<p>Many studies have used images of novel objects as experimental materials. Existing novel object databases do not provide diverse exemplars, and many studies need to manipulate or examine the diversity of exemplars. To fill this gap in experimental materials, the present study introduces the Novel Prototype and Exemplar (NPE) database. This database contains 108 prototypes and 2592 exemplars of viewpoint and shape variants. The present study conducted four experiments to standardize the database and validate nine dimensions of the novel objects in the NPE database. Experiment 1 standardized familiarity, visual complexity, naming difficulty, reality, and comfort. The results revealed that the prototypes in the NPE database performed well in these attributes. Experiment 2 used the same methodology and reported that the prototypes have high absolute novelty, high relative novelty, low name agreement, and low identity agreement. Experiment 3 used multidimensional scaling (MDS) and Monte Carlo simulations and revealed that the similarity between prototypes was low. Experiment 4 generated exemplars by varying viewpoints and shapes and used the same methodology to evaluate the similarity between prototypes and exemplars and between different exemplars. Finally, Experiment 4 used these evaluation results to select appropriate typical and atypical exemplars. In conclusion, the NPE database contains the largest number of novel 3D object images, controls for the largest number of subjective dimensions, and is the first to develop abundant exemplars. As an open-source novel object database, the NPE database will greatly contribute to research in psychology, linguistics, cognitive science, ergonomics, artificial intelligence (AI), etc.</p>