Re: [osg-users] PolytopeIntersector LIMIT_NEAREST (or LIMIT_ONE) different distances for different groupnodes

2016-06-01 Thread David Knipp 
Hi robertosfield,

Thank you for you reply. 

This absolutley understandable. Every group can have it's own transformation 
and so every group has it's own local coordinates. Quite clear.

So i guess there is absolutley no way to caluclate distances in 
Worldcoordinates?

I admit, I don't understand much about how the IntersectionVisitor works and I 
still searching for the right lines in the osgUtil Project. 

I've just written simpel boolean collision code before.

PS: Would you mind to remove my email address from the qoute? : D I actually 
like that to be private. : ) 

Cheers,
NoxxKn

--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=67326#67326





___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org

Re: [osg-users] PolytopeIntersector LIMIT_NEAREST (or LIMIT_ONE) different distances for different groupnodes

2016-06-01 Thread Robert Osfield 
Hi David,

Way too much detail in a difficult form to follow when skimming over
things to spot what might going amiss.

The best I can suggest right now is think about the coordinate frame
that the distance and coordinate values are being reported in.  If you
have transforms in your scene graph then the coordinates generated on
one subgraph can be different to another subgraph as each subgraph now
has it's own coordinate frame.  This means the same point in world
coordinates can have different local coordinates.

Robert.

On 1 June 2016 at 15:07, David Knipp  wrote:
> Hi,
>
> I'm very sorry that i have to create a new post. I'm currently facing a 
> problem i can't solve on my own.
>
> Im working on a single Model which has a lot of group nodes and geometry 
> nodes. This model is categorized in a few "main" groups.
>
> In the beginning i worked only with one of this main groups at once.
>
> The application does view this group and handle pickings on them. Everything 
> worked nice and cleanly. After i was mixing some of the groups the results of 
> the picker got strange.
>
> Groups got different distances and also the differences change if i just 
> change the angle i'm looking at the model. (orbit manipulator! I don't change 
> the distance!)
>
> The strange behavior is that i start picking nodes that where BEHIND of the 
> nodes i wanted to pick (in viewer).
>
> I'm working with a picker like so:
>
> int x = GET_X_LPARAM(lParam); // windows mouse pos
> int y = GET_Y_LPARAM(lParam); // windows mouse pos
>
> float _x = static_cast(x);
> float _y = static_cast(y);
>
> _x = 2.f * (_x - 0.f) / (window_width - 0.f) - 1.f;
> _y = -(2.f * (_y - 0.f) / (window_height - 0.f) - 1.f);
>
> double w = 0.005, h = 0.005;
> osgUtil::PolytopeIntersector * picker = new 
> osgUtil::PolytopeIntersector(osgUtil::Intersector::CoordinateFrame::PROJECTION,
>  _x - w, _y - h, _x + w, _y + h);
>
> picker->setIntersectionLimit(osgUtil::Intersector::LIMIT_NEAREST);
> picker>setPrecisionHint(osgUtil::Intersector::PrecisionHint::USE_DOUBLE_CALCULATIONS);
> picker->setDimensionMask(osgUtil::PolytopeIntersector::AllDims);
>
> osgUtil::IntersectionVisitor iv(picker);
> iv.setTraversalMode(osg::NodeVisitor::TRAVERSE_ALL_CHILDREN);
>
> viewer->getCamera()->accept(iv);
>
> if (picker->containsIntersections()) {
>const osg::NodePath& nodePath =
>picker->getFirstIntersection().nodePath;
>unsigned int idx = nodePath.size();
>while (idx--) {
> // bla bla specifc
> }
> }
>
> What i mean is described in Attachments.
>
> First Image "pick-error.png":
>
> Red: Calculated nearer (but are in viewer BEHIND!)
> Green: Calculated further away (Actually the node i wanted to pick)
>
> Second Image "pick-error2.png":
>
> I just change the angle of the orbit viewer for about 30 degrees and don't 
> change the distance to object. The Distances just explode to much higher 
> values. (Note the Model should only be about "2.f" heigh!)
>
> I guess the second image error was allways present, but i didn't realised it, 
> because the picker worked allways right.
>
> Another side node, the model seems to doesn't stand in the right position. It 
> should be in 0 0 0 but is located anywhere like 1 -11 -180 or something like 
> that. Which is very strange but never made a problem.
> For loading i do:
>
> // ... osg::MatrixTransform mt
> const double angle = osg::DegreesToRadians(90.);
> const osg::Vec3d axis(1, 0, 0);
> mt->setMatrix(osg::Matrix::rotate(angle, axis));
> mt->addChild(model);
> root->addChild(mt);
> ..
>
> I can't upload the model because it's not for public use.
>
> I hope the informations are enough that some kind person can give me a hint.
>
> I worked to all coordinate frames. Only Projection seemd to work for me. 
> Never found the right matrices to make View or Model working. Precision Hint 
> can be float which has the same errors. Intersection Limit can be "NO_LIMIT" 
> and wouldn't change a thing.
>
> A LineSegmentIntersector never found a result.
>
> Cheers,
> NoxxKn
>
> --
> Read this topic online here:
> http://forum.openscenegraph.org/viewtopic.php?p=67319#67319
>
>
>
>
> Attachments:
> http://forum.openscenegraph.org//files/pick_error2_941.png
> http://forum.openscenegraph.org//files/pick_error_165.png
>
>
> ___
> osg-users mailing list
> osg-users@lists.openscenegraph.org
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org

[osg-users] PolytopeIntersector LIMIT_NEAREST (or LIMIT_ONE) different distances for different groupnodes

2016-06-01 Thread David Knipp 
Hi,

I'm very sorry that i have to create a new post. I'm currently facing a problem 
i can't solve on my own. 

Im working on a single Model which has a lot of group nodes and geometry nodes. 
This model is categorized in a few "main" groups. 

In the beginning i worked only with one of this main groups at once. 

The application does view this group and handle pickings on them. Everything 
worked nice and cleanly. After i was mixing some of the groups the results of 
the picker got strange. 

Groups got different distances and also the differences change if i just change 
the angle i'm looking at the model. (orbit manipulator! I don't change the 
distance!)

The strange behavior is that i start picking nodes that where BEHIND of the 
nodes i wanted to pick (in viewer).

I'm working with a picker like so:

int x = GET_X_LPARAM(lParam); // windows mouse pos
int y = GET_Y_LPARAM(lParam); // windows mouse pos

float _x = static_cast(x);
float _y = static_cast(y);

_x = 2.f * (_x - 0.f) / (window_width - 0.f) - 1.f;
_y = -(2.f * (_y - 0.f) / (window_height - 0.f) - 1.f);

double w = 0.005, h = 0.005;
osgUtil::PolytopeIntersector * picker = new 
osgUtil::PolytopeIntersector(osgUtil::Intersector::CoordinateFrame::PROJECTION, 
_x - w, _y - h, _x + w, _y + h);

picker->setIntersectionLimit(osgUtil::Intersector::LIMIT_NEAREST);
picker>setPrecisionHint(osgUtil::Intersector::PrecisionHint::USE_DOUBLE_CALCULATIONS);
picker->setDimensionMask(osgUtil::PolytopeIntersector::AllDims);

osgUtil::IntersectionVisitor iv(picker);
iv.setTraversalMode(osg::NodeVisitor::TRAVERSE_ALL_CHILDREN);

viewer->getCamera()->accept(iv);

if (picker->containsIntersections()) {
   const osg::NodePath& nodePath = 
   picker->getFirstIntersection().nodePath;
   unsigned int idx = nodePath.size();
   while (idx--) {
// bla bla specifc
}
}

What i mean is described in Attachments.

First Image "pick-error.png":

Red: Calculated nearer (but are in viewer BEHIND!)
Green: Calculated further away (Actually the node i wanted to pick)

Second Image "pick-error2.png":

I just change the angle of the orbit viewer for about 30 degrees and don't 
change the distance to object. The Distances just explode to much higher 
values. (Note the Model should only be about "2.f" heigh!)

I guess the second image error was allways present, but i didn't realised it, 
because the picker worked allways right.

Another side node, the model seems to doesn't stand in the right position. It 
should be in 0 0 0 but is located anywhere like 1 -11 -180 or something like 
that. Which is very strange but never made a problem. 
For loading i do: 

// ... osg::MatrixTransform mt
const double angle = osg::DegreesToRadians(90.);
const osg::Vec3d axis(1, 0, 0);
mt->setMatrix(osg::Matrix::rotate(angle, axis));
mt->addChild(model);
root->addChild(mt);
..

I can't upload the model because it's not for public use.

I hope the informations are enough that some kind person can give me a hint.

I worked to all coordinate frames. Only Projection seemd to work for me. Never 
found the right matrices to make View or Model working. Precision Hint can be 
float which has the same errors. Intersection Limit can be "NO_LIMIT" and 
wouldn't change a thing.

A LineSegmentIntersector never found a result.

Cheers,
NoxxKn

--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=67319#67319




Attachments: 
http://forum.openscenegraph.org//files/pick_error2_941.png
http://forum.openscenegraph.org//files/pick_error_165.png


___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org