to make is easier to write code that can work on DrawElementUByte, UShort or UInt.
Changed the osgTerrain::GeometryTechnique so that it automatically chooses
the use of DrawElementUShort or DrawElementsUInt accordining to the size of the tile.
Because there already exists the option to convert slow path geometry to the fast path by computing an internal fast path alternative, I added a new optimizer option that automatically does this. To check the results I also made some changes to the statistics gathering and rendering.
Somewhat unrelated, but also part of the optimizer I disabled removal of CameraView nodes during RemoveRedundantNodes optimization.
As discussed on the ML, CameraViews were removed from the scenegraph. This solves that issue.
Summary:
-Geometry::areFastPathsUsed now also looks at internalOptimizedGeometry
-Added Optimize option to make all slow path geometry compute their internal fast path alternative
-Added fast geometry counter to the statistics
-Disabled removel of CameraViews in optimizer
"
circumstances under which this bug occur are rather specific, but the
basic problem occurs when one translation unit other than libosg.so
constructs an object that is a subclass of osg::Shape and another
translation unit other than libosg.so tries to perform a dynamic_cast or
other RTTI-based operation on that object. Under these circumstances,
the RTTI operation will fail. In my case, the translation units involved
were an application and osgdb_ive.so. The application constructed a
scene graph that included instantiations of subclasses of osg::Shape.
Depending on how the user ran the application, it would write the scene
graph to an IVE file using osgDB::writeNodeFile(). The dynamic_cast
operations in DataOutputStream::writeShape() would fail on the first
subclass of osg::Shape that was encountered. This is because there were
two different RTTI data objects for all osg::Shape subclasses being
compared: one in the application and one in osgdb_ive.so.
The fix for this is simple. We must ensure that at least one member
function of each of the subclasses of the polymorphic type osg::Shape is
compiled into libosg.so so that there is exactly one RTTI object for
that type in libosg.so. Then, all code linking against libosg.so will
use that single RTTI object. The following message from a list archive
sort of explains the issue and the solution:
http://aspn.activestate.com/ASPN/Mail/Message/1688156
While the posting has to do with Boost.Python, the problem applies to
C++ libraries in general."
The fix was to convert the osg::State to use C pointers for the set of applied PerContexProgram objects, and use the osg::Oberver mechanism to avoid dangling pointers for being maintained in osg::State.
methods
getProjectionMatrixAsOrtho()
getProjectionMatrixAsFrustum()
getProjectionMatrixAsPerspective()
getViewMatrixAsLookAt() (2x)
are now const, as they only call const methods of osg::Matrixf/d.
"
returns true if (the extension string is supported or GL version is greater than or equal to a specified version) and
non extension disable is used. This makes it possible to disable extensions that are now
available as parts of the core OpenGL spec.
Updated Texture.cpp is use this method.
/** Convinience method that sets the update callback of the node if it doesn't exist, or nest it into the existing one. */
void addUpdateCallback(NodeCallback* nc);
/** Convinience method that removes a given callback from a node, even if that callback is nested. There is no error return in case the given callback is not found. */
void removeUpdateCallback(NodeCallback* nc);
... and the same for Event and Cull callbacks methods."
--------------------------------------------
Descripton:
The patch does provide a new class PixelDataBufferObject which is capable of allocating memory on the GPU side (PBO memory) of arbitrary size. The memory can then further be used to be enabled into read mode (GL_PIXEL_UNPACK_BUFFER_ARB) or in write mode (GL_PIXEL_PACK_BUFFER_ARB). Enabling the buffer into write mode will force the driver to write data from bounded textures into that buffer (i.e. glGetTexImage). Using buffer in read mode give you the possibility to read data from the buffer into a texture with e.g. glTexSubImage or other instuctions. Hence no data is copied over the CPU (host memory), all the operations are done in the GPU memory.
--------------------------------------------
Compatibility:
The new class require the unbindBuffer method from the base class BufferObject to be virtual, which shouldn't break any functionality of already existing classes. Except of this the new class is fully orthogonal to existing one, hence can be safely added into already existing osg system.
--------------------------------------------
Testing:
The new class was tested in the current svn version of osgPPU. I am using the new class to copy data from textures into the PBO and hence provide them to CUDA kernels. Also reading the results back from CUDA is implemented using the provided patch. The given patch gives a possibility of easy interoperability between CUDA and osg (osgPPU ;) )
--------------------------------------------
I think in general it is a better way to derive the PixelBufferObject class from PixelDataBufferObject, since the second one is a generalization of the first one. However this could break the current functionality, hence I haven't implemented it in such a way. However I would push that on a stack of wished osg 3.x features, since this will reflect the OpenGL PBO functionality through the classes better.
"
* When used PDS RenderStage::runCameraSetUp sets flag that FBO has already stencil,depth buffer attached. Prevents adding next depth buffer.
* Sets correct traits for p-buffer if used PDS and something goes wrong with FBO setup or p-buffer is used directly.
* Adds warning to camera if user add depth/stencil already attached through PDS.
* Sets blitMask when use blit to resolve buffer.
There is also new example with using multisampled FBO."
I have not reverted added Compiler options. I assume that one may want to have warnings enabled for the application but may not want to see them while OSG libraries and examples compile.
Modified files:
osg/Export - now explicitly includes osg/Config to make sure OSG_DISABLE_MSVC_WARNINGS is read
osg/Config.in - declares OSG_DISABLE_MSVC_WARNINGS flag to be added to autogenerated osg/Config
CMakeLists.txt - declares OSG_DISABLE_MSVC_WARNINGS as option with default ON setting
"
facilitate the subclassing of Image providing interactive behaviours so as
used in the vnc interactive VncImage class.
osgViewer::InteractiveImageHandler provides an event handler that convertes osgGA
mouse and keyboard events into the coordinate frame of an image based on ray intersection with geometry in
the associated subgraph.
Changed the ordering of events processing in Viewer and CompositeViewer to allow
scene graph event handlers to take precidence over viewer event handlers and camera manipulators
To facilitate this possibility I have added a template copy constructor and a template assignment operator. These work for safe conversions like those in my previous example, but fail as desired for non-safe conversions since the template instantiation fails during the raw pointer copy/assignment.
The lack of this facility has appreciably handicapped development of generic algorithms applied to the scene graph, and I believe that this alteration will be welcomed by most of the community. If there are issues with antiquated compilers not supporting template member functions then perhaps a small piece of conditional compilation is in order.
I have made the change using the latest release version (2.6) as the base. The full file is attached.
I have not tested building the osg wrappers and the script binding code. I have tested the legal and illegal conversions of both copying and assignment to ensure that behave as expected on MSVC7.1, MSVC8, MSVC9, GCC 4.3."
set.
The optimization is based on the observation that matrix matrix multiplication
with a dense matrix 4x4 is 4^3 Operations whereas multiplication with a
transform, or scale matrix is only 4^2 operations. Which is a gain of a
*FACTOR*4* for these special cases.
The change implements these special cases, provides a unit test for these
implementation and converts uses of the expensiver dense matrix matrix
routine with the specialized versions.
Depending on the transform nodes in the scenegraph this change gives a
noticable improovement.
For example the osgforest code using the MatrixTransform is about 20% slower
than the same codepath using the PositionAttitudeTransform instead of the
MatrixTransform with this patch applied.
If I remember right, the sse type optimizations did *not* provide a factor 4
improovement. Also these changes are totally independent of any cpu or
instruction set architecture. So I would prefer to have this current kind of
change instead of some hand coded and cpu dependent assembly stuff. If we
need that hand tuned stuff, these can go on top of this changes which must
provide than hand optimized additional variants for the specialized versions
to give a even better result in the end.
An other change included here is a change to rotation matrix from quaterion
code. There is a sqrt call which couold be optimized away. Since we divide in
effect by sqrt(length)*sqrt(length) which is just length ...
"
they are written to disk, either inline or as an external file. Added support for
this in the .ive plugin. Default of WriteHint is NO_PREFERNCE, in which case it's
up to the reader/writer to decide.
"I made several modifications:
* The cause of my errors was that my OSG source directory path contains spaces. To fix this issue I wrapped all paths with quotes, as stated in doxygen documentation.
* I also received some warning messages about deprecated doxygen settings, which I fixed by updating the doxygen file, i.e. running \u2018doxygen \u2013u doxygen.cmake\u2018. By running this command deprecated doxygen options are removed, some option comments have changed and quite some options have been added (I kept their default settings unless mentioned).
* I was surprised to find that the doxygen OUTPUT_DIRECTORY was set to \u201c${OpenSceneGraph_SOURCE_DIR}/doc\u201d, which does not seem appropriate for out of source builds; I changed this to \u201c${OpenSceneGraph_BINARY_DIR}/doc\u201d. (On the other hand, maybe a cmake selectable option should be given to the user?)
* Fixed two warnings I received about unexpected end-of-list-markers in \u2018osg\AnimationPath and \u2018osgUtil\CullVisitor due to excess trailing points in comments.
* Fixed a warning in osgWidget\StyleInterface due to an #include directive (strangely) placed inside a namespace.
* Fixed a warning in osg\Camera due to the META_Object macro that confused doxygen. Adding a semi-colon fixed this.
* Removed auto_Mainpage from the INCLUDE option, because I am positive that this file does not belong there; It never generated useful documentation anyway.
* I added the OSG version number environment variable to the PROJECT_NUMBER option so that the version number is now shown on the main page of generated documentation (e.g. index.html).
* Changed option FULL_PATH_NAMES to YES, but made sure STRIP_FROM_PATH stripped the absolute path until the include dir. This fixed an issue that created mangled names for identical filenames in different directories. E.g. osg/Export and osgDB/Export are now correctly named.
* Changed option SHOW_DIRECTORIES to yes, which is a case of preference I guess.
"
New attribute DatabasePager::_expiryFrames sets number of frames a PagedLOD child is kept in memory. The attribute is set with DatabasePager::setExpiryFrames method or OSG_EXPIRY_FRAMES environmental variable.
New attribute PagedLOD::PerRangeData::_
frameNumber contains frame number of last cull traversal.
Children of PagedLOD are expired when time _AND_ number of frames since last cull traversal exceed OSG_EXPIRY_DELAY _AND_ OSG_EXPIRY_FRAMES respectively. By default OSG_EXPIRY_FRAMES = 1 which means that nodes from last cull/rendering
traversal will not be expired even if last cull time exceeds OSG_EXPIRY_DELAY. Setting OSG_EXPIRY_FRAMES = 0 revokes previous behaviour of PagedLOD.
Setting OSG_EXPIRY_FRAMES > 0 fixes problems of children reloading in lazy rendering applications. Required behaviour is achieved by manipulating OSG_EXPIRY_DELAY and OSG_EXPIRY_FRAMES together.
Two interface changes are made:
DatabasePager::updateSceneGraph(double currentFrameTime) is replaced by DatabasePager::updateSceneGraph(const osg::FrameStamp &frameStamp). The previous method is in #if 0 clause in the header file. Robert, decide if You want to include it.
PagedLOD::removeExpiredChildren(double expiryTime, NodeList &removedChildren) is deprecated (warning is printed), when subclassing use PagedLOD::removeExpiredChildren(double expiryTime, int expiryFrame, NodeList &removedChildren) instead. "
implementation of the atomic increment and decrement into a implementation
file.
This way inlining and compiler optimization can no longer happen for these
implementations, but it fixes compilation on win32 msvc targets. I expect
that this is still faster than with with mutexes.
Also the i386 gcc target gets atomic operations with this patch. By using an
implementation file we can guarantee that we have the right compiler flags
available."
"I have taken the liberty of updating a few files so that there is no longer any derivation from std::vector. I have done this by adding a new file osg/MixinVector and by updating only two others: osg/PrimitiveSet and osg/Array. You will notice that this actually removes what is acknowledged as a \u2018hack\u2019 in osg/PrimitiveSet.
With the original code I did manage to find memory leaks with some compiler options on VC 8 and 9, as well as Intel compiler. I determined the leak existence by instrumenting the destructor code, and by use of a garbage collector as a leak detector (in a similar manner to the Firefox project). Hence in contrast to what I said originally, it is exhibiting symptoms on at least some platforms.
Since I am trying to be a good OSG citizen I got out my editor and started hacking! I have built and tested on Linux (Ubuntu) with GCC 4.x and Windows VC 8 SP1. It appears that nothing is broken, and that I\u2019m using less memory J"
via StateSet::setNestedRenderBin(bool) whether the new RenderBin should be nested
with the existing RenderBin, or be nested with the enclosing RenderStage.
GL_GENERATE_MIPMAP_SGIS is very slow (over half a second for a 720*576
texture). However, glGenerateMipmapEXT() performs well (16ms for the
same texture), so I have modified the attached files to use
Texture::generateMipmap() if glGenerateMipmapEXT is supported, instead
of enabling & disabling GL_GENERATE_MIPMAP_SGIS."
Notes, from Robert Osfield, I've tested the out of the previous path using
GL_GENERATE_MIPMAP_SGIS and non power of two textures on NVidia 7800GT and
Nvidia linux drivers with the image size 720x576 and only get compile times
of 56ms, so the above half second speed looks to be a driver bug. With
Muchael's changes the cost goes done to less than 5ms, so it's certainly
an effective change, even given that Michael's poor expereiences with
GL_GENERATE_MIP_SGIS do look to be a driver bug.
- Solves issues of loading image data into the texture memory
- Print a warning if images are of different dimensions or have different internal formats (GL specification requires images to be the same)
Patch is tested and seems to work fine. It shouldn't break any other functionality. It should go into include/osg and src/osg
"
multi-threaded paging, where the Pager manages threads of reading local
and http files via seperate threads. This makes it possible to smoothly
browse large databases where parts of the data are locally cached while
others are on a remote server. Previously with this type of dataset
the pager would stall all paging while http requests were being served,
even when parts of the models are still loadable virtue of being in the
local cache.
Also as part of the refactoring the DatabaseRequest are now stored in the
ProxyNode/PagedLOD nodes to facilitate quite updating in the cull traversal,
with the new code avoiding mutex locks and searches. Previous on big
databases the overhead involved in make database requests could accumulate
to a point where it'd cause the cull traversal to break frame. The overhead
now is negligable.
Finally OSG_FILE_CACHE support has been moved from the curl plugin into
the DatabasePager. Eventually this functionality will be moved out into
osgDB for more general usage.
from: DEEP_COPY_STATESETS = 8,
to: DEEP_COPY_STATESETS = 1<<3,
showing clearly that this isn't the _value_ 8, but the _bit_ 8. this is an old pattern i see (and like to promulgate) to make code a bit more readable and maintainable.
"
From Robert Osfield, refactored the FrameBufferObejcts::_drawBuffers set up so that its done
within the setAttachment method to avoid potential threading/execution order issues.
Introduced code in BoundgingSphere, BoundingBox, ProxyNode and LOD to utilise the above settings.
Added Matrix::value_type, Plane::value_type, BoundingSphere::value_type and BoundingBox::value_type command line
options that report where the types of floats or doubles.
and a new scheme for computing the scaling when using autoscale that introduces smooth
transitions to the scaling of the subgraph so that it looks more natural.
Attached is a fixed version of OverlayNode.cpp. I fixed CustomPolytope::cut( osg::Plane ) method. Bug was apparent in such scenario:
Let P1 be some random frustum polytope
Let P2 be the polytope that was created from P1 bounding box (P2 contains P1 entirely)
Then ignoring precision errors: P1.cut( P2 ) == P2.cut( P1 ) == P1. But this condition was not always met. Cut failed when some of the polytope reference points happened to lie exactly on some intersecting planes in both P1 & P2 (plane distance was = 0).
I only use CustomPolytope for my shadowing stuff so I did not test how this affects rest of OverlayNode.cpp.
----2----
Also attached is a minor precision improvement for osg::Plane intersect method (double version).
----3----
I have also one observation regarding osg::Plane - There are two intersect vertices methods (float and double flavour):
inline int intersect(const std::vector<Vec3>& vertices) const
inline int intersect(const std::vector<Vec3d>& vertices) const
I guess osg::Plane won't compile when someone changes default vec3 typedef to vec3d. Shouldn't the first method be changed to use vec3f explicitly ? Ie:
inline int intersect(const std::vector<Vec3f>& vertices) const"
pbuffer functions or exactly ask for the extensions we need to call the
apropriate glx extension functions for and around pbuffers extensions.
The glx 1.3 version of this functios are prefered. If this is not pressent we
are looking for the glx extensions and check for them.
Prevously we just used some mix of the glx 1.3 functions or the extension
functions without making sure that this extension is present.
"
creating subclasses of osg::Array that referenced data
stored an application's internal data structures. I took
a stab at implementing that and ran into a couple of
downcasts in Geometry.cpp. Enclosed is my take at fixing
those along with a simple example of how to do this."
StateSet::removeAssociatedModes(const StateAttribute*)
and a
StateSet::removeAssociatedTextureModes(unsigned, const StateAttribute*)
call. These funktions are just missing for a complete api IMO."
highlighted problems with Light, ClipPlane and Hint usage in osg::State's usage of cloneType
and reassignment of target/num in StateSet/these StateAttributes.
- GL2Extensions, Program and Program.cpp
Features:
- Support for fragment output binding. (e.g. You can now specify in the fragment shader varying out vec3 fragOut; fragOut = vec3(1,0,1); to write to the fragOut variable. In your program you call glBindFragDataLocation(program, 1, "fragOut") to bind the fragOut variable with the MRT 1 - GL_COLOR_ATTACHMENT1_EXT)
- new methods Program::add/removeBindFragDataLocation Program::getFragDataBindingList
"
- Implementation of integer textures as in EXT_texture_integer
- setBorderColor(Vec4) changed to setBorderColor(Vec4d) to pass double values
as border color. (Probably we have to provide an overloading function to
still support Vec4f ?)
- new method Texture::getInternalFormatType() added. Gives information if the
internal format normalized, float, signed integer or unsigned integer. Can
help people to write better code ;-)
"
Futher changes to this submission by Robert Osfield, changed the dirty mipmap
flag into a buffer_value<> vector to ensure safe handling of multiple contexts.
local function pointer to avoid compiler warnings related to case void*.
Moved various OSG classes across to using setGLExtensions instead of getGLExtensions,
and changed them to use typedef declarations in the headers rather than casts in
the .cpp.
Updated wrappers
"A new texture class Texture2DArray derived from
Texture extends the osg to support the new
EXT_texture_array extensions. Texture arrays provides
a feature for people interesting in GPGPU programming.
Faetures and changes:
- Full support for layered 2D textures.
- New uniform types were added (sampler2DArray)
- FrameBufferObject implementation were changed to
support attaching of 2D array textures to the
framebuffer
- StateSet was slightly changed to support texture
arrays. NOTE: array textures can not be used in fixed
function pipeline. Thus using the layered texture as a
statemode for a Drawable produce invalid enumerant
OpenGL errors.
- Image class was extended to support handling of
array textures
Tests:
I have used this class as a new feature of my
application. It works for me without problems (Note:
Texture arrays were introduced only for shading
languages and not for fixed function pipelines!!!).
RTT with Texture2DArray works, as I have tested them
as texture targets for a camera with 6 layers/faces
(i.e. replacement for cube maps). I am using the array
textures in shader programming. Array textures can be
attached to the FBO and used as input and as output."
stereo format to work. It's a good thing I tested these on a TV
before submitting them since I did indeed have a bug. One thing I
did not test was to see how this would work in windowed mode. Does
the interlaced stereo code have support for 'absolute' positions?
For example a given pixel on the screen is always shown in a given
eye no matter where the graphics context is placed?
"
to the view to be done during syncronous updateTraversal().
This feature can be used for doing things like merging subgraphs that have been loaded
in a background thread.
