On 19. Apr 2012, at 9:56, stahn1 [via Software] wrote: > So, if the observer is inside of the wall, then the opening angle of the > frustum reaches 180°, and thus the size of the displayed objects all become > zero?
Yes, although you'll reach NaN if it's exactly in the wall. > >> and if he goes even nearer, the object reappears and from here on grows in > >> size the closer he goes. > >Now he is behind the wall, which can't happen in reality (he would be > >looking back at the wall). > Well, it happens in our CAVE, so we have a misconfiguration somewhere. The > distance of the point in the middle of the CAVE to every wall is actually > more than one meter. The object vanishes when we are one meter away from the > point in the middle. Are there any values in a config-file which we could > change so that the virtual point inside of the wall is the same as the real > one? The wall corners are in meters and have to match your physical walls, i.e., in your case they should be bigger than one meter. > >The near clipping planes always have to be in front of the eye. > Now that you mention it... so I can't just change a fixed value, the near > planes have to adapt to the observers position. ...and the scene observed. Naturally you will get near clipping in a CAVE, when you stick your head into an object. > >The distance of the wall is V(eye->wall center), so you can set the near in > front of the wall using Channel::setNearFar or Frustum::adjust_near, see > Figures on http://www.equalizergraphics.com/documents/design/immersive.html. > Hm, now I am a little confused. From the said figures it looks like the near > plane is already adjusted automatically in respect to the observers position. It's always relative (in front of) the eye, by the OpenGL glFrustum definition. Eq does not adjust it, but eqPly does. > Or to ask in a different manner: What do we have to do if we want to > integrate immersive visualization (with the help of head tracking) into our > existing framework of our CAVE? The only thing we altered so far is that we > set the observers headMatrix (with the matrix of a tracking device) at the > beginning of the Config::startFrame-method. Does the setHeadMatrix-method > already take care of everything(e.g. adjusting the near planes), or do we > have to add some additional stuff? The near-far plane discussion is somewhat independent of the head tracking. GL has a limited depth precision (24 or 32bit), that's why you want to set your near/far planes to fit the model as tightly as possible. One can also use a fixed near/far, e.g., 0.1...1000, in which case you can't see anything closer than 10cm or further than 1km. The head matrix will influence the left,right,top,bottom of the frustum and the head transformation. It will not modify your near/far values - Note that your frustum is moving, and therefore the eye-relative near/far is moving around in the scene. HTH, Stefan. -- View this message in context: http://software.1713.n2.nabble.com/Head-Tracking-in-CAVE-tp7470111p7479760.html Sent from the Equalizer - Parallel Rendering mailing list archive at Nabble.com. _______________________________________________ eq-dev mailing list [email protected] http://www.equalizergraphics.com/cgi-bin/mailman/listinfo/eq-dev http://www.equalizergraphics.com
