3D Technologies

We have released an open-world 3D environment covering the entire Shizuoka Prefecture, built through the advanced processing and refinement of VIRTUAL SHIZUOKA's massive point cloud data. Enjoy a flight over Shizuoka, recreated through high-quality real-time rendering. To keep the experience engaging, the in-simulator time progresses 24 times faster than reality, allowing you to witness the dynamic transition between day and night. 

Note: The flight speed is calculated based on real-time parameters.

"Point Cloud" format data is widely used when digitizing the shapes of terrain and buildings in real-world spaces. Point Cloud is a collection of points discretely placed within a 3D coordinate space, representing the surface shapes of the ground and objects. Sensors such as laser scanners and LiDAR are often used to measure real-world spaces, and the output information from these sensors is in point cloud format. By taking photos with an optical camera (ordinary camera) at the same time as the LiDAR, it is possible to assign colors to each point in the point cloud through subsequent alignment processing. Displaying this colored point cloud data in a 3D virtual space provides a high sense of presence. The data from VIRTUAL SHIZUOKA mentioned in the previous section also corresponds to colored point cloud data. However, there are challenges when using this virtual space for simulation purposes. For example, when attempting to run an autonomous robot, the ground and objects are represented by points, meaning the tires of the robot's wheels come into contact with "points" rather than a "surface," making it difficult to replicate realistic physical behavior. Furthermore, point clouds have large data volumes and tend to result in high processing loads. For instance, the application displaying the 3D virtual space becomes heavy. Therefore, we are conducting research and development on algorithms to convert point cloud data into lightweight polygon (polygonal) data. These algorithms are also utilized for processing open data such as VIRTUAL SHIZUOKA. Furthermore, during data conversion, we are not only performing simple replacement processes but also challenging ourselves to develop technologies for noise reduction and for estimating and interpolating missing areas.

While our basic policy is the automatic generation of 3DCG, there are cases where manual 3DCG production is required depending on the application or conditions. This applies to situations where the precision of open data is insufficient, or where actual measurements cannot be taken because equipment installation is still in the planning stage. When creating 3DCG for outdoor spaces, we utilize actual measurement results from open data and other sources. Depending on the purpose of the simulation or visualization, the accuracy of that data may be inadequate, or it may be impossible to take measurements in the first place if equipment is about to be installed on-site. In such cases, we may produce the 3DCG manually. For example, in the research and development of smart agriculture, we produce 3DCG of farms to conduct simulations.