A sphere doesn't seem that bad, and would naturally provide a set of terraces with good views / access to the central open space. The outermost section could be scaffolded to provide a big flat band near the equator (with lots of space beneath for offices, storage, industry, or whatever. Having different gravity levels might be an advantage. The design is one of the three top designs, and is called a Bernal Sphere.
Something I recently read: A problem with rotating cylinders is that they want to convert the rotation from along the long axis to along a short axis. I'm sure there is a statistical / kinematic reason why that happens. The tendency means that active spin management is needed, independent of hub friction problems.
Does "convert the rotation from along the long axis to along a short axis" mean that the cylinder would tend to wobble over time? (without active intervention)
I don't know the exact physics, but I wonder if, with a sphere, there might be problems with different bands of air rotation being in contact with one another. The air rotating at the equator of the sphere would be moving faster than the air rotating along north and south of the equator.
That happens on Earth too, and makes for a lot of entertaining weather patterns you probably wouldn't want to duplicate on a space station, even on a smaller scale.
I think at the scale these projects were imagined for that's very likely. On the one hand, there could be psychological benefits to changing weather and benefits to creating a rain system to support your agriculture. Then again, thunder and lightning are probably things you don't want on a space station.