For sense of depth, you need multiple views of the same feature. Accomplishing this with one retina is absolutely possible, you'd just need a way to tell the view A of feature A apart from view B of feature A ( ... and from other views of features B-Z, but this problem also exists with two separate eyes).
Temporally this might be easiest - just one iris is open at any moment, but the frequency of switching between the (however many) irises is high enough to still work for depth on objects in motion. (Or the iris(es) move, or it is one iris-band that is undulating, ...)
Chromatically - different irises have different color-filters, easy, but might also be easy for prey to out-evolve (dazzle patterns in colors matching the filters)
By shape of bokeh might be fancy, but also well suited to neural processing: Every iris has a different shape (circle, triangle,...) - so the bokeh of every feature is coded, thereby binding feature to iris
If the irises are so far apart that the images do not even overlap, that would also work.
To direct your gaze, you would need to move your long-eye/head, or maybe just the irises: Imagine a cylinder of retina (in the extreme case, like an all-eye stalk), interspersed with chromatophore-like iris-lens-assemblies (ILA), that can be opened and closed. That way you would only have to open the irises pointing in the right direction, and voila! - There would be more blindspots (in the extreme case, as there would be no retina on the ILAs) per impromtu 'eye', but that could be ameliorated by having more than two 'eyes' facing in any given direction.
Compound eyes of course would also work to give your character that Geordi-LaForge-visor-look, but frankly, it's cheating - with compound eyes any distribution on your body is fair game.