It has become common in foundational discussions to say that we have a variety of possible interpretations of quantum mechanics available to us and therefore we are faced with a problem of underdetermination. In ref [1] Wallace argues that this is not so, because several popular approaches to the measurement problem can't be fully extended to relativistic quantum mechanics and quantum field theory (QFT), and thus they can't reproduce many of the empirical phenomena which are correctly predicted by QFT. Wallace thus contends that as things currently stand, only the unitary-only approaches can reproduce all the predictions of quantum mechanics, so at present only the unitary-only approaches are acceptable as solutions to the measurement problem.Wallace's arguments about the difficulties of extending approaches which are not unitary-only to QFT are quite compelling, but on the other hand the Everett interpretation and the other extant unitary-only interpretations have a number of serious epistemic problems which arguably have not yet been satisfactorily resolved (see refs [2,3]), and thus we are faced with a dilemma: if neither of these obstacles can be overcome then it would seem that at present we have no viable solution to the measurement problem at all! Or at least, none of the well-studied, mainstream interpretations or modificatory strategies will suffice -and we suspect that many less well-known proposals would also be difficult to extend to non-relativistic quantum mechanics, or would face the same epistemic problems as the unitary-only interpretations. We therefore consider that it remains an urgent outstanding problem to find a viable solution to the measurement problem which can be extended to relativistic quantum mechanics and QFT.In this article we seek to understand in general terms what such a thing might look like. We argue that in order to avoid serious epistemic problems, the solution must be a single-world realist approach. We also argue that any singleworld realist solution which is able to reproduce the predictions of relativistic quantum mechanics will probably have the property that observable reality does not supervene on dynamical, precisely-defined microscopic beables. Thus we suggest three possible routes for further exploration: observable reality could be approximate and emergent, as in relational quantum mechanics (RQM) with the addition of cross-perspective links, or observable reality could supervene on beables which are not microscopically defined, as in the consistent histories approach, or observable reality could supervene on beables which are not dynamical, as in Kent's solution to the Lorentzian classical reality problem.We conclude that once all of these issues are taken into account, the options for a viable interpretation or modificatory strategy for quantum mechanics are significantly narrowed down. In light of this fact we have renewed optimism that it might eventually be possible to arrive at a definitive solution to the measurement problem, although the remaining options ...