Home > Virtual Reconstruction Model > Level of Details
Virtual Reconstruction Model
Level of Details

The interactivity of real-time simulations presents a unique challenge to the computer modeler. With computer models built expressly for animation clips, it is only necessary to include as much detail as will be seen from a pre-determined path at a pre-determined viewing distance. If the rear fa?ade of a building is not going to be seen in the animation, there is no reason to include it in the model. In contrast, real-time simulation models must include every possible detail because the user is given the freedom to choose his or her own path through the environment, driven only by their curiosity. In the case of the Herodian Temple Mount model, one user may choose to fly overhead, circling the platform. Another may choose to enter the Royal Stoa and examine the moulding of the Attic column base. As a result, every building element must be modeled to as high a degree of detail as possible, always mindful of the restrictions imposed by the rendering requirements of real-time technology.

The solution is to build a series of nested models of the same object, each level more complex than the next. Each version of the object is then known as a level of detail. During the simulation, the user's distance from the object determines which level of detail is displayed. When seen from far away, only the simplest level of detail is displayed. When seen close up, the most complex level of detail is displayed.

The columns in the Royal Stoa are a good example of this technique. Five different levels of detail are built into each of the column. From a distance over 200 meters, the user sees only the lowest level of detail - 6 simple triangles. From a vantage point five meters away from the column, the user sees the highest level of detail - 828 total triangles. This approach is very data efficient. Consider the effect when the user is standing inside the western entry to the Royal Stoa: a maximum of six columns are being rendered at their highest level of detail (828 triangles), a dozen are being rendered at the middle level of detail (340 triangles), and the rest are rendered at the second-lowest level of detail (82 triangles). For this configuration, the computer has to render 20,856 triangles every 33 milliseconds to display the columns. If the columns were built with only one level of detail (the highest), the computer would have to render an additional 113,280 triangles for the same scene.

Back to top