Gaze-guided improvement of Volume Rendering in Large Virtual Reality Systems

 

The visualization of volumetric data models is usually addressed through the use of a GPU-based ray-casting scheme.
For large datasets, real time framerates are usually not achieved. This is especially true in the case of Virtual Reality
Systems, where two renderings (one for each eye) is required.
The purpose of this project is to focus on the user's gaze in order to deliver higher quality in the direction of sight
and eventually improve framerates by reducing the details in regions not directly pointed by the user's attention.
Although similar systems exist for surface-based approaches, mainly based on multiresolution models, volumetric
models are not easily decomposed into parts or objects (usually a single model is visualized) and therefore, adaptive
multiresolution is difficult to achieve.