Understanding the Grid Positions

[mdfeinstein]

When I set up a set of geometries with different mediums in a simulation and then plot the epsilon values to confirm the geometry is set up correctly, there are often extra lines (I use the contour plot for clarity). If I increase the resolution of the plot, the lines go away and my simulation looks as expected. I assume this is because the geometries are not quite lining up with the grid lines, and gaps or overlaps are occurring. This is a problem I am familiar with from Lumerical FDTD, but in that program, I could see the computational grid lines and understand how to set the geometries to match.
So my question is: where are the grid positions in meep?
For instance, if I have a resolution of 50 nm and a structure that is 50 nm tall, should I center that structure at the 0, so that it's edges are at +/- 25 nm? or should I center it at 25 nm, so that its edges are at 0 and 50 nm?