Organizations involved: SINTEF (Norway)
IQmulusViz is a desktop application targeting the marine showcases involving large seafloor data sets in IQmulus. The outstanding feature of this visualization client is the ability to interactively render LR B-splines which are used as efficient representation of the large seafloor data.
The term "large data" can refer to volume or complexity, for the case of marine scenarios using LR B-splines we are addressing both. The number of data items, surfaces and point clouds, can be very large and the viewer is mainly used for quality control and to analyse the configuration of the data items. This implies that there is a need for keeping track of individual entities and select, deselect and modify them individually.
The viewer supports a user initiated offloading of the most distant geometries, which removes the corresponding data from the CPU RAM & GPU RAM. When navigating the model the geometries are then dynamically offloaded and loaded according to their location in the scene. This can be useful when handling larger models where the amount of data can result in a low framerate and thus a less interactive experience.
FIGURE: IQMULUSVIZ VISUALIZATION OPTIONS: BOUNDARIES OF POLYNOMIAL PATCHES FROM WHICH THE SURFACE IS DEFINED (LEF). DATA POINTS COLOURED ACCORDING TO THEIR DISTANCE WITH RESPECT TO THE SURFACE (RIGHT).
The sizes of the LR B-spline surfaces are small compared to the distance fields. Typically, a 1.8 GB point cloud is represented by a collection of surfaces of size 4 MB to 13 MB depending on the requested accuracy. Thus, the main visualization effort is dedicated to showing point clouds coloured with respect to the distance to the corresponding surface. The size of the point clouds influence the memory consumption, the time spent in reading the input data and the frame rate. In most situations, points that are less accurately represented by the surface are most interesting to study. To gain performance, IQmulusViz offers the option to store and show only the points that are further away from the surface set than a specified distance. This distance is set in the menu using a slider and it must be set prior to reading the point set. The Figure above shows some of the visualization options of IQmulusViz. In the left picture of the figure, the list of surfaces belonging to the workflow result reveals that only one surface is created. If we chose to render the boundaries of the polynomial patches from which this surface is defined, we can see that the degrees of freedom vary between the different areas of the surface. In the right picture, the data points given as input to the workflow, are shown. The combined data set has a very non-uniform pattern. It is a clear correspondence between dense data points and many degrees of freedom in the surface, but also areas with large height variation between the points will give rise to small polynomial patches. In the right picture, the data points are coloured according to their distance and configuration with respect to the surface. Blue points lie above the surface, red points below and green points lie within a distance to the surface selected using a slider in the GUI.