> Cool, any chance of an auto scheme converter (replace TS normals with the equivalent RA2 indecies)? Also, are you going to work out the approx angles of all the other indecies in the RA2 scheme for use in 3DS to voxel conversion and/or a more advanced autonormals algorithm.

The problem with converting TS normals into RA2 normals is that we only used a subset of the TS normals.(26 out of 36)
The remaining 10 could cause havoc on your model when you convert it to RA2 style.

As for the complete spectre of the RA2: it is going to be a tedious task to define all 240 normals, on the assumption they are distributed on a 240 faced sphere...
Maybe Will or Vipr can find the algorithm that distributes the indexes on that sphere, by taking the main directions front/back/left/right/top/bottom. I'm pretty sure those are correct, because they can be easily identified on a Westwood voxel.

If you take my old TS diagram, that orientates a voxel when it is facing South-North ( http://www.eclipsemods.com/flyby/NORMALS/TS_normals_APC_green.png ), you'd get this :

front 106
back 7
left 153
right 147
top 232
bottom 16