
handle(const PointerInfo&, const osgGA::GUIEventAdapter&, osgGA::GUIActionAdapter&)
returns false.I think it is not correct...i think handled should be changed only when it is really handled."
OSG_DEBUG << "Hello world!\n";
the underlying stream is not automatically flushed. It is flushed
if endl is queued instead of \n:
OSG_DEBUG << "Hello world!" << std::endl;
The notify macros do:
stream->setCurrentSeverity(severity);
return *stream;
So, if a function do:
OSG_DEBUG << "This is a debug message\n";
OSG_NOTICE << "This is a notice message" << std::endl;
the debug message will be classified as a notice message.
It is a problem when the application uses a NotifyHandler. The notify
method of the handler is called with:
severity = NOTICE
message = "This is a debug message\nThis is a notice message"
The attached Notify.cpp contains an automatic flush of the stream when
the current severity changes.
"
If the index is more than the size of _children vector, the new child is pushed at the end of the vector, but the index value is unmodified, so an incorrect value is passed to the "childInserted" method."
1. Respect "noMatrixTransforms" option for matrices attached to meshes, not only those attached to nodes
2. traverse_nodes variable did the exact opposite of it's description.
3. Always try to load referenced textures, even if we can't find the files (in order to the the registry's ReadFileCallback have a stab at it)
"
See attached my ported code, which fulfills:
- using of setUserValue()/getUserValue()
- user-value will be attached to Geode or Drawable
- actually 3 values will be added: SMC (surface), FID (feature) and IRC (IRcolor). Its UserValue-names are < UA:SMC >, < UA:FID > and < UA:IRC > (without spaces!)
- user-value will be attached to Geode or Drawable
- has an option (named preserveNonOsgAttrsAsUserData) for control whether the UserValues should be attached (Default: no UserValues will be attached)
Reading out this values is very simple. Here is an example (uservalue-name without spaces!):
signed short smc;
if (myGeode.getUserValue("< UA:SMC >", smc))
{
doSomethingWithTheGeodeAndItsSmc(myGeode, smc);
}
else
{
// geode has no SMC
}
Header update
FindFFmpeg.cmake has been changed in order to support the new header include format for FFmpeg. In the 1.0 release, a new file had been added with the name “time.h” in the avutil library. The previous method of adding includes caused conflicts with the ANSI C “time.h” file. Now the include directive will only use the main include folder. All files using the old include format have been updated to reflect the change.
Added __STDC_CONSTANT_MACROS define to CMakeLists.txt
Since there is no guarantee that FFmpegHeaders.hpp will be included before stdint.h is included, the define has been moved from FFmpegHeaders.hpp to be part of the CMakeLists.txt for the FFmpeg plugin. This will allow the define to work on all compilers regardless of include order.
Replaced AVFormatParameters with AVDictionary
AVFormatParameters is no longer supported in FFmpeg and has been replaced with a key/value map of strings for each setting. FFmpegParameters and FFmpegDecoder has been updated to reflect this.
Replaced av_open_input_file with avformat_open_input
FFmpeg now opens files using avformat_open_input. Since the av_open_input_file method is deprecated, the FFmpegDecoder class has been updated to reflect this change.
Added custom AVIOContext field to options
Since some formats and inputs may not be supported by FFmpeg, I have added a new parameter that allows a user to allocate their own AVIOContext. This class will allow for creating a read, seek, and write callback if they desire.
Checking for start_time validity
It is possible for some file formats to not provide a start_time to FFmpeg. This would cause stuttering in the video since the clocks class would be incorrect.
Removed findVideoStream and findAudioStream
The new FFmpeg release already has a function that will find the best audio and video stream. The code has been replaced with this function.
Updated error reporting
Some functions would not log an error when opening a file or modifying a file failed. New logs have been added as well as a function to convert error numbers to their string descriptions.
decode_video has been replaced
The old decode_video function would remove extra data that some decoders use in order to properly decode a packet. Now av_codec_decode_video2 has replaced that function.
Picture format changed from RGBA32 to RGB24
Since most video will not contain an alpha channel, using a 24 bit texture will use less memory."
