[osg-users] Intersections with multiple Cameras and Views

[Danyel Kemali]

Hi,

I'm currently trying to figure out something and hopefully one of you can help me to get the issue here. So I have  two Views which I'm adding to a compviewer. My scene consists of several geodes and also includes a volume, we are working with a custom volume rendering here at my workplace. I also have four additional cameras which are running with the second cameras ViewMatrix as ray starting and ending cameras for the volume rendering. These cameras are updated continously with the ViewMatrix of the second views camera, as it is intended to render the volume only on the second view. Therefore, i cull the volume group on the first camera, respectively on the first view. Additionally, i'm doing some intersection. 

My Problem is, that my intersector is getting things messed up. Im using 


Code:
if (mFirstView->computeIntersections(screenPosX,screenPosY,intersections,mask))



to compute the intersections but it seems like the second cameras ViewMatrix is mixed up with the first camera's ViewMatrix somehow resulting in false positioned intersection results. The screen positions I'm picking are correct and triggered only on my first view. I have already figured out that everything runs fine when I'm updating the four cameras for the ray starting and ending positions (left and right eye), which are used for the volume rendering, with the first views camera's matrix. 

Therefore, i could just update the matrices just before the intersection part is done, but i would really like to know why my intersector is using matrices from cameras which I am not using for intersections computation. I might be missing here some of the basics on that topic.

I also attached a snapshot of my SceneGraph's first part. As you can see, six cameras are included there, four of them are used as the ray starting and ending cameras and two cameras are the camera's of view 1 and view 2. I was already wondering if the IntersectionVisitor might be the culprit here. I'm using: 


Code:
osgUtil::IntersectionVisitor intersectionVisitor(pLineSegmentIntersector);
mFirstCamera->accept(intersectionVisitor);



to set the Visitor for the the camera of the first view. Is it possible that my four cameras for the volume are  also accepting the same IntersectionVisitor? I'm attaching them to the MainTransformationNode as children. I have also tried to specify the camera for the Intersections with:


Code:
if (mFirstView->computeIntersections(mFirstCamera),cf,screenPosX,screenPosY,intersections,DIRECT_VOLUME_RENDERING_MASK))




but the result remains the same. Hopefully someone got my issue here and can help me out with some ideas on that.

... 

Thank you!

Cheers,
Danyel

------------------
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=67409#67409




Attachments: 
http://forum.openscenegraph.org//files/scenegraph_119.png