Erra, UgoCapece, Nicola FeliceAgatiello, RobertoAdrien Peytavie and Carles Bosch2017-04-222017-04-2220171017-4656https://doi.org/10.2312/egsh.20171003https://diglib.eg.org:443/handle/10.2312/egsh20171003We present a feed-forward neural network approach for ambient occlusion baking in real-time rendering. The idea is based on implementing a multi-layer perceptron that allows a general encoding via regression and an efficient decoding via a simple GPU fragment shader. The non-linear nature of multi-layer perceptrons makes them suitable and effective for capturing nonlinearities described by ambient occlusion values. A multi-layer perceptron is also random-accessible, has a compact size, and can be evaluated efficiently on the GPU. We illustrate our approach of screen-space ambient occlusion based on neural network including its quality, size, and run-time speed.I.3.3 [Picture/Image Generation]Display AlgorithmsI.3.7 [Three Dimensional Graphics and Realism]Colorshadingshadowingand texture10.2312/egsh.2017100313-16