This article investigates how images are understood inside and outside heritage institutions. It focuses on information specialists in libraries, archives, and museums and on a very specific yet substantial end-user group for visual heritage material: university scholars in the humanities. Based on a survey on the production and use of descriptive metadata, this study discloses that there is an ontological divide between these two groups, and that the extensive production of descriptive metadata does not match the needs and interest of researchers in the humanities, but rather other end users. An increased dialogue is needed between these two groups concerning what metadata should be attached to images. This potentially could lead to a broader and more extended scholarly use of visual heritage material.