I am glad you understand, and I would be happy to explain what I think I know to be true.
I think we agree that MAME is versatile enough to accommodate analog and digital devices to be mapped to any kind of game input, with varying degrees of successful control based on how well the two match up. However, in order for a 2-axis analog device to be mapped to a digital input for a game, I surmise that its field must be discretized. From using HeadKaze's joystick mapping utility, I gather that the field of a 2-axis device is split up into a 9x9 grid, or 81 distinct combinations of voltage values for the field defined by the two axes. Without a customized mapping, the first 3x3 group (NW) would be devoted to NW movement, the next 3x3 group to due north, and so on, with the middle center group defined as a deadzone. The logic of course being that if the voltage values of the input device fall within the bounds of one of those 9 blocks in one of the 3x3 groups, only the command associated with that group will be delivered to the game. With a customized mapping, however, some or all of those 9 blocks in a 3x3 group can be assigned to a different direction. So in the case that I present, only the very first block of the first 3x3 group (top leftmost) would be assigned to NW, 3 blocks to N, 3 blocks to W, and the remaining two blocks down the diagonal of that group to "sticky". [1+3+3+2=9] The values that you see in the ini line I posted is a row-wise serialized encoding of the matrixed sensitivity map for that 9x9 grid. Each of the different numbers is associated with one of the 8 directions. An s indicates a "sticky" value, which I suppose has some logic tied to it for four-way games.
Edited by Tron1001 (04/05/13 01:33 AM)