Advances in seismic interpretation using new volume visualization techniques

International audienceAs the use of 3D seismic interpretation continues to become part of the main stream work process with the industry, visualization techniques also continue to evolve as software and hardware improves. In the past 10 years, volume rendering tools have been progressively adopted by the geophysical community as the emergence of high-end graphics workstations with 3D texture capabilities made real-time volume rendering possible. Many interactive volume rendering packages are now available for seismic interpretation. However, interpretation is still mostly done in 2D. Using classical volume rendering with high spatial frequencies of seismic data make it very difficult to produce meaningful volume images and often results in cluttered useless images. Volume rendering tools are now part of most seismic interpretation packages. However, most do not provide better insight into the 3D structures of seismic data because the noise and high spatial frequencies in the data prevent classical volume rendering from capturing relevant information. As an improvement, we propose more suitable high quality volume rendering algorithms based on a pre-integration of the transfer function (i.e. colour map) that use the capabilities of the recent programmable graphics processing units (GPUs) of new graphics cards. This results in a versatile multimodal (multi attributes) volume rendering system. This system is dedicated to the efficient combined visualization of several volumes. Coupled with high quality volume rendering algorithms, it makes it possible to visualize isosurfaces interactively and paint them with another attribute. This is done without explicit extraction of the surface. By this way, isosurfaces of distance maps to faults or well paths can be interactively extracted and painted with the seismic data. In this article, we adapt high quality volume rendering algorithms from the computer graphics industry to improve the imaging. We have found these algorithms more suitable for seismic data analysis than classical ones. They use the capabilities of the recent programmable graphics hardware. In addition, we will present a versatile multimodal volume rendering system that enables the efficient co-visualization of several volumes