New True 3D Chess

I've recently graduated from a game design program and was laying back for awhile; decided to watch some of the classic Star Trek series and looked into it's classic tri-dimensional chess. Started wondering what it'd be like if someone made true three dimensional chess; did a bit of research and found that a few variants of true 8x8x8 grided chess had been created; but largely they never got very far; weren't popular and were hard too difficult for players to understand; the most popular one involving having 8 boards and a full standard set excluding the king on each (the king being on board 4 for white and board 5 for black), it's easy to see how this got confusing. But, when I set the game up as a real cube in a 3D game engine; I end up with something that players could more directly and easily read.

where the big screen is the board as a translucent 3D cube and the two smaller cameras are a single "layer" of it (the upper cam from the top and the lower one from the left side). I'm now working on developing a new variant based on this archetecture and am looking for input before I publish the game to Unity webplayer.

Things that are noteworthy about how I've been determining things so far:

The queen now has a total of 26 directions in which she can move. The bishop can move 8 and the rook only 6. Due to this big gap in material value and the high piece count (128 (2x8x8) total on each side), I've been considering adding new pieces for the variant; though I'm not certain how much resistance that would meet.

There are 64 total pawns on each side; which depending on how you interpret the capture behavior in 3D (I haven't made up my mind) either protect 4 spaces or 8 and either can form a single wall as normal or two separate stitches of 32 (like bishops half the pawns couldn't help the other half).

Simply put; any thoughts? Would you like to play this? etc.

First version I publish will be local multiplayer only (in other words white and black move using the same computer and there is no versus computer option) this is because I don't have the funds to set up a server for a game that I'm not certain will go anywhere for non-local multiplayer and having the computer have good AI to play against the player would take much longer for me to write than the rest of the project. Noteably it's actually harder for me to make a reasonable difficulty AI, than either nearly impossible for a non-chessmaster to beat or almost completely random/just understands the rules and makes moves it 'can'.

looks cool and fun... i would play

The idea is kinda cool, but I'm already lost with 16 pieces, I doubt I could handle 128

This looks all insanely complicated. I doubt many people will take a serious look into it. 

reminds me of this game: http://en.wikipedia.org/wiki/Taikyoku_shogi

If it is turn based, then it has no hope. Thinking with 64 (!) pawns, and any other pieces you have, and 512 squares (!) is beyond our human capacities. Heck, only the top engines can beat the top players around the world (read win not draw) and that's with only 64 squares and 16 units!

The number of units is large, but considering half of the units are Pawns, this does not need to be a fatal flaw. Chu Shogi has 46 pieces per side, and is a very playable game.

Your association of diagonals to body diagonals is kind of arbitrary, and applied inconsistently: A Queen is no longer the compound of a Rook and Bishop. In 3d there are 3 different main directions (i.e. to neighbor squares), so it seems there should be 3 basic sliders, and you made your Queen a compound of all 3 (no doubt to make it a slider version of the King, i.e. a "Kingrider").

The main problem with 3d variants is that 26 escape directions for a King makes it exceedingly difficult to checkmate it. This makes the game extremely drawish, and thus uninteresting. (Especially in combination with an excessively long playing time.) The game would improve a lot if the number of King moves would be limited. So you could consider having the King move in only 14 directions (6 Rook and 8 Bishop directions), and the Queen likewise.

Whether your claim that this would be 'true 3d Chess' is valid is debatable. There is more than one 'natural' generalization to higher dimensionality possible, and the one you choose is in fact a bit unnatural, as you arbitrarily drop one of the 3 main directions. In stead of seeing the sliders from orthodox Chess as 1-dimensional movers, you could also consider them as (N-1)-dimensional movers with N=2. In that case, generalizing to N=3 would make pieces that move in planes the equivalent of sliders. This would make the relative properties of the pieces much closer to normal Chess (e.g. two Rooks moving in planes parrallel to the cube faces would be able to deliver checkmate). Pieces moving in a plane could be implemented similar to the Hook Mover of Tai Shogi: an arbitrary number of steps, and allowing to change direction once (in general N-2 times). This would still makes them reasonably easy to block (2 obstacles would do this, and deny them access to a large part of the plane). Bishops would then move in planes containing 2 body diagonals.

Pawn shileds are probably best only 'loosely knit', to make breakthroughs not impossibly difficult.

I am not sure why you are sold to 8x8x8. With just 6 piece types this would lead to every type being present in an enormous number of copies, which would give a very un-Chesslike feel. One could also argue that the natural extension keeps the number of board cells constant, e.g. 4x4x4 = 64.

An AI for any of these proposals should not be any more difficult to make than for any other Chess variant. The micro-Max AI for orthodox Chess, for example, is only ~100 short lines of C code, and would hardly need any modification.

Is this chess inside 8x8x8 cube basically? 

Something like this:

Looks cool! I've had this Idea for pocket billiards, in a cube :D But that would only be playable in space and in a hologram (how else are you going to shoot inside the cube?)

Personally I think you should just do the normal set (8 pawns + 8 pieces). But as said that could give drawish games. How about: the king can only move on its horizontal plane, and if it wants to go up or down it can only go straight up and straight down.

But on the other hand: there is an extra dimension. And I think it requires new pieces or an adjustment to the old pieces that doesnt change the relative capabilities to much. (eg: bisshop should stil be about as strong as the knight, and 2 rooks should be a little better than 1 Q etc.

This is an idea that must have crossed many minds, but you're the first to make this work! Really cool!

That is a nice setup. If I understand correctly: a piece can move of the diagonal plane, move around on its level and 'join' the diagonal plane again? And when its on the diagonal plane it may move over the diagonal plane as if the plane was horiontal?

That could make it work without changing any piece.

*edit* if a pawn moves off that diagonal plane it can never return :/ And I assume that a pawn can only promote when it reaches the end of the diagonal plane, not the vertical plane. That would make it a game of stopping pawns. And the first thing to do is bring your king to safety on you horizontal plane. The opponent must bring pieces to that plane in order to deliver a checkmate..

Here:

when does a pawn promote?

I don't know how to multi-quote on this forum (or even if the forum can so.

HG Muller: "I am not sure why you are sold to 8x8x8. With just 6 piece types this would lead to every type being present in an enormous number of copies, which would give a very un-Chesslike feel. One could also argue that the natural extension keeps the number of board cells constant, e.g. 4x4x4 = 64."

4x4x4 either starts with pawns practically or literally already touching the opponent's side or the King exposed. Not a great way of starting. As of right now; finding an appropriate way of an 8x8x8 starting setup is my plan; that said; I'll leave the game able to use any board size (technically from 3x3x3 to 1024x1024x1024; though no one in their right mind would try either extreme) and any starting setup (including a separate starting setup editor that saves a readable file of your setup for later use). I want to open the doors to customization to try to perfect the new variant, but I also want to try to come up with the best starting standardization I can. That said; when I used to play regularly; the highest I ever was was the Class C range on USCF ratings and that's been years, so I'm unsure of my skill at present, so masters and greater might stand better odds at editing it to more tournament capable. Let's say I'm programming/manufacturing the board and other people can make pieces, setups, openings, etc.

HG Muller:  There is more than one 'natural' generalization to higher dimensionality possible, and the one you choose is in fact a bit unnatural, as you arbitrarily drop one of the 3 main directions. In stead of seeing the sliders from orthodox Chess as 1-dimensional movers, you could also consider them as (N-1)-dimensional movers with N=2. In that case, generalizing to N=3 would make pieces that move in planes the equivalent of sliders. This would make the relative properties of the pieces much closer to normal Chess (e.g. two Rooks moving in planes parrallel to the cube faces would be able to deliver checkmate). Pieces moving in a plane could be implemented similar to the Hook Mover of Tai Shogi: an arbitrary number of steps, and allowing to change direction once (in general N-2 times). This would still makes them reasonably easy to block (2 obstacles would do this, and deny them access to a large part of the plane). Bishops would then move in planes containing 2 body diagonals.

As it stands right now; because there are multiple ways of interpolating into 3D linearly I'm creating new piece types that are basically different interpolations of classic pieces; like you could interpret Bishops linearly into moving 8 diagonals two different ways or all 16 diagonals.

Planes movement isn't something I considered for long because it either needs to be interpreted as literally teleporting or two moves in quick succession in one turn. The former would be broken and the latter; is feasible but will be harder for the average player to strategize (and for me to code the computer into understanding when check and checkmate are). I do admit the 'planes movement' DOES make the material values remain similar instead of being completely thrown through the loop AND there is only one way of interpretting each slider into 3D that way so...it may be better IF people can play, understand it. I'll patch that mode in after I finish the linear interpolations-based one.

HG Muller: An AI for any of these proposals should not be any more difficult to make than for any other Chess variant.

Didn't say it would be; said it would be easy for me to make stupid player or normal player, but in between is difficult. Stupid would literally be randomized legal moves and take no time at all to code; normal player wouldn't necessarily have a good endgame but would be able to play defensively pretty well.

El Kitch: Looks cool! I've had this Idea for pocket billiards, in a cube :D But that would only be playable in space and in a hologram (how else are you going to shoot inside the cube?)

Billiards in a 3D video game isn't too terrible of an idea and is relatively easy to project; actually could probably set that one up in no time flat. O_O; (though yea, in real life would require holograms or a weightless environment (and would have a high chance of being hit with balls in the latter).

ajttja: when does a pawn promote?

In any of the versions I'm working on the answer is the same: when it reachs the plane the opponent's color started on. (Same as standard, only it's a plane of squares rather than a line).

Have you already figured out how many (26-directional) Queens you would need to checkmate a single (26-directional) King, on 8x8x8? Would KQQQK be a win, or would it still be a dead draw?

I guess Queens still aren't too bad, mainly because their single-step moves for an inpenetrable wall. The real problem would be to mate with Rooks.

Not sure but i think it should be possible to drive the lone king to the edge of the board and deliver checkmate.

http://cp4space.wordpress.com/2013/03/30/three-dimensional-chess/

2 Queens protected by pawns (or anything) are enough for middle of the board. 1 Queen protected by pawn is enough for a corner.

For Rooks alone, erm...middle of the board would be 6 of them alone and probably not that feasible (hard to set up the six cube without the king taking one of them). Corner would be 3.

My favourite 3D-chess variant is Raumschach (5x5x5):

http://en.wikipedia.org/wiki/Raumschach 

I like the idea of 3d chess in a 3-board environment - one set of pieces and easier but not easy to play - I don't think anyone has made a move generator for this game - if you are good enough to program that - I would be grateful, I am tired of the same old chess and really like cube-chess and its much more playable - I am a former chess master  I'm a shut-in from illness and this would pass the hours , my friends are dead mostly and I can't go to the chess club to play , I don't like online chess because people cheat.  A move generator for my 3d chess set would be fun.

chess is not a video game and is fine as it is. 

 next you'll be saying you need to give chess better graphics. this idea should be scraped. don't fix things that aren't broken.  

Let's play 4x4x4 = 64 chess by the classical board with 6 pawns, 2 rooks, queen and king for each player! Are there at least two real chess players?