I think it depends on how it is licenced. If it is under a creative commons licence, where people can use/reuse/extract data it may be useful to generate morphometric comparative data sets. If it is just "look and see but dont use to generate new data" type of copyright, then I wouldnt see the use beyond classroom teaching (although it would be useful for that).
While I definitely agree about the freely available for all uses. I think a database of brain images could be really helpful to anyone who works on species that don't have neuroanatomical atlases available. The ability to examine multiple images (not just the ones in the published manuscripts) could help in the determination of regional homologies in your species of interest. I think this could (in theory) improve the quality of the research and help everyone address specific questions of interest. I personally feel that the lack of neuroanatomical detail can really hinder the ability of future studies to make meaningful comparisons and the ability of everyone to make strides towards understanding the functionality or evolution of specific regions. I think that a database of brain images could help eleviate some of the difficulties with neuroanatomy on non-traditional species.
So in conclusion, Yes, I would find a database of brain images useful. I would also be willing to be a significant contributor to this database. In addition, maybe there could be a location on this database where people could upload pictures of brain regions that they are having difficulty identifying. Although my neuroanatomical knowledge is limited to one species of fish, I would be happy to help people try and identify regions as well.