Deep Rendering Graphics Pipeline

The graphics rendering pipeline is key to generating realistic images, and a vital process of computational design, modeling, games, animation. Perhaps the largest limiting factor time; processing required for each pixel inevitably slows down produces bottleneck which limits speed potential pipeline. We applied deep generative networks complex problem an animated 3D scene. Novel datasets annotated image blocks were used train existing attentional adversarial network output renders environment. Caltech-UCSD Birds-200-2011 dataset served as baseline comparison loss quality. While our work does not yet generate production quality renders, we show how method using machine learning architectures novel text has produce functioning framework