Improving terrain rendering using Hardware Tessellation for Ordered Point Clouds (OPCs)

Improving terrain rendering using Hardware Tessellation for Ordered Point Clouds (OPCs)

Time seems to run faster in the field of computer graphics and visualization because of rapid development of new hardware and algorithms. Therefore the advantages of such algorithms have to be presented attractively to industrial partners to keep their interest in innovation and improvements. VRVis created a prototype to visualize Ordered Point Clouds (OPCs), which are used for terrain rendering, more efficiently by the use of the state of the art approach Hardware Tessellation. It showed that the effort of improving already optimized methods with completely new algorithms can create a future-proof base for further projects.

Challenges in the fast paced world of computer graphics

In computer science many developments seem to be out-of-date straight after they hit the market. On the other hand many industries use applications and techniques for a surprisingly long time. Especially internally wide-spread tools are often difficult to change because teaching the use of new ones seems more expensive than keeping older and slower applications.

Therefore presentation of state-of-the art approaches is really important. It is obvious that such solutions have to be developed and created with content known by specialists who will have to work with these approaches.

Especially in visualization many questions, e.g. if the change is really needed because the current system still works fine, are often asked. Only proof of concept can explain the real advantages of new technologies because solutions are highly specialized.

Current laser scanners create a huge amount of data for each virtually reconstructed environment. VRVis already created some highly advanced methods to explore such datasets in real-time. Because technical progress has since advanced, new methods have to be prepared to be presented for the specific needs of industrial partners.

Hardware Tessellation for more efficient use of graphics resources

After some research we discovered that Hardware Tessellation would meet our needs for more efficient visualization of OPC datasets. It is a method which creates more details directly in the computer’s graphics hardware which leads to a much more efficient usage of resources. Especially memory is always a big problem with high tessellation laser-scan datasets. Level of detail solutions, which only show information needed at the moment, are essential. There are still problems with older approaches of these techniques like over-tessellation close to the viewer’s camera or noticeable changes between the fixed levels. For Hardware Tessellation the amount of vertices and therefore the use of memory is highly reduced. Displacement maps are used to keep all information of the dataset highly memory-efficient. With the combination of this data a high tessellated rendering is possible. The closer the user is to the data he is exploring, the more information is shown by the graphics hardware. Figure 1 shows the much higher tessellation closer to the viewer and how the implemented, seamless Level of Detail approach works by lowering details farther away from the camera.

Fig. 1: Hardware Tessellation with seamless level of Detail

Impact and effects

The use of Hardware Tessellation proves that it can highly increase the exploration efficiency of OPC datasets. The industry partners of VRVis are easier to convince by the advantages of such approaches and that the investment of time and money will lead to a better base for further development. The technique is planned to be used for the visualization of terrain or tunnel surfaces, which both already exist as OPCs created by Joanneum Research.

Fig. 2: Rendering of terrain created by a laser scan

Finally the decision, if and how to use this work, is made by industrial partners. The improvements in graphical attractiveness and memory efficiency cannot be denied and technological progress must not be ignored if they try to stay competitive in the fast-paced field of real-time visualization.