We begin with the evaluation of rendering results. Figure 4(a) shows the result of rendering a virtual view using only the first layer of the LDI and without hole filling. Figure 4(b) shows the result of rendering the same scene but this time applying a simple hole filling technique, namely, horizontal depth-based extrapolation, which can be implemented in real-time. More advanced hole filling techniques can achieve better results, but they cannot recover the girl’s face—when we rely only on one 2D+Z data,crucial data may be missing. Figure 4(c) shows result rendering the same scene but now using the proposed LDI-based graphics architecture. Note that the girl’s face can be seen in the background.