OK, so this is a revised 16x16 setup:

I moved the Warriors up to the 2nd rank, so they make a pre-fabricated 'pawn shield' for the King. The backward Knight captures of the Warriors are then quite useful: they protect the part of the back rank that is not shielded from attacks by the flying pieces. When an Eagle or Falcon (flying Q or R) could capture an unprotected piece there, this would deliver checkmate! So it is important that the entire back rank remains sufficiently protected against aerial intrusion.

The Warriors also provide solid protection of the pieces in front of them, on 3rd rank. They start heavily protected themselves, by putting a Prince in front of them, a Dragon King behind, and a Dragon Horse besides them, while also the Knight, Nightrider, Elephant, Rook, Frog, Archer and War Machine contribute to their defense. The Prince also helps protecting the minors on 4th rank. A strategically placed Leo initially protects three back-rank pieces and 3 Pawns.

I moved the Frog to the back rank, to provide more protection for the piece in the corner. Normally I would put minors in the most advanced locations. But the 4th rank cannot contain all of those anyway, and the Frog can move forward pretty fast through its 3-leap (which initially protects the otherwise unprotected War Machine).

So on the back-rank a1 is protected by Warrior (c2), Frog and Leo, b1 by Warrior (d2), Dragon King and Zebra, c1 by Warrior (a2), Leo and Cannon, d1 by Warrior (b2) and Leo, e1 by Warrior (c2), Dragon Horse, Amazon and Bat. The Frogs at d1/l1 thus are the weakest spots. Extra protectors can be arranged for them after the Zebra or Camel have been developed: a War Machine or Leo at d3, or moving the Falcon to e3 to discover the Vao.

With such a compact setup the pieces must not obstruct each other's development too much. The idea here is that the e- and l-file can be used as an exit corridor from the camp for orthogonal sliders, and the diagonals f1-a6 and k1-p6 for diagonal sliders. The Knight and Zebra on these files, and the Camel on these diagonals can all leave the camp in a single move, to quickly clear the corridors. The Bishop and Dragon Horse already start on these diagonals, and the Cannons on the files. The Griffon and Rhino are set up to slide into the file or diagonal, respectively. After the latter are gone, a second diagonal can be cleared by developing the War Machine, allowing Queen and Archbishop to deploy. The Lion and Chancellor can be developed through their Knight jumps, when some space on 4th rank becomes available.

The King was moved to the back rank, to allow it to (fast-)castle to behind the Warrior shield. With the King on a central file the corner piece would end deeply burried after fast-castling, though. To prevent that, the King was placed quite asymmetrically, on the k-file, and the corner pieces were made Dragon Kings rather than Rooks. Fast castling would then transport the Dragon King to k1, just beside the intended exit file, which it can reach even without clearing away the Cannon.

The Rams were all placed on the back rank. Like the Warriors these are pieces intended for use later in the game, to break through Pawn chains that have been formed. Their Grasshopper non-captures should allow them to get out and to the front line easily and quickly, once the camp is mostly cleared. The pieces behind the Warriors are mostly jumping pieces (Nightrider, Ram, Frog) so they can leave without the need to compromise the wall of Warriors.

The Advancer capture I gave to the Terror might be too strong. On a sparsely populated board there would usually be no defense against it other than moving a threatened piece away. And the forking power of a Queen is such that you could almost always attack multiple pieces. So the game would degenerate into a giant plunder raid. Good for shortening it, but probably not very much fun, as the outcome would be almost impossible to predict, and thus depend mainly on luck. I therefore changed the 'super ability' of the Terror (i.e. what it has over a normal Queen) to hit-and-run capture on an adjacent square through a double King move. (But no double capture; the second leg must be a non-capture to another square than where the Terror came from!) This makes the possibility for threatening two pieces with locust capture dwindle when the board gets empty, and thus allows defense by moving away the threatened piece.

I don't think the (for chess) unusually high piece density hurts a lot, because the Warriors are best kept in reserve until the board population thins enough to have a chance to promote them. And using the Rams early in the game is also troublesome: they would hardly contribute fire power, and would take many more moves to reach the frontline on a crowded board than on a sparsely populated one. So 12 of the 80 pieces will initially not compete for the available space in the battle area, reducing the effective density to a more normal 53%.