gmane.comp.graphics.openscenegraph.cvs

osgTextBase bugs

John Swan-Stone — 2012-05-25T21:07:37

Hi,

I ran into an issue with osg::TextBase.cpp not computing the bounding box correctly.  The problem manifested whenever you used user rotations of non-multiple-of-90 degree angles.

I’ve attached a program demonstrating the problem and a patch against the 3.0.1 release that fixed the issue.

John.#include <osg/Geode>
#include <osgViewer/Viewer>
#include <osgGA/TrackballManipulator>
#include <osgText/Text>
#include <osg/PolygonMode>
#include <osg/BlendFunc>

struct BBDrawer : osg::NodeCallback
{
    inline BBDrawer(osg::Group* root) {
        _geom = osg::createTexturedQuadGeometry(osg::Vec3(0,0,0),osg::Vec3(1,0,0),osg::Vec3(0,1,0));
        _geom->setUseDisplayList(false);
        _geom->setUseVertexBufferObjects(false);
        _geom->getOrCreateStateSet()->setAttributeAndModes(new osg::PolygonMode(osg::PolygonMode::FRONT_AND_BACK, osg::PolygonMode::LINE));
        osg::Geode* geode = new osg::Geode;
        geode->addDrawable(_geom);
        root->addChild(geode);
    }

    virtual void operator()(os

Re: Small fix for Uniform serializer

Robert Osfield — 2012-05-25T16:07:22

Thanks Rui, fix merged and submitted to svn/trunk.

On 22 May 2012 11:32, Wang Rui <wangray84-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:
_______________________________________________
osg-submissions mailing list
osg-submissions-ZwoEplunGu0hajLcUbyfC12AsgEQdTeF< at >public.gmane.org
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Re: Three dragger fixes in osgManipulator

Robert Osfield — 2012-05-25T08:57:43

Hi Japp,

I have slept on the issue of how to set the scale of the dragger and
feel the first step is to merge your proposed changes as they are a
good first step and are the least intrusive way to provide the
customisability your require.  These changes are now merged and
submitted to svn/trunk.

Cheers,
Robert.

Re: Three dragger fixes in osgManipulator

Robert Osfield — 2012-05-24T18:17:52

Hi Jaap,

I've reviewed and merged change items A, themissing break, and B,
missing removeDraggerCallback implementation.  These changes are now
in svn/trunk.  I haven't yet merged item C as will need to reflect on
the topic further.

I would recommend updating to svn/trunk so that you can work on all
the additional changes checked into osgManipulator since the 3.0.1
release.

Cheers,
Robert.

On 24 May 2012 16:04, Jaap Glas <Jaap.Glas-T0UFp3o85ygAvxtiuMwx3w< at >public.gmane.org> wrote:
_______________________________________________
osg-submissions mailing list
osg-submissions-ZwoEplunGu0hajLcUbyfC12AsgEQdTeF< at >public.gmane.org
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Re: Three dragger fixes in osgManipulator

Robert Osfield — 2012-05-24T16:52:57

Hi Jaap,

I've just started a review, only looked at TabPlaneDragger change
right now, but one issue that I've spotted it would be worth flagging
for further discussion, I'll add my comments after the section below.
FYI, I'm not the original author of osgManipulator so didn't do the
original design/implementation work, so I like yourself have to learn
about the design/implementation by reading the code.  Unfortunately
the original author isn't active in the community right now so we'll
have to make do without their input on original design intent.    I'll
add my specific comments after the section your original comments
below:

On 24 May 2012 16:04, Jaap Glas <Jaap.Glas-T0UFp3o85ygAvxtiuMwx3w< at >public.gmane.org> wrote:

I've just looked at the virtual Dragger::setupDefaultGeometry() method
and it doesn't have a bool parameter, so the
TabPlaneDragger::setupDefaultGeometry(bool) hides this, which is a
good programming approach, so I think it would be worth us fix this at
the same time as merging your other chang

Three dragger fixes in osgManipulator

Jaap Glas — 2012-05-24T15:04:26

Dear all,

I am an employee of dGB Earth Sciences, and we are currently porting
the 3D visualization of our open-source seismic interpretation package
OpendTect to OpenSceneGraph.

I want to submit three minor dragger fixes in osgManipulator.

It concerns the files Dragger.cpp, TabPlaneDragger and TabPlaneDragger.cpp.
The changed files have been drawn from the last public release OSG 3.0.1,
and have been attached to this email.


A> Missing break in switch statement at following function of Dragger.cpp:

bool Dragger::handle(const osgGA::GUIEventAdapter&, osgGA::GUIActionAdapter&)

The case osgGA::GUIEventAdapter::PUSH has no terminating break statement.
This was discovered because I was twice receiving a DraggerCallback with
its MotionCommand-stage set to osgManipulator::MotionCommand::START.


B> Missing body implementation at following function of Dragger.cpp:

Dragger::removeDraggerCallback(DraggerCallback* dc)

While using DraggerCallbacks in my application, this omission resulted
into a crash at remova

Small fix for osg::copyImage (ImageUtils.cpp)

Marcel Pursche — 2012-05-23T12:52:23

Hello Robert,

I found a bug in the osg::copyImage function when I was trying to write
a programm to create a very large texture atlas and then save it into
small tiles (for virtual texturing). The guard clauses at the beginning
of the copyImage function seem to be buggy.
I wasn't able to copy a small part of a big image(the atlas) into a
smaller image (one tile). I got error messages that the input height was
to big. But it should easily be possible to copy a 256x256 part of the
atlas into a 256x256 output image. On the other hand I was able to
accidently trigger a bad memory access when I was trying to copy a small
image(input texture) into a big one(the atlas), when the destination
offset was not calculatetd correct. The guarding clauses should have
prevented this.
Another small bug is the return value of the function. It is not
documented but I think it should return false if an error occured and
true if the image was copied with no problems. But in the last else case
where the color format needs to be t

Small fix for Uniform serializer

Wang Rui — 2012-05-22T10:32:24

Hi Robert,

A very small but maybe fatal problem was found when I saved models with
shader and uniforms (with new double types support) to osgb format and
tried to read it again. The application will crash here. And the serializer
file should be slightly changed to fix it. Please replace the original file
in osgWrappers/serializers/osg/Uniform.cpp.

Cheers,

Wang Rui
_______________________________________________
osg-submissions mailing list
osg-submissions-ZwoEplunGu0hajLcUbyfC12AsgEQdTeF< at >public.gmane.org
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Re: PrimitiveSet improve

David Callu — 2012-05-21T09:39:24

Hi Robert

Yes, I add an extra level of call but this is just to avoid code duplication
so the most complexe function is call by the less complexe

PrimitiveSet::draw(State& state, bool useVertexBufferObjects, unsigned int
numInstances) const
is call by
PrimitiveSet::draw(State& state, bool useVertexBufferObjects) const

This part of code of OSG is there for a long time, and probably will never
change so code
duplication will probably not hurt futur maintainers.



 I work on SpeedTree integration, and I need to draw a forest in
Multi-thread mode,
with, for example, one view, two camera, two GraphicsContext, so
with CullThreadPerCameraDrawThreadPerContext
threading model. To do this, I need to call the same
osg::PrimitiveSet::draw(...) function to draw a tree, in different thread.
But camera 1 don't see the same number of tree that camera 2. So
numInstance have to be adjusted by each camera.
I could push a mutex in my drawable but this is always good to avoid mutex
usage when we can.
I could duplicate osg::P

Bug in osgUtil::ReversePrimitiveFunctor.cpp

Geoff Michel — 2012-05-17T07:59:08

Robert

The v3.01 implementation of the ReversePrimitiveFunctor is incorrect for triangle_strips - at least with odd numbers of vertices since the last vertex will be swapped with data beyond the end of the index array. In general the method is also incorrect in that the triangulation of the strip will change in the reversed strip - important if the strip is not plane. The attached code corrects this by merely reversing the indices of the strip (sharing code with reversing lines & polygons) - however this will still have an error if the normals are per_primitive since the reversed strip has not reversed the associated normals. 

Any suggestions on how to find the normals associated with the reversed strip(s) appreciated - the normals of course are attached to the osg::Geometry not the primitiveset. Should we pass into the reversefunctor the entire geometry?

Geoff/* -*-c++-*- OpenSceneGraph - Copyright (C) 1998-2006 Robert Osfield 
 *
 * This library is open source and may be redistributed and/or modified un

Android CMake fix for osgViewer

Thomas Hogarth — 2012-05-17T00:11:41

Hi Robert

Attached is a fix for osgViewers CMakeList.txt when configuring for Android
on OSX. Before it was defaulting into the IF(APPLE) branch when selecting
WindowingSystem.

Then when comparing the "macosx10.7" variable an error was occurring.
Presumable as I was targeting Android.

Cheers
Tom
# FIXME: For OS X, need flag for Framework or dylib
IF(DYNAMIC_OPENSCENEGRAPH)
    ADD_DEFINITIONS(-DOSGVIEWER_LIBRARY)
ELSE()
    ADD_DEFINITIONS(-DOSG_LIBRARY_STATIC)
ENDIF()

SET(LIB_NAME osgViewer)

SET(HEADER_PATH ${OpenSceneGraph_SOURCE_DIR}/include/${LIB_NAME})
SET(TARGET_H
    ${HEADER_PATH}/CompositeViewer
    ${HEADER_PATH}/Export
    ${HEADER_PATH}/GraphicsWindow
    ${HEADER_PATH}/Renderer
    ${HEADER_PATH}/Scene
    ${HEADER_PATH}/Version
    ${HEADER_PATH}/View
    ${HEADER_PATH}/Viewer
    ${HEADER_PATH}/ViewerBase
    ${HEADER_PATH}/ViewerEventHandlers
)

SET(LIB_COMMON_FILES
    CompositeViewer.cpp
    GraphicsWindow.cpp
    HelpHandler.cpp
    Renderer.cpp
    Scene.cpp
    ScreenCaptureHandler.

Bugfix to prevent crash for NVTTProcessor.cpp

Tassilo Glander — 2012-05-16T13:42:14

Hi Robert,

the current NVTTProcessor.cpp crashed on my system in the convertRGBAToBGRA method. For me it looks as if it accidentially uses the n variable to access the red channel. I fixed it by reading column*4+0 as it is done for the other channels. The error was introduced with the change to non-contiguous images in OSG

old code: 
[code]
// Convert RGBA to BGRA : nvtt only accepts BGRA pixel format
void NVTTProcessor::convertRGBAToBGRA( std::vector<unsigned char>& outputData, const osg::Image& image )
{
    unsigned int n=0;
    for(int row=0; row<image.t(); ++row)
    {
        const unsigned char* data = image.data(0,row);
        for(int column=0; column<image.s(); ++column)
        {
            outputData[n] = data[column*4+2];
            outputData[n+1] = data[column*4+1];
            outputData[n+2] = data[column*4+n]; // TG crash here
            outputData[n+3] = data[column*4+3];
            n+=4;
        }
    }
}
[/code]

fixed code: 
[code]
// Convert RGBA to BGRA : nvtt only accepts BGRA

Added Max PagedLOD Level property to CullSettings

John Vidar Larring — 2012-05-15T12:58:56

Hi Robert,

With refernce to email "[osg-users] Strategies for reducing visual 
terrain-tile-popping with osgTerrain" (2012-04-24), I've made a small 
patch that can alleviate some of the worst "in-your-face" popping by 
adding CullSettings::setMaxPagedLODLevel(unsigned int). In conjunction 
with LODScale, this setting can be tuned to minimize terrain tile 
popping for flight paths that do not vary to much in altitude.

The changes are made against the 3.0.1 tag. Note that the changes have 
only been tested in SingleTreaded mode

If the addition/changes are acceptable, please feel free to make the 
changes you deem necessary for inclusion in OSG:-) Would have preferred 
not to introduce a dynamic_cast in PageLOD::traverse()....

Best regards,
John

Re: ViewDependentShadowMap improvements

Wang Rui — 2012-05-15T04:18:41

Hi Robert,

Sorry for a little late. I'd like to submit a newer version of my VDSM
implementation which provides PSSM/VSM support and a flexible shader code
generator. There are also some more functionalities which can the light
frustum and output the shadow textures as wish. I was about to add some
more changes like ESM and fix possible problems of VSM (doesn't work with
some very simple scenes), but as there are plenty of work for you to do
before I could have enough time so I would submit current implementation
for the community to consider and discuss.

I'm also interested in a newer implementation called QVSM (Queried Virtual
Shadow Maps, a kind of tiled shadow maps) which appears in ShaderX 5 for
the first time. It is a deferred shadowing technique and seems to perfectly
solve the aliasing problems without rendering the scene for multiple times.
Anyone who has experience about that and could help me finish an OSG
implementation would be appreciated. :-)

Cheers,

Wang Rui


2012/5/3 Wang Rui <wangray84

IOS CMake adjustment for tutorial

Thomas Hogarth — 2012-05-14T23:16:52

Hi Robert

Attached is a change to the main CMakeLists.txt which just sets the IOS sdk
paths to default to the new /Application/Xcode path. It also sets the
default sdk version to 5.1 and the default min version to 4.0

This is to fit in with the new XCode and the new tutorial on the osg site.

I've seen the thread on automatic selection of the sdk path, which I'm
planning to look into, but this will make things easier on new comers till
then.

Cheers
Tom
set(CMAKE_ALLOW_LOOSE_LOOP_CONSTRUCTS TRUE)
set_property(GLOBAL PROPERTY USE_FOLDERS ON)

IF(WIN32)
    CMAKE_MINIMUM_REQUIRED(VERSION 2.4.6 FATAL_ERROR)
ELSE(WIN32)
    IF(APPLE)
        CMAKE_MINIMUM_REQUIRED(VERSION 2.6.0 FATAL_ERROR)
    ELSE(APPLE)
        CMAKE_MINIMUM_REQUIRED(VERSION 2.4.4 FATAL_ERROR)
    ENDIF(APPLE)
ENDIF(WIN32)

if(COMMAND cmake_policy)
    # Works around warnings libraries linked against that don't
    # have absolute paths (e.g. -lpthreads)
    cmake_policy(SET CMP0003 NEW)

    # Works around warnings about escaped quotes in

a few nvtt fixes

Bob Kuehne — 2012-05-05T19:18:39

hi robert,
attached are a few nvtt fixes for the latest nvtt, osg(r13078), and osx.
bob

osgText bugs

John Swan-Stone — 2012-05-05T01:15:24

Hi,
I’ve run into three bugs with osgText.cpp with OSG 3.0.1.

The first happens when you use setCharacterSizeMode(osgText::Text::OBJECT_COORDS_WITH_MAXIMUM_SCREEN_SIZE_CAPPED_BY_FONT_HEIGHT); and you place text in any plane other than osgText::Text::XY_PLANE  The essence of the issue is the auto-sizing code doesn’t work correctly for text in any plane other than XY_PLANE.  I’ve attached the program and a screen shot showing the problem. The fix was simply scaling all 3 dimensions instead of just X and Y.

The second showed up with using osgText::Text::FILLEDBOUNDINGBOX draw mode with osgText::TextLYGON_OFFSET backdropImplementation with a osglygonOffset stateAttribute attached.   osgText.cpp wasn’t respecting the PolygonOffset in the current State.   I also noticed some incorrect values in a couple glPushAttrib calls.

I’ve attached a patch file generated by SVN.

Thanks,
John.#include <osg/Geode>
#include <osgViewer/Viewer>
#include <osgGA/TrackballManipulator>
#include <osgText/Text>

void main(

Add key binding help for osgmanipulator

Thrall, Bryan — 2012-05-03T20:09:55

Add key bindings to the help and add on-screen help for osgmanipulator to make it easier to figure out how to activate the draggers.

Modified from r13078.
--
Bryan Thrall
Principal Software Engineer
FlightSafety International
bryan.thrall-A7WBat7Xm6L28lPXxEEMUQ< at >public.gmane.org
  


_______________________________________________
osg-submissions mailing list
osg-submissions-ZwoEplunGu0hajLcUbyfC12AsgEQdTeF< at >public.gmane.org
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Re: ViewDependentShadowMap improvements

Wang Rui — 2012-05-02T19:58:33

Hello Robert,

I'm on a business trip at present and will be in my own home this weekend.
I would submit some recent changes for VDSM at that time (and also check
the osgb multi-platform problem which is reported in another thread). Hope
it will still be in time.

Cheers,

Wang Rui


2012/5/2 Robert Osfield <robert.osfield-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>

_______________________________________________
osg-submissions mailing list
osg-submissions-ZwoEplunGu0hajLcUbyfC12AsgEQdTeF< at >public.gmane.org
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Re: ViewDependentShadowMap improvements

Robert Osfield — 2012-05-02T13:22:25

Hi Rui,

Do you have any latest updates?  I'm about to dive in and do a review.

Cheers,
Robert.

On 23 April 2012 16:52, Wang Rui <wangray84-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:
_______________________________________________
osg-submissions mailing list
osg-submissions-ZwoEplunGu0hajLcUbyfC12AsgEQdTeF< at >public.gmane.org
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Text3D vertical layout lines count

Mihail Baranov — 2012-04-26T15:24:04

Good day!

Could you pls consider a change to text3D implementation:
In case of VERTICAL layout we use linelength variable but never increase its value during charcode iterations.
What I suggest:


Code:
diff --git a/3rdparty/osg/OpenSceneGraph/src/osgText/Text3D.cpp b/3rdparty/osg/OpenSceneGraph/src/osgText/Text3D.cpp
index a4ae866..e7c2108 100644
--- a/3rdparty/osg/OpenSceneGraph/src/osgText/Text3D.cpp
+++ b/3rdparty/osg/OpenSceneGraph/src/osgText/Text3D.cpp
< at >< at > -362,6 +362,7 < at >< at > void Text3D::computeGlyphRepresentation()
previous_charcode = charcode;
}
+linelength++;
}
}
else





Modified Text3D.cpp attached.

Cheers,
Mihail

------------------
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=47323#47323




Attachments: 
http://forum.openscenegraph.org//files/text3d_438.cpp