Deep Volumetric Ambient Occlusion

We present a novel deep learning based technique for volumetric ambient occlusion in the context of direct volume rendering. Our proposed Deep Volumetric Ambient Occlusion (DVAO) approach can predict per-voxel data sets, while considering global information provided through transfer function. The neural network only needs to be executed upon change this information, and thus supports real-time interaction. Accordingly, we demonstrate DVAO's ability occlusion, such that it applied interactively within To achieve best possible results, propose analyze variety function representations injection strategies networks. Based on obtained results also give recommendations applicable similar scenarios. Lastly, show DVAO generalizes modalities, despite being trained computed tomography only.