Despite decades of study of memory, it remains unclear what makes an image memorable. There is considerable debate surrounding the underlying determinants of memory, including the roles of semantic (e.g., animacy, utility) and visual features (e.g., brightness) as well as whether the most prototypical or most atypical items are best remembered. Prior studies have also relied on constrained stimulus sets, preventing a generalized view of the features that may contribute to memory. Here, we collected over one million memory ratings (N=13,946) for THINGS (Hebart et al., 2019), a naturalistic dataset of 26,107 object images designed to comprehensively sample concrete objects. We uncover a model of object features that is significantly able to predict image memorability, covering over half of the explainable variance. Within this model, we find that semantic features have a stronger influence than visual features on what people will remember. Finally, we examined whether memorability could be accounted for fully by the atypicality of the objects, by comparing three complementary measures using human behavioral data, object feature dimensions, and deep neural network features. We discover, surprisingly, that the relationship between memorability and typicality is more complex than a simple positive or negative association, however, generally, prototypical objects are the most memorable. Taken together, our findings reveal important structural features underlying the organization of information in memory.