Hello everyone and thank you very much for the introduction.Now I‘d like to presents the results of two master students, [CLICK]
Themainmotivationforourwork was to examinethe power of todays mobile devicesusingtheexample of an 3D virtualenvironment.We‘vedonethisusing a simple example of a high-ordervisualizationapproach: thevisualization of 3D points of interests.Thismeansnotonlyrendering a specificscene.
This talk mainly deal with two approaches for visualizing points-of-interests which can be located on- and offscreen.One approach is called 3D halo circle and the other 3D halo projection depicted here.We combined both approaches with a simple form of active occlusion management.[CLICK]
Before I proceed I‘d like to give a brief overview of this talk.At first, I present the problems and challenges to visualize points-of-interest in 3D geovirtual environments on mobile devices.Then I present two approaches that can accomplish this task in combination with a simple form of occlusion management.After that, I briefly describe our implementation before I discuss our approaches and conclude this talk.[CLICK]
The resulting images of this rendering can look as depicted here:Screenshot A shows the visualization of 3 points-of-interest which are located outside the current view frustum.Screenshot B shows this visualization of a point-of-interest that is on the screen and a point-of-interest that is located in the left half space behind the camera.[CLICK]
Again, I‘dliketoreferetothepaperformoredetails on that.[CLICK]
Thisvideodemonstratethedirectapplicationofthecirclevisualizationto a virtual 3D city model.Youcanrecognizetwomajorissuesatthispoint:1.2.[CLICK]
Zusammenspiel von occlusionmanagementandpoivisualization
After the main concepts are explained, I‘d like to drop some lines on the implementation.As target platform we choose Apples iPhone and iPod touch because they were available at this time.It provides a dedicated GPU with around 11 MB video memory out of 128 MB main memory.For software development Apples SDK was used, to develop a rudimentary scene graph system which is based on OpenGL 1.1.For optimizing the rendering speed, the scene graph API provides per-object view-frustum culling.[CLICK]
Let me give a brief overview of the rendering processthta is displayed here in pseudo code.The following code is executed on a per-frame basis.For each object to render, we first distinguish between visible and invisible objects.The, given all POI; occlusion management is applied to object.[CLICK]If the object is categorized as occluder, the rendering mode is changed (here: to wireframe)In the case that a object is not visible but classified as POI, [CLICK]We select the characteristic sigma and delta function and determine the ARP.Finally, we render the proxy object (projected or circle halo).
Totestourprototypicalimplementationweusetwo different synthetictestdatasets.
Thank you very much for your attention.I‘d like to welcome any questions you may have on this topic.