To make easier "lazy apply" on the customer OpenGL shaders, the easiest way was to add an accessor to current OSG state's UniformMap.
I've also added accessors for modes and texture, since it could be usefull in the same way.
All methods are const, so I think there is no side-effects."
I've also checked the modified files still build ok with other
compilers (Linux gcc, Windows Visual Studio).
osgDB/OutputStream.cpp and osgPlugins/lws/SceneLoader.cpp require
stdlib.h for atoi use.
In osg/Uniform.cpp the compiler complains that base_class is unknown
unless I add a class name qualifier.
Not a build fix, but I spotted a typo in osgUtil/SceneView."
Provided are lua, python and V8 (for javascript) plugins that just open up enough of a link to the respective libs to run a script, there is no scene graph <-> script communication in current implementation.
But for glPrograms, in order to get all osg's uniform management system to work, I had to subclass osg::program::PerContextProgram.
Here is a modified version of this class, which add some "virtual" method to allow easy subclassing."
osgViewer::Renderer doesn't use these enum settings so now no longer has a calls StateSet::clear() or StateSet::setGlobalDefaults() on the osg::Camera's StateSet. Previously these were being
called and breaking the ability to attached state to Camera's StateSet.
OpenSceneGraph/include\osg/BufferObject(701): warning C4138: '*/' found outside of comment (E:\osg\osgSvn\OpenSceneGraph\src\osg\Array.cpp)
adding a space before /* fixes the problem
void removeClient(osg::Object * /*client*/) { --_numClients; }
"
"The idea of this new OpenGL feature is :
- set RestartIndex = "n"
- draw elements strip
-> when the index is "n", the strip is "stopped" and restarted
It's very usefull for drawing tiles with a single strip and a "restart" at the end of each row.
The idea a an OSG StateAttribute is :
Usually we use to build geometry from code, because software modelers rarely support it (and 3d file formats doesn't support it) :
-RootNode <= "PrimitiveRestartIndex=0" // So now, we know that our restart index is 0 for all drawables under this node
|
- Drawable 1 : triangles => as usual
|
- Drawable 2 : triangles strip => as usual
|
- Drawable 3 : triangles strip + "GL_PRIMITIVE_RESTART" mode = ON => use the restart index
|
- Drawable 4 : triangles strip + "GL_PRIMITIVE_RESTART" mode = ON => use the restart index
|
- Drawable 5 : triangles strip => as usual
With a StateAttribute, it's easy for the developper to say "0 will be my restart index for all this object" and then activate the mode only on some nodes.
The main problem is if you set and restart index value which is not included in the vertex array (for exemple set restart index = 100 but you have only 50 vertex). There is no problem with OpenGL, but some OSG algorithms will try to access the vertex[100] and will segfault.
To solve this, I think there is two ways :
1/ add restart index in osg::PrimitiveSet and use this value in all algorithms. It's a lot of work, maybe dangerous, and it concern only a few situations : developpers who use this extension should be aware of advanced OpenGL (and OSG) data management
2/ use a StateAttribute, and choose a "correct" restart index. In my applications, I always use "0" as a restart index and duplicate the first vertex (vertex[0] = vertex[1]). So there is no difference for OpenGL and all OSG algorithms works properly.
"
"The attached file contains:
- a per-context read counter in GLBufferObject::BufferEntry
- a global client counter in BufferData
- the glue between Texture* and Image client counter
"
I think this is necessary on OpenGL 3.2+ since this is no more "default" locations in the OpenGL specs.
The default behaviour stay the same.
There is a few new methods on osg::State :
- resetVertexAttributeAlias : reset all vertex alias to osg's default ones
- set**Alias : set a vertex attribute alias configuration
- setAttributeBindingList : set the attribute binding list (allow to specify an empty list if you're using "layout" qualifier in glsl code to specify the bindings. This save some CPU operations)"
// 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."
following in the qopengl.h header:
# include <QtGui/qopengles2ext.h>
# ifndef GL_DOUBLE
# define GL_DOUBLE GL_FLOAT
# endif
# ifndef GLdouble
typedef GLfloat GLdouble;
# endif
Unfortunately, when building for normal OpenGL (not GL/ES!) on Windows
with MSVC2012, GLdouble is not defined (it is not a macro but typedef)
and the code above produces a conflicting definition, making the
compile fail. I am attaching a bit hackish workaround for this problem
in osg/GL "
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.
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 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
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."
with startThreading/stopThreading the _drawQueue and _availableQueue
are not reset properly. This can lead to a deadlock when threading is
started again. So before threading is started again the queues must be
reset. This deadlock is also reported earlier by someone else in here:
http://forum.openscenegraph.org/viewtopic.php?p=43415#43415"
* resthttp/osc: encapsulate RequestHandler-classes in their own namespaces to prevent class-name-lookup-errors in the debugger/code (had some weird crashes)
* QTKit: fixed a compile-bug for gcc and blocks
* osgPresentation: click_to_* will fire on RELEASE, only if the drawable received a PUSH beforehand
* p3d/osgPresentation: implemented "forward_mouse_event_to_device"-tag, which will forward mouse-events to all registered devices of a viewer, if an intersection occurs. The mouse-coordinates get reprojected
* present3d: all devices get registered with the viewer
* osgViewer: only devices which are capable of receiving events are queried for new events.
* GraphicWindowIOS: added a flag to GraphicWindowIOS::WindowData to set up a retained backing buffer (defaults to false) This will enable read-back of the render-buffer with glReadPixels even after the renderbuffer got presented
* curl: added an optimized check for file-existance, now only the headers are requested and checked, instead of reading the whole file and handle it with a ReaderWriter
* p3d: fixed a bug, where the existence of a local file may prevent the remote loading of a file with the same name.
"
The set method modify the buffer object of the BufferData while the get method returned the buffer object of the Image.
I've also removed the _bufferObject member of Image (not used anymore)."
--This line, Lionel Lagardeand those below, will be ignored--
M include/osg/Image
* ZeroConfDevice does now return FILE_NOT_HANDLED instead of FILE_NOT_FOUND
* present3D supports multiple devices per env-var P3D_DEVICE, separate multiple device with a space
I refactored parts the p3d-plugin, the curl-plugin and parts of Registry and ReaderWriter. Currently the p3d-plugin tries to open all remote files with the help of the curl-plugin.
I added a new method to Registry called getReaderWriterForProtocolAndExtension. which will return a ReaderWriter which is capable in handling the remote file for the given protocol and extension. If no readerwriter is found for the given extension, a list is built of all readerwriters supporting the given protocol and this list is checked for support of wildcards (extension = "*"). If anything matches it get returned.
I added this principle also to the Registry, so now it's possible to register a generic ReaderWriter which can handle all filetypes for a given protocol, similar what curl is doing. All you have to do is to load the plugin at startup. The curl-fallback is still in place.
With these changes it is now possible to reference a movie inside a presentation without a server-address, read the presentation (with curl) and stream the movie with the correct plugin (e.g. QTKit)
"
* avfoundation: added support for IOS (CoreVideo-support is still in development, works only for SDK >= 6.0, set IPHONE_SDKVER in cMake accordingly)
* zeroconf: added ZeroConf-device-plugin (Mac/Win only, linux implementation missing) to advertise and discover services via ZeroConf/Bonjour, on windows you'll need the Bonjour SDK from Apple
* osgosc: modified the example to demonstrate the usage of the ZeroConf-plugin (start the example with the command-line-argument --zeroconf)
* SlideShowConstructor: enable/disable CoreVideo via a environment variable (P3D_ENABLE_CORE_VIDEO)
* RestHttp: mouse-motion-events get interpolated
* RestHttp: unhandled http-requests get sent as an user-event to the event-queue, all arguments get attached as user-values to the event
* modified some CMakeModules to work correctly when compiling for IOS
* fixed a compile-error for IOS in GraphicsWindowIOS
* some minor bugfixes"
This issue can be reproduced:
1. Create osgViewer window,
2. Push right&left mouse buttons on the osgViewer window,
3. Move mouse out of window, and release right&left mouse buttons.
osgViewer window handle only first mouse release, as result window thinks that we did not released second mouse button.
I attached fix for this issue."
I fixed some bugs and did some more tests with both of the video-plugins. I integrated CoreVideo with osgPresentation, ImageStream has a new virtual method called createSuitableTexture which returns NULL for default implementations. Specialized implementations like the QTKit-plugin return a CoreVideo-texture. I refactored the code in SlideShowConstructor::createTexturedQuad to use a texture returned from ImageStream::createSuitableTexture.
I did not use osgDB::readObjectFile to get the texture-object, as a lot of image-related code in SlideShowConstructor had to be refactored to use a texture. My changes are minimal and should not break existing code.
There's one minor issue with CoreVideo in general: As the implementation is asynchronous, there might be no texture available, when first showing the video the first frame. I am a bit unsure how to tackle this problem, any input on this is appreciated.
Back to the AVFoundation-plugin: the current implementation does not support CoreVideo as the QTKit-plugin supports it. There's no way to get decoded frames from AVFoundation stored on the GPU, which is kind of sad. I added some support for CoreVideo to transfer decoded frames back to the GPU, but in my testings the performance was worse than using the normal approach using glTexSubImage. This is why I disabled CoreVideo for AVFoundation. You can still request a CoreVideoTexture via readObjectFile, though.
"
Added template readFile(..) function to make it more convinient to cast to a specific object type.
Added support for osgGA::Device to osgViewer.
Added sdl plugin to provides very basic joystick osgGA::Device integration.
macro, which could set version within brackets and reset it after
that. All related serializers are also modified so that the
backward-compatibility bug reported by Farshid can be fixed.
"
From Robert Osfield, removed the use of osg::Referenced and creating the proxy object on the heap.
I worked with a osg::Constraint and found strange part of code:
class OSGMANIPULATOR_EXPORT Constraint : public osg::Referenced
{
public:
...
virtual bool constrain(ScaleUniformCommand& command) const { return constrain((MotionCommand&)command); }
virtual bool constrain(const Rotate3DCommand& command) { return constrain((MotionCommand&)command); }
...
If i use osgManipulator::Rotate3DCommand then method Rotate3DCommand::accept(const Constraint& constraint) calls Constraint::constrain(MotionCommand&) instead Constraint:: constrain(const Rotate3DCommand&).
If you replace
virtual bool constrain(const Rotate3DCommand& command) { return constrain((MotionCommand&)command); }
on to
virtual bool constrain(Rotate3DCommand& command) const { return constrain((MotionCommand&)command); }
then all works correctly.
"
I tested it with clang 3.1 and it seems that clang is enforcing the use of the same type for all parameters in this builtin. Looking at the function declaration [1]
bool __sync_bool_compare_and_swap (type *ptr, type oldval type newval, ...)
it seems to be doing the right thing: here the same type is used for *ptr, oldval and newval.
[1] http://gcc.gnu.org/onlinedocs/gcc/_005f_005fsync-Builtins.html#g_t_005f_005fsync-Builtins
"
and TabPlaneDragger.cpp:
TabPlaneDragger(float handleScaleFactor=20.0f);
The reason for this is that the default OSG tab sizes are way bigger than
those we used in our application so far. And since handleScaleFactor_
is already a (constant) class member, I see no objection against making
it user defined."
My previous patch for Atomic Counter Uniform provide new template implementation
of Matrix{2,3,4}x{2,3,4}{fd}. This new implementation use Column-Major Matrix.
Original code define matrix as Row-Major matrix like other Matrix in OSG, and
my matrix implementation break compatibility with previous code.
For example osg_normalMatrix define in osg::State report by Roland Hill.
Thanks to Paul Martz to spot me when the bug appear."
- add non square matrix
- add double
- add all uniform type available in OpenGL 4.2
- backward compatibility for Matrixd to set/get an float uniform matrix
- update of IVE / Wrapper ReadWriter
implementation of AtomicCounterBuffer based on BufferIndexBinding
add example that use AtomicCounterBuffer and show rendering order of fragments,
original idea from geeks3d.com."
I have also made changes to the RotateCylinderDragger to provide a cylinder ring with a thickness. It is totally optional, but IMHO makes the default behavior work better than a solid cylinder (which typically obscures the geometry you are trying to drag). Gives it a bit more to grab, especially in the case where eyepoint and cylinder axis are near parallel.
"
* If the eyepoint and cylinder axis are close to parallel (given some tolerance), then it uses a plane perpendicular to the cylinder axis.
* Otherwise it uses a plane parallel to the cylinder axis oriented towards the eyepoint (previous behavior). This gives decent behavior and is the only path that was taken in the previous code. I kept with previous behavior and that allowed a good bit of code to be removed, simplifying things. There is now no need for the _onCylinder flag, but since there is a public accessor, I wasn't sure how to handle it for backwards compatibility, so I left it in. NOTE - there is no default initialized value, so if it is kept in, it should be set to 'false' to keep same behavior as before. I am not quite sure how the _onCylinder case was supposed to behave as even forcing that path gave undesirable behavior, even with carefully controlled dragging.
"
in osg::Program::PerContextProgram :
typedef std::vector<UniformModifiedCountPair> LastAppliedUniformList;
should be
typedef std::map<unsigned int, UniformModifiedCountPair> LastAppliedUniformList;
Intel driver can use index uniform value > 200000.
With a std::vector, this index uniform value generate an out of memory error
Nothing in OpenGL or GLSL specification define index uniform value rules.
And all other implementation that deal with uniform index in osg::Program
use a std::map.
This fix could have a little performance impact but this is the cost
to pay to work with
all driver."
Bug description:
Let's say we have class A
namespace Bug
{
class A : public osg::Object
{
public:
//...
typedef std::vector<osg::ref_ptr<A> > AList;
protected:
AList _alist;
//...
}
}
REGISTER_OBJECT_WRAPPER( A,
new Bug::A,
Bug::A,
"osg::Object Bug::A" )
{
ADD_LIST_SERIALIZER(A,Bug::A::AList);
}
Bug:
We create say 3 instances of class A: A1,A2,A3 and then we add A2 and A3 and A1 as child instances of A1 so we get next structure:
A1
|- A2,A3,A1
we call osgDB::writeObjectFile(A1,"/data/a.osgt") -> saved correctly( third element in list is saved as unique id that references parentClass
now we call
A1 = osgDB::readObjectFile("/data/a.osgt");
Everything is deserialized correctely except last element in list which should be same instance as parent A1.
The attached code resolves this issue by passing UniqueID in readObjectFields method and saving object in _identifierMap as soon as we have valid object instance so we can make reference to parent object from any child instance.
"
I think the best way to achieve this is to overwrite the DatabasePager::addDatabaseThread() method within the customized database pager. However this method is not 'virtual' yet, so I propose to make the method 'virtual'."
Without the change the application does not work properly. First I get the notification that an OpenGL error occured. After some more of this error messages I see broken textures on the screen. With the changes attached to this message my application works as intended."
Note from Robert Osfield, changed the Image::supportsTextureSubloading() to be const and to be implemented in the .cpp rather than inline.
The initialisation of glu low level tessellator is done in osgUtil::Tessellator::beginTessellation()
This function is not virtual, preventing any customization of the tesselation.
In particular, there in an option in glu tesselator that force the generated primitives to be triangles and that I'd like to use (GLU_TESS_EDGE_FLAG).
"
"I've made a few changes to osgUtil::PolytopeIntersector so that it
actually uses double precision floating point numbers everywhere (as
long as OSG_USE_FLOAT_PLANE is not defined).
I needed double precision intersections in a project I am working on.
These changes fixed the problems I was having -- this is all testing I
have done.
Notice that I have changed
osgUtil::PolytopeIntersector::Intersection's members to use doubles
(osg::Vec3d, instead of osg::Vec3). I could have added #ifdef's there
too, but I think it is better to not change the types of stuff in the
public interface depending on some preprocessor definition.
The modified files are attached. A diff also follows, for those who like it."
With the following changes from Robert Osfield:
"I've just reviewed your changes and have just tweaked them a little to
streamline them. What I have done in the PolytopeIntersector header
is add:
typedef osg::Plane::Vec3_type Vec3_type;
And then use this typedef in the definition of the vertices rather
then Vec3d as you did. Next changes were to PolytopeInteresector.cpp
where to the PolytopeIntersectorUtils defintions of the Vec3_type, and
value_type which now simply read:
typedef osg::Plane::Vec3_type Vec3_type;
typedef Vec3_type::value_type value_type;
This way I was able to complete avoid any if def's and have essential
the same implementation as you achieved. Changes now checked into
svn/trunk."
retessellatePolygons was applying the winding and boundary option.
Moved the gluTessProperty calls into beginTessellation().
There's a comment typo fix, removing an unused VertexPointList
typedef, and allocates one _tobj instead of one per tesellation.
Protections were added to check that _tobj was allocated in the few
remaining places it wasn't being checked.
---
On a side note, I would like to avoid the 'new Vec3d' in
Tessellator::addVertex for each call to
gluTessVertex(tess, location, data).
The RedBook leaves it ambiguous if the location pointer must
remain valid after gluTessVertex or not.
http://www.opengl.org/sdk/docs/man/xhtml/gluTessVertex.xml
says that changing location is not safe, so being conservative, I'll
leave it as is, even though the Mesa GLU library copies the data not
the pointer, so it is currently safe."
- apply() and reset() methods made virtual to allow overriding
- added apply(StateSet&) to make more easier to gather StateAttribute
statistics in user-derived classes
"
I found that some of the items that had been paged in were being expired on the first frame that they were not visible (as the cache was full). This resulted in excessive paging every time the view was moved. With the following changes I could only allow children to be expired if they had not been used for e.g. 30 seconds or 60 frames."
