Figure 1: Common labeling as used in many AR browsers (Left) compared to our image-based approach (Right). Position of the labels can be automatically optimized, but also their appearance, including depth cue for the labels' anchor or their leader lines.