Hey, Tom, thanks for the comment. I'm actually quite happy with the elephant, because it does keep its value [exactly!] relative to the warmachine. Both pieces' moves have the same size and shape in that each moves 1 or 2 cells along 3 mutually perpendicular axes in this 3D game. They share 1 axis, and the elephant's move is rotated 45 degrees around this axis relative to the warmachine. Thus, there are 3 possible elephants in this game. I am only using 1, breaking symmetry; this is where the 'cheat' comes in. I am not cramming the board with pairs of all 3 elephants, so I have established a preferred direction, an asymmetric field. 

Since there is already a preferred direction created by setting up 2 opposing sides with pawns, using the elephant that lines up with this preferred direction and ditching the other 2 elephants keeps the game simple. Part of my philosophy is to let the board star in these sorts of games, not the pieces. The players should be able to enjoy the extra connectivity of the board without getting lost in a clutter of pieces or in the potential ramifications of the move of a tricky piece like the sissa or rose. [Anybody want to try to draw the 3D and 4D movement diagrams of the rose and the sissa? Anyone who is desperate for a headache?*] Anyway, with only 10 pieces and pawns per side, the starting piece density is some 18-19% [40/216], making it possibly playable by mere mortals. At least, that's my hope. 
*Actually, I know one or two people who prefer complex, difficult to visualize, effectively chaotic, pieces to nice, simple, boring, pedestrian pieces - I have no idea why.

Charles Gilman's Redistribution 3d Chess uses many of the same principles, although in different ways. The key idea is the same: use of short range pieces [right down to the 1 or 2 step pieces] to reduce chaos and maintain better control of the game. Both games use small numbers of pieces, keeping down complexity. These limits, on movement and range, were applied the same way. Other limits were applied differently, to much the same end. 3D GtS further limits the shortrange pieces move potential on a 6x6x6 board by eliminating most diagonals from piece moves, and laming the knight. R3d keeps and uses all the diagonals, so cuts down the movement potential of individual pieces by cutting down the size of the board to 4x4x6, less than half a 6x6x6. 

I see 2 noticeable differences in the games. One is that 3D GtS has a 'traditional' start, with pawns and a good amount of room to spread out in, before moving into contact with the enemy. R3d, with its lack of pawns [and thus promotion and 'terrain'] and strong board directionality [strictly toward the other player] starts 'in the middle' of the game. This, I see as a relatively minor difference. From a design standpoint, R3d merely jumps the game ahead a bit. 

The significant difference between the 2 games, I believe, is in the piece design. R3d has complex pieces, 3D GtS has simple pieces. Well, at least relative to each other, regardless what one may think of the absolute simpleness or complexity of the pieces. I believe it would be instructive to play both games side by side, to compare and contrast them. Who knows, we might find out something unexpected from this sort of comparison.