Roll20's hex grids are "mathematically drawn", with neither axis being exact or remotely corresponding to the grid size, and yet the hexes are somehow also not regular hexagons, or even a regular tiling. They are thus impossible to cleanly align anything with in any way shape or form. The v-hexes and h-hexes are not even remotely similar dimensions either, for no reason I can discern. I cannot see this as anything but a serious bug, and cannot understand why it (apparently) remains completely unaddressed for most of a decade (particularly considering the geometry is not that complicated). You can zoom in and put a ruler against your screen, and directly measure how irregular the edges are. For example, the vertical edges of v-hexes are about a tenth longer than the diagonal ones, and this is very significant. This would, however, not be a problem if the widths and heights of the hexes were conversely distorted, but they are not. These edges, however, are but a symptom of what is likely the real issue: I am aware that the hex-grids are actually disguised offset square/rectangular grids, and these grids are spaced ... incorrectly . Using the centers of hexagons (and thus the centers of the corresponding hidden squares), adjacent rows of v-hexes need a vertical distance between their center-lines of exactly sqrt(3)/2 times the horizontal distance between the centers of two adjacent hexes in a single row. That horizontal width could easily be left the regular grid width of 70px exactly, such that a v-hex is displayed exactly as wide as a square grid, making alignment and scaling trivial. (H-hexes would, of course, just swap horizontal for vertical, rows for columns, widths for heights, etc., in all this.) The accuracy of the hexagons drawn around the centers could be left eternally incorrect and yet proper spacing of the centers alone would almost entirely fix the actual issue (... as long as the drawn hexes at least remained centered). Here is a helpful diagram for the v-hex case (the h-hex diagram is identical but on its side): (Note that this image was produced using 76/44 as a ratio approximation for the square root of 3. It's a pretty decent one, too.) This works equally well using the spacing of vertices belonging to the underlying grid instead of cell-centers as reference points, if the squares work that way (which is likely). Yes, the underlying "squares" need to deform into rectangles for the hex-grids, and tokens would clearly need to ignore the squashed axis for "unit size"; I trust this was never an issue and is the case currently. Every single hexagonal vertex (for a regular hexagon) is also just 2/3rds of the vertical separation away from the hex-center directly above or below it, making the hexagon edges easy to locate. If Roll20 has anyone trained in maintaining the grid code at all, please have them finally fix this, and ensure they are given the time needed to do so. If the hex-grid code is incomprehensible spaghetti, please just rewrite it with a sensible geometric framework in mind this time. Anything less could only be understood as a complete lack of pride in the quality of this application, which does not encourage one to spend time with it, let alone money on it. ... Or recommend it to others. (Every time I talk about Roll20 with people, now, all I can do it rant about how incomprehensibly bad the current hex grids are ...)