Background. Particle systems render groups of points in space that are used to add various effects to a scene. When rendering particles, there are various advantages and disadvantages for rendering an effect in either object-space or image-space. The former uses calculations in a 3D environment, while the latter adds particles to a 2D image. There is a lack of existing research to understand the trade-off in performance and in effort when applying these two techniques. Objectives. This thesis aims to compare the performance difference between particle systems in object-space versus image-space. The intent is to do this using a C++ DirectX11 implementation. Methods. The primary methodology of this thesis was an implementation complemented by a performance evaluation. Two particle systems were made to render the same particle effect. The systems were run at five different particle amounts and the FPS was recorded for 30 seconds while the simulations ran. Results. The data presented displays the average FPS recorded, with different line graphs for the various particle amounts. The results show an overall better performance of the O-system. Conclusions. While the O-system performed better, the various techniques and optimizations available to particle systems that render in image-space would allow avastly different result. This thesis would be well complemented with further research. It would also be furthered by comparing performance while additional geometry interacts with the particles.

The digitalization leads to that many physical solutions are replaced by digital once. Especially in the surveillance and security business have humans been replaced by cameras which are monitored from a remote location. As the power of the computers has increased, live video can be analyzed to inform the controller about anomalies which a human eye could have missed. SAAB – Air Traffic Management has a digital solution to provide Air Traffic Service which is called Remote Tower. This project will come up with a recommendation on how SAAB can dynamically render overlay based on live video, which marks the runways and taxiways on the airfield.

Volume rendering is quite an old concept of representing images, dating back to the 1980's. It is very useful in the medical field for visualizing the results of a computer tomography (CT) and magnet resonance tomography (MRT) in 3D. Apart from these two major applications for volume rendering, there aren’t many other fields of usage accept from tech demos. Volumetric data does not have any limitations to the shape of an object that ordinary meshes can have. A popular way of representing volume data is through an algorithm that is called volume raycasting. There is a big disadvantage with this algorithm, namely that it is computationally heavy for the hardware. However, there have been vast improvements of the graphic cards (GPUs) in recent years and with the first GPU implementation of volume raycasting in 2003, how does this algorithm perform on modern hardware? Can the performance of the algorithm be improved with the introduction of GPGPU (DirectCompute) in Directx 11? The performance results of the basic version and the DirectCompute version was compared in this thesis and revealed significant improvement in performance. Speedup was indeed possible when using DirectCompute to optimize volume raycasting. / Implementation, optimering och prestandamätning av en volume rendering algoritm som heter volume raycasting. Optimeringen är utförd med hjälp av DirectCompute i Directx 11.