What do you think the fix is for this? When people understand that the token's cell position is determined by the top left corner the current behaviour kind of makes sense... but it feels and looks off.

Are you going to change the cell position to be cursor based? Or center of token when dragging?

Are we too far down the rabbit hole to get an option to set a token's cell position as center?

Also (and this is separate issue but should be in sight when looking at this behaviour), when "Hide Mouse Pointer while dragging" is disabled, the cursor is visible during token layer drags, but not on any other layers.

What do you think the fix is for this? When people understand that the token's cell position is determined by the top left corner the current behaviour kind of makes sense... but it feels and looks off.

Are you going to change the cell position to be cursor based? Or center of token when dragging?

Are we too far down the rabbit hole to get an option to set a token's cell position as center?

Also (and this is separate issue but should be in sight when looking at this behaviour), when "Hide Mouse Pointer while dragging" is disabled, the cursor is visible during token layer drags, but not on any other layers.

The only real problem here is that the token image isn't rendered in the right place. If it were a half cell up-and-left from where it currently is, it would look and feel quite natural. The fact that it's offset just messes with expectations of where the token should end up. If we fix that, the token will naturally snap according to its center, which is the same as saying that it will snap to the cell that the token is mostly within (for Medium tokens anyways).

As for options / configurations for this behaviour, we are actually well set up to support that. Internally we already query the token for its "snap point", which today accounts for the token size, grid properties, layer, etc. Details for a configurable alternative would of course depend on the specific FR, but seems feasible.