Skip to Content

procedural board generation

12 replies [Last post]
Joined: 01/28/2017

hey all

I fancy having a go at designing a board made up of individual tiles that can be generated procedurally.

any starting point tips anyone can offer?


ElKobold's picture
Joined: 04/10/2015
Too wide question, I believe.

Too wide question, I believe. How you approach this will depend on the game's mechanisms. Plus, not every game will benefit from a randomized board.

In general though, I would start with a preconstructed board first, and decide on how to split it later.

Joined: 01/28/2017
its a procedurally generated

its a procedurally generated board Im after (or more accurately rules to generate more than one) rather than randomised

just wondering if there are existing approaches before starting reinventing the wheel

ElKobold's picture
Joined: 04/10/2015
krone9 wrote:its a

krone9 wrote:
its a procedurally generated board Im after (or more accurately rules to generate more than one) rather than randomised

just wondering if there are existing approaches before starting reinventing the wheel

Same advice really. To understand how to write the rules for generating the board, you need to understand how your "proper" board should look like.

That being said, I`m not sure if it's such a great idea to begin with as it will probably increase your setup time.

Joined: 01/28/2017
yeah I am concerned about

yeah I am concerned about that - would need to be super simple.

I have the vaguest inkling of an idea on how to do this so will have a play and report back to get feedback

Iyidin_Kyeimo's picture
Joined: 05/09/2017
One of the first thing to

One of the first thing to consider is the procedure you are using to generate the board. Will it be player generated during the game (a la Archipelago), or generated during setup at the start (Food chain magnate for example).

Does every tile need to be able to join to every other tile, or will it be more restrictive?

Are there paths that need to link up (i.e. have roads/rivers/rails always in the centre of an edge so they are continuous)

What is the art style going to be? Will it matter if it looks a bit clunky, like in Roads and Boats, where the different terrain types are more abstract?

Or, is it even to do with terrain at all? Are you generating action spaces?

I do like a variable board though, particularly something like FCM (which is a real favourite of mine) as the different boards offer hugely differing strategies and working that out is a significant part of the puzzle.

ssm's picture
Joined: 04/06/2017
The first things I did

The first things I did was-
1. Figure out what I want the final size to be for a complete play area.
2. Figure out if I want it contained or free-flowing. (Will it be a square/rectangle or will it just be random and hope people have enough room.)
3. Figure out the size of each tile depending what I want the player to do on that tile. (Will the player make 2-3 moves on that tile or is it sort of single use.)

BHFuturist's picture
Joined: 11/01/2008
Good question

As others have stated there are things that would be helpful to know going into such a venture.

The type of possible movement the game will have and what sort of pathing or obstacles can block or limit player movement.

The ideal distance between points of interest (locations).

The method of generation during or prior to game start. (random draw from a stack of tiles, scenario book with a set of numbered tiles).

The procedure side is the simple part once the design goals are known. The hard part is not making it take a long time to do the steps of the procedure and integrating such a process into the fun of the game.

It is just the rules for how tiles are placed to form the board. If drawn randomly as the game is played the rules will be similar to that of Carcassonne where the rule are focused on ensuring tile edges match and what to do when no tile will fit in an empty space. If tiles are numbered and players need to build a board using information from a game scenario book, setup time will be quite long.

On the other hand, if you are trying to gain a dynamic world where locations and features are handled by rules to make sure they are not random then you may need to have more than one layer of tiles and tokens. Some things might also be done where tiles are drawn from typed decks based on the current game state and the edge conditions presented by the side of the tile the player is exiting.

Something like this could be easily tested with Carcassonne tiles. Just have rules like:

When leaving from a grass edge roll a 1d6. and have a table of results for what deck of tiles to draw from.

This is a large city so when leaving from a city edge draw from the city tiles until the city is 4 tiles (the fifth tile from the city edge deck).

This could be simplified by using the same methods used in the creation of Memoir 44 dice (table probability reduced to a set of custom dice with non-random faces).

Also, having the procedure can be given variance through randomness without making the board random generation. Having the generation be purely procedure based will place all of the burden of generation on the players. If that is your intent I would recommend making the procedure as deterministic as possible (based on some sort of player choice). This makes the generation part of the game rather than part of the setup or upkeep of the game.

This is a very large topic as procedural generation is a function of computer logic and not very interesting for humans to walk through manually unless it is integrated into the fun of the game as in Carcassonne.

My advice is to keep it as simple as possible where there are clear rules for how tiles can connect and provide some very easy "if/else/until" type rules for things that might confuse the players. "If the tile you need to place does not fit discard and draw a new tile" or "Until you place a city tile you cannot place a tile with X icon" or "Save any tile with X icon until you can place a tile with Y icon, then place X tile connected to the tile with the Y icon."

If you give more information as to the goal of the board generation within the scope of the type of game you want to make and the role the board serves in gameplay, then we might be able to better focus our feedback and suggestions.

I hope this halps in some way, best of luck!


pelle's picture
Joined: 08/11/2008
I can't remember playing a

I can't remember playing a game with procedural generation rather than just random maps being generated or generated semi-randomly by players placing tiles somewhat freely.

One idea I have had is to use a variation of a maze-generation algorithm to make terrain. I have tested it a bit in software, but I think with modifications it could be simple and fast enough to use for creating a boardgame map as well, either during setup or during play as terrain needs to be revealed. Not 100% sure about what combinations of tokens/tiles would be useful to make the process as streamlined as possible, and I am not quite prepared to give away all the secrets about that idea (heh), but this book is excellent for some inspiration (although most algorithms it describes are only useful if you want something that just looks like a maze...):

In general procedural generation tends to use algorithms that you really need a computer to use though, so there might not be many options that are feasible to use in a boardgame.

Joined: 08/09/2014
Starting Points

I agree with what was said above, you need to know what you want the final board to look like and design creation rules to achieve that.

Almost every tile laying game is procedurally generated. Otherwise they would just be randomly placed tiles. Two games came to mind when I read this.

The first is Xia, which has you draw a random map tile when you explore a new space, the location of the tile is determined by the direction the player is exploring, but the orientation of the tile is determined by symbols on the tiles. I'm not sure if this is just to reduce the number of decisions a player needs to make and speed up the game or if the symbol placement is intentionally designed to prevent or cause certain adjacency issues.

The second is Betrayal at House on the Hill, which again has you draw random room tiles when you explore but the back of each tile shows which floor(s) it can be placed on, basement, first floor, or second floor. Some can be placed on two floors or all three. But it prevents things like the furnace from being anywhere but the basement. The orientation of the tile is also partially restricted because you must connect as many doorways as possible.

I could see a system with some information on the tile backs working for an even more restrictive set of placement rules.

Another game that just came to mind is Eclipse. There are 3 levels of tiles that are placed based on if it is the inner ring, mid ring, or outer area of the board. So each area has a different distribution of planets and aliens.

Good luck with your design.

lewpuls's picture
Joined: 04/04/2009

I'd only ask, what makes you think the procedural construction will be significantly better than random construction? The time it takes for PC must make the game better.

Joined: 01/27/2017
Procedural includes random

In principle, simply pulling from a shuffled deck is procedural construction, but it's pretty clear something more involved is being considered here.

I think it's essential to keep player decisionmaking to a minimum UNLESS this is intended as part of the game"s strategy.

For example, it would be fine to have icons at the tile edges, and direct players to draw the next tile from a deck with that icon, matching as many edges as possible but requiring a match on the edge that prompted the draw. Then it. Scones a case of play testing how game-breaking mismatched edges could be.

Joined: 01/28/2017
Thanks for all the commments

Thanks for all the commments guys - interesting reading (and I might pick up that Mazes for Programmers book)

I have made a fair bit of progress on this but its purely in spreadsheets/in my head at the moment so will be doing a chunk of playtesting to see how well it works.

Here's my thoughts....

My game is a dungeon crawler called Caverns of Doooooom. Imagine a set of tiles. These tiles are laid down as they are revealed by the players based on the range of their torch.

Random layout - simply draw a tile and lay it next to the next based on permissable permutations. Like Sorcerors Cave or Betrayal at House on the Hill.

Procedural layout - play tiles in a specific predetermined order to create a map as designed by the creator. Through simple rules the player creates complicated maps that are exactly how the designer intended.

Procedural is better than Random as long as that process can be kept simple and intuitive, because it means you can create a designed experience. I am however hoping to use the same tools to create random layouts too.

One way - for example - is to have cards which depict tiles in certain formations. These cards have numbers on the back. If I list a sequence of numbers referencing specific card layouts (eg 1,4,2,9,3,5) then I can create a simple procedural map which is created by the players in the order I specify, very simply and with a suitable in game rationale as to why the map is revealed in chunks (eg a door is opened). Through this process I can create a large variety of maps, with different characteristics.

Unlike Descent, where you layout the entire board at the start of the level, I want the suspense of discovery throughout the game. (I focus on immersion and achieving certain experiences at the heart of games so this is important to me)

Syndicate content

forum | by Dr. Radut