The Mapmaker's Companion

[Tarkeel]

This thread is meant as a resource for anyone interested in making custom maps for Civ4. While I have some experience in the area, I don't proclaim anything herein as facts, but more a guideline to get started. Hopefully it will also serve as a place for other map-makers to share their tricks, which I can work into the text.

Part 1: Concepts
Part 2: Basic Balancing
Part 3: Starts
Part 4: World Design
Part 5: Worldbuilder Basics
Part 6: Worldbuilder Hacks
Addendum: Tips and Tricks
Addendum: Resources

[Tarkeel]

Part 1: Concepts

What exactly does making a map for cIV entail? There's no set answer, as it depends on what kind of map is being requested. The most important step is deciding what kind of map you want to make, and what they players want to play on. You go into this process with an idea (subconciously or not) of how your map should look. Roughly ordered by amount of work needed, the kinds of map can be classified as:

1. Random (see PB64 and PB69)
   
2. Modified (see PB61 and PB63)
   
3. Crafted (see PB60 and PB70)
   

Random usually means rolling two dozen or so maps, and pick the ones with most interesting shapes for further analysis. Main focus of analysis should be neighbor-count and areas of expansion, the goal of these games is usually more towards playable fun than perfectly balanced. In general these games should be modified as lightly as possible, just ensuring access to the basic strategics of copper, horse, iron and oil, with possibly light start editing.
Example: For PB64, Mjmd rolled a map, and made sure everyone had access to copper/iron/horse for bare minimum balance.
Example: For PB69, Mjmd rolled 10 tectonics maps, picking out the two most interesting shapes for lurker feedback. Through discussion we settled on a minimum amount of changes to the starts to ensure all were playable. Other changes were limited to adding copper and removing ice for navigation.

Modified starts like random by rolling maps until an interesting candidate is found, but just how much is modified varies from "not much" (like PB73) to "most of the map" (PB63). The main benefit is that you get a naturally randomized map to work from, limiting the amount of decisions needed.
Example: For PB61 I rolled pangeas untill I found a pleasant shape; I knew I wanted a bay in the north or south to divide the two central starts. I then disregarded the naturally rolled start locations and placed them roughly equi-distant around the coast, some of them just inland. I find it very helpful to mark the midpoints to clearly see where the natural borders show up, and then started working on the borders. Since this was a greens game we wanted some naturally defensible terrain for borders that wasn't so lush as to spark instant conflict.
Example: For PB63 I wanted to recreate the feeling of PB62 with a single large landmass with interspersed bodies of water. To do this I rolled a large Lakes map, and editing the worldbuilder file to cut off the top and bottom 8 rows, and some more magic to remove columns. I should have just rolled a Torusland map instead.
Calculated start locations to form a skewed grid resulted in this, which after heavy editing of borders and water end up looking like this.

Crafted on the other hand start with a blank slate, and is then hand-drawn. This takes a lot of effort and a good idea of what type of map you're looking for. Due to the clunkiness of the in-game editor you're best off sketching the early drafts by hand to get a feel for the layout, before starting to calculate the distances involved. One thing to remember here is that civ is likely to misbehave if the map dimensions are not a multiple of 4. If the map involves mirroring, this can set a hard constraint on which dimensions are avilable.
Example: For PB60 I was inspired by my early misconception of how the PB58 map was: pairs of players sharing a each continent, with isthmuses leading to the other players. An early paint-mockup led to a very mirrored map. I was at a loss for a good way to connect the plates, but looking at the overview gave me the idea of shifting each plate by a third of the map, eventually forming a double helix donut pattern, in which everyone was neighboring each other.
Example: Since PB70 featured five players, I wanted them arranged in a off-set grid to harmonize number of contacts. Since the number of players (5) does not have any denominators in common with the base dimension size (4), each dimension has to be a multiple of 4x5=20. Using Excel I laid 40x60 and 40x40 examples for visualization. Experimenting with the worldbuilder yielded this water-heavy whirlwind pattern, before I settled on the more classic strip.

One important decision that I've mentioned briefly is wrapping. Toroids give better neighbor balance but some people have trouble wrapping their head around it.

Map Scripts
Some suggestions for mapscripts to use.

• Tectonics produces rather dry maps in intersting twisting shapes, usually wrapping east-west. Suitable for any player count, but probably best for 6 or more.
  
• Pangea prouduces a single landmass, with usually players starting along the coast. Can produce very uneven neighbor-balance, best for no more than 8.
  
• Big and Small produces several smaller landmasses with varying amount of players on each, along with lots if islands. Needs careful checking of luxuries.
  
• Torusland is a very versatile script than can create various shapes that in general defy natural expectations, all of which wrap in both directions.
  

Lastly, the closest thing to a hard rule I have while making a map: Always assign starts to players randomly, preferably by someone else (except when starts are drafted: See PB68 and PB73). While crafting the map, always refer to starts by some identifier (numeric or alphabetic) to avoid implicit bias from the player slipping in.

[Tarkeel]

Part 2: Basic Balancing

What exactly does balancing mean? Again there is no set answer, but the main factors to consider are:

• Quality of starting location
  
• Access to strategic resources
  
• Access to luxury resources
  
• Expansion options (including islands)
  
• Contact with neighbors
  

Before diving into each of these in detail, it's worth taking a moment do discuss degrees of balance. On the most perfectly balanced side is a mirrored game, where everyone theoretically has the exact same resources and thus chances of winning. In practice mirroring can lead to stalemates and less fluid games. The other extreme is mostly random games slightly tweaked so nobody has a truly unwinnable position. This is an area where it's important to communicate with the players what they want and expect, but my preferred setting is to balance for fun and fair.

Quality of starting location
Balancing starts is such an essential task that it gets a separate chapter, but in short you need to remember that food is the most important resource in the early game and not every food resource is created equal. In particular animal resources have a very high research cost, while seafood resources have an extra cost in producing workboats but payout in higher commerce. This means that of the four tehnologies that unlock improving food, you should avoid picking more than two; usually one of hunting and agriculture combined with fishing or animal husbandry. The reason for this is to limit the amount of beakers the player is forced to invest in food techs, giving them more options for other goals.

I find it very helpful to decide on a set of guidelines for what each start should contain. Below is an example from PB63 that shows how I defined guiding principles for the starts. You do not need all of these parts to be equal in every game!

The principles I've used are that each start should have the following resources:

• One primary food, inner ring: wet plains-wheat, grassforest deer or lakeside fish. (5f1h vs 6f2c)
  
• One secondary food, outer ring: dry plains-wheat, plains deer or lakeside crabs/clams. (4f1h vs 4f2c)
  
• One bananaforest, inner ring (3f1h)
  
• One oasis, outer ring (3f2c)
  
• One grasshill pig, outer ring (5f1h)
  

This should result that everyone has two 5f and one 4f tile; one of which will be 2c and the others 1h.

Terrain-wise I aim for the starts to have:

• 9 river-side tiles, of which 5 are flat grasslands, 3 hills and the bananaforest.
  
• 6 forests, atleast one of which is a plainshill
  
• 6 hills, with atleast one each of plains and grass that isn't forested, including the pigs.
  

This comes in addition to the capital site itself, which will be a river-adjacent plainshill.

Lastly, the player can opt to move their settler, so you should have a quick look this doesn't break anything too much: either with orphaned seafood, or picking up too many good tiles.

Access to strategic resources
Copper is usually the first strategic resource players go looking for, as it's one of the most important in the early game. Players usually prefer that their copper can be claimed together with a new food in the first ring, but It's most important that every player has similar access here. Don't give one player wet corn with their copper while the other has no food. Another thing to consider is direction towards neighbors; two players with their copper pointing towards each other are more likely to go to war than if the copper was in their backlines. Don't be afraid to give everyone two coppers pointing in opposite directions to lessen this effect. Lastly consider what is needed to connect the resource to the capital: a player whose copper is on a river that connects to the capital can forego researching the wheel, so it's often best to place it on a river-less tile.

Horse is another resource players look for early, and can determine whether they try for a chariot rush or not. Placing this in an awkward position can slow down early long-range aggression.

Stone and Marble can have a big influence on early wonders. Most importantly, they are not interchangeable: you should not give half the field stone and the other half marble. Stone is essential for defensive walls and the Moai national wonder, as well as early wonders like Stonehenge and Pyramids. Marble is more a mid-game resource boosting wonders like TajMahal and the two epics.

Iron should be accessible to all, but don't need to be that close as it's importance comes in the mid-game. Again make sure the players have similar access, but try to avoid placing on riverside grasslands as they are most likely contain mature cottages.

Oil is the singular most important late-game resource, and care should be taken that everyone has similar access to defensible oil. While oil can be placed in water as well as land, accessing land oil is much easier and earlier than water.

Coal, Uranium and Aluminum are all important that they player can pick up, but usually don't need too much thought in their placement as they appear late.

Access to luxury resources
Luxuries can be lumped into two rough categories, Ancient and Classical, and each player should have natural access to a similar amount of distinct resources in each category. A good baseline seems to be two of each, but this is an area where you can take some liberties, especially with contested resources. Just remember that playing without an early luxury hooked means any non-capital city is unhappy (at size 4) from any whipping, and is quite miserable to play around.

Another thing to remember is that not every luxury is equal. The ancient resources are pretty equal in terms of happyness; ie all metals are doubled with forge, while furs and ivory are doubled with markets. There's a big difference in tile yields though, where you'd prefer working improved ivory over furs most of the time, and silver is very inferior to gems and gold. Trying to balance out these yields is left as an exercise for the reader (in other words, it's hard). Note that access to gems can be delayed by using jungle.

There's a much larger difference between the classical resources, both in unlocking tech (calendar, monarchy and optics), happyness yield (spices, sugar and wine give health instead) and even tile yield. Among the calendar luxuries, silks dyes and incense are mostly interchangeable, while the same is partially true for spices and wines, leaving sugar and whales as the odd ones out.

If all of the luxuries are present on a map, there's a good chance for happyness to become a non-factor in the late game, so you should try to leave some out; about four seems to be a good balance.

Expansion options
This was briefly covered under strategic resources, particularly copper. Games are most interesting when players have several viable options, so make sure they aren't locked in to only one site for their first city. The main tools at your disposal here are placing plainshill for 2h cities, "dead" tiles such as tundra and desert the player will want to settle on, and resource placement. Using the in-game dotmap tool you should try to settle the inner ring around the capital, and if possible you could try to have the "optimum" sites have their resources in the outer ring for a CREative benefit.

Access to islands can be very powerful due to how intercontinental trade-routes (ICTRs) work, so everyone should have similar access to islands, or at least be a concious choice to break this.

Contact with neighbors
Lastly and possibly most importantly, players should have similar access to neighbors. One of the worst things that can happen to a map is having a player with only one neighbor, which means that is the only avenue they have for expansion. Some examples are PB52 where Superdeath had a larger peninsula to himself, with Pindicator as the only neighbor, leading to inevitable war, while PB66 had several players with only one neighbor (at the very least, Superdeath, Amicalola, Alhazard, Plemo&Piccadilly). Oversea neighbors can to a certain degree compensate, but not fully, and more towards the later stages.

Another reason for balancing this out is making contacts to earn known-tech bonus on researching. Consider a pangea with an unlucky player in the middle, they would atleast have easy contacts with everyone for more research bonus.

[Tarkeel]

Part 3: Starts

A customary start will reveal the 5x5 area centered on the starting units, regardless of which tiles are actually visible. By custom anything inside the BFC should match what is in the final game, while the corners (and fog) might not, but the players should be informed what is the case. The defining part of a start is which resources are present, and which techs are required to improve them, but other features such as hills, forests, river and water needs to be considered.

Starting Tile
Let's begin by talking about the city tile. Any city will produce at least 2/1/1 regardless of the underlying tile, but if that tile has a higher value so will the city. Note that features such as forests, jungle and flooplains will be removed by the city and thus don't count.

The classic example for altering this yield is the plainshill, which unimproved produces 2h. The other side of settling on a non-default tile is what improved yield you lose out on, which for the mined plainshill is just a 4f tile: nothing special. The other way of improving the yield is to add resources: While any food resource on a grassland flat will yield a 3f city, most of these will make you lose out on something too big later on. However bananas and sugars (and possibly dry rice) are less of a loss and more viable to settle on. For commerce, you can combine a river with most luxury resources to create a 2c capital. Once again, calendar luxuries or wines are the best bet to use here.

But why does this matter so much? Short answer is, a single foodhamer can speed up the first worker or workboat and gets the snowball rolling faster.

Opening Moves
Consider a city working an unimproved 3 foodhammer tile, which is common and easy to find. Most openings start with a worker costing 60 foodhammers, and with only 1 spare foodhammer from the city it will take 60 / 4 = 15 turns for the worker to finish. Add in a second food or hammer from the city tile and the timing drops to 60 / 5 = 12 turns. If the start also has a 4 foodhammer tile (such as forested deer in a hunting start) the time drops even further to 60 / 6 = 10 turns.

This neatly illustrates why it's so hard to balance hunting vs agriculture starts; deer without a forest has similar yield to a dry corn or wheat, but the forest (bring it up to a wet) itself will skew the base yield and speed up the worker. The classic solution is to make sure everyone has a 4 foodhamer tile to work, with forested ivory being a favorite. Since improving the ivory is a high priority for both yield and happyness, this adds tech requirement and takes away choice from the start. My personal favorite is to use bananaforest instead, which while isn't a naturally occuring tile does give some interesting option how you improve the tile down the road: You can either chop and cottage it early, or turn it into a lumbermill or plantation later in the game.

Workboat openings are different though, since they are normal hammer-built units costing 30 hammers. With a plainshill city working a first-ring forested plainshill providing you can produce one in 30 / 5 = 6 turns while not growing, and then swap to working the 5 or 6 food tile to build your worker. Alternatively you can use your food surplus to grow; working the aforementioned superb 3f1h tile for 8 turns you grow to size 2, and can then finish the workboat on T9 working two hammer-focused tiles. This is slower workboat is nevertheless better as you've also grown to size 2 and can work two tiles for the worker.

The takeaway here is that workboat are much worse off if they don't have a forested plainshill or a 3f1h tile available.

Resources
Just how many food resources should a start have? Let's start by defining this as a tile that yields at least 4 food after improvements. Most starts with two of these is playable, and a third starts becoming juicy. These should require no more than two different technologies to improve, so as to not dictate all of the early research goals. Animal resources distracts the player from the dogmatic opening of bronze working first, and should at the very least be combined with huning or agriculture resources since they are the necessary prerequisites. Combining animals with seafood is heavily discouraged (even if I did just that in my first map. Whoops).

A run-through of the starts I've made before:

• PB60: 2x 4f + 3f: Cattle and clam/crab with lakeside whale. Option to move to grab a 5f1h deer or wheat instead of the clam/crab
  
• PB61: 2x 5f + 4f + ivory and hillside animal: Very juiced starts for a greens game.
  
• PB63: 2x 5f + 4f +2x 3f: One of the 5f is a hillside pig, and the idea was they rest of start tiles were good enough to not need early AH.
  
• PB68: 2x 5f + 4f + 2x 3f: Intentionally juiced starts for quick-starting Always-War.
  
• PB70: 3x 5f + 3f: Slightly too juiced?
  

Looking back on them, I see a tendency towards juicy starts, and in hindsight many of the starts would be improved by dropping the animal. Juicy starts have some benefits as they speed up the otherwise boring early turns, as well as making the players feel good. On smaller maps it might skew the balance somewhat, giving a too large benefit to any player that can eliminate a neighbor. I think a good guideline is three tiles in total that provide 3 or more food, with atleast two of them providing 4f or more. As I wrote about the PB60 start, where I considered adding in an extra grassland pig (6f):

I did a test-play up to T50 with and without pigs, settling in place as Mehmed of the Vikings. The pigs was a great boon, but mainly for getting the second and third cities up and running. It did feel more interesting to play without them, so I'm not putting them in.

For easy reference the improved food-yields (excluding hammers and commerce) for the common resources are:
6f: Wet corn, lakeside fish, grassland pigs
5f: Dry corn, wet wheat, wet rice, grass deer, fish, lakeside clam/crab, grassland sheep, grasshill pigs
4f: Floopdplains, dry wheat, dry rice, plains deer, clam/crab, grass cow, grasshill sheep
3f: Oasis, lakeside whales, plains cow, plainshill sheep

Remember that items on the same line are not substitutes for each other in the long term, as most will produce hammers or commerce. Dry corn/wheat/rice can usually be improved in the mid-game with civil service, and seafood will usually be improved by early lighthouses. Any resource covered by jungle can be discounted for the purpose of start balancing; likewise resources that can't be improved from the capital. (The list of 3f items is not exhaustive, and is limited to the early game)

Lastly, the positioning of resources matter: Outer-ring resources can't be worked before the capital borders expands after 5 (or 3 with CRE) turns and takes at least a turn for the worker or workboat to arrive. If a start needs to be made just slightly worse, position the resources so that the worker can't move between them efficiently. This often has the added benefit of boosting civs that start with wheel, since the worker can build half-roads on each move. Outer ring resources are also easier to share with other cities.

Other Features
Forests and hills are important to provide non-whip production, and especially the amount of chop-able forests needs to be taken into account. It's good to have at least one hill without forest for mining before bronze working if needed.

Rivers are important both for the bonus commerce on the tile, but even for so for trade connections without having to build (and research) roads.

[Tarkeel]

Part 4: World Design

Working with the map usually entails working with the in-game worldbuilder, whether you modify an existing map or create one from scratch. The worldbuilder itself can be accessed by hitting CTRL+W while playing, but only if the "Lock modified assets" option is not set.

Remember to save often, preferably using multiple files for versioning. I find it helpful to use the mapmaking thread to keep a log what I'm working on between versions. Since looking at pictures takes much less effort than starting a custom game you get a lot more engagement from the other lurkers by posting pictures as well as the WBS file. If possibly try to be explicit in what you're looking for feedback on.

Similarly you should use the sign feature to note key areas: every start should be clearly marked, and on most maps I also mark the midpoints between relevant players. If you want a sign to be shown regardless of which player you're currently controlling, you can use the "Landmark" feature in the worldbuilder.

Plot Anatomy
A plot (or tile) has four layers of information: The plot-type determines the height (mountain, hill, flatland or ocean) with an associated terrain-type (grassland, plains, desert, etc). Not all terrains are compatible with all plot-types. The tile can then be further refined by a feature (forest, jungle, floodplain, oasis, etc) and resource (food, strategic, luxury) which combined gives you the yield for the tile.

Natural Variety
There's nothing more boring to look at than uniform strips of the same terrain. Try to keep in mind how water interacts with mountains and rivers; try to leave consistent dry side on either the east or west side of highlands, while areas bordering a river should usually be greener than their surroundings.

Drawing good-looking rivers is something of a dark art, and it just takes practice. To draw a river you need to select a start and end tile, so it will always draw atleast two pieces at a time. Personally I tend to start at the river mouth, and then working my way up to the source with deleting and adding each piece.

Coastlines should preferably be curving or jagged and not straight, don't be afraid to add small coves or peninsulas jutting out.

Borderlands
Borders between players that narrow into chokes can easily form natural defensive which are very hard to break through, leading to static gameplay. Similarly making the borders too rich in contested resources are likely to lead to early flashpoints, and huge rewards for whichever player settles first (a big boon for IMP). Lakes in the border area can have very interesting effects, giving increased vision and reduced mobility for both players.

One method for creating more dynamic borders is to use belts of jungle or desert. Jungled lands are usually very variable later on, but require a commitment in technology and worker labor to make profitable. It's best to prune out a few jungle tiles so scouts don't get completely bogged down when traversing. Desert belts on the other hand are usually not settled, and can lead to open areas that players can freely move through to attack, possibly in an unexpected blindspot.

Similarly, don't discount the value that peaks have on extending vision and limiting movement.

Astronomy and Circumnavigation
Most water-maps have at least one continent that can only be reached after teching Astronomy, as a reward for what can otherwise be a niche tech. If water is already sufficiently strategically important (see PB66 and PB70 for examples) this is not necessary. This land can be of a higher quality than elsewhere to make up for it being settled later, but should not be overdone (see PB56) especially if ocean access is not equally distributed between players. Similarly you need to consider how hard it will be for players to circumnavigate the map before caravels become available.

[Tarkeel]

Part 5: Worldbuilder Basics
Civ4 stores map in a WorldBuilderSave file, which is human readable text arranged in a specific way so the game can read it. CivFanatics' wiki has an in-depth documentation, but this guide will cover the options you're most likely to have to edit manually to finalize a map. Just open the file in your favorite proper text-editor, like Notepad++ or Visual Studio Code. The file is organized into parts:

BeginGame
This part holds the settings for the game as a whole. The easiest way to set this correctly is to generate the maps with the requested settings, or copy them over from a new game where you've set them. You do not have to set the ModPath variable.

BeginTeam
Defines each team for the game, and in most games there will be one team per player. This is where starting technologies and contacts with other players are stored. In general, you want to make sure there are no ContactWithTeam entries here, delete as necessary (Exception: Team52 is the barbarians). For the Tech entries it's important to remove Archery from all the AI civs, and you most likely have to remove Hunting from a few as well.

BeginPlayer
Defines each player for the game, and you're going to have to edit most of the options from LeaderType to ArtStyle. The easiest way to do this, is creating a new custom game in which you set the options, leader and civilization for each player on the correct IDs. You can then copy the options into BeginGame, the techs into BeginTeam, and the players into BeginPlayer. Personally I like replacing the LeaderName with the player's name to avoid confusion on game startup.

BeginMap
Defines the map itself, with the most important attribute being world size defining tech and maintenance costs. Most games are played on WORLDSIZE_STANDARD, but communicate with the players around this: Smaller sizes have lower tech costs but higher maintenance.

BeginPlot
Lastly, each single tile (or plot) is described in intimate detail. You shouldn't mess around too much with this, but since each unit is stored on the plot (using BeginUnit), you can do a search-and-replace for UnitOwner=X to change where players are. Just remember to update tile vision in the worldbuilder afterwards!

[Tarkeel]

Part 6: Worldbuilder Hacks

Since the WorldBuilderSave (WBS) file is plain text it can easily be manipulated using advanced text editors or tools like perl.

This area will expanded in the future to describe other hacks.

Transposition
Transposition is the art of swapping rows for columns in a dataset, and can easily be done using a text-editor that supports regular expressions. We take advantage of the fact that the game will read plot info unordered, even if it will save them in ascending order. Using something like Notepad++ you can do the following search and replace to transpose the map:

Search: x=(\d+),y=(\d+)
Replace: x=$2,y=$1

Remember to update the grid width/height and num plots written attributes as needed.

[Tarkeel]

Addendum: Tips and Tricks

Q: How can I create a map with a given set of dimensions?
The easiest way is to use Sevenspirits' Torusland mapscript which let you set dimensions on generation.
If you for some reason don't want to do that, you can use very cumbersome method I used for PB63 (lightly edited):

I wanted to take the middle part (without all the tundra) and turn it into a 48x36 map. This was done by:
1. Roll a huge lakes map, dimensions 84x52
2. Use a perl script to shift every Y down 8. (52 = 36 + 8 + 8)
3. Edit the file to add in signs to indicate starts
4. Move all starting units to the indicated starts to avoid the player auto-dying on loading
5. Load the map in civ and save it again (*)
6. Edit the file, search for x=48 and remove all tiles from there untill the end. Remember to update the grid width/height and num plots written attributes.
7. Transpose the rows and columns using regex replace in Notepad++, search for "x=(\d+),y=(\d+)" and replace with "x=$2,y=$1"
8. Load the map in civ and save it again (*)
9. Edit the file, search for x=36 and remove all tiles from there untill the end. Remember to update the grid width/height and num plots written attributes.
10. Transpose the rows and columns using regex replace in Notepad++, search for "x=(\d+),y=(\d+)" and replace with "x=$2,y=$1"

(*) This step is so that civ will re-order the tiles.

Q: How do you make the map overviews?
They are actually taken from the replay screen. Making 100% sure I've saved the map, I settle in place and end turn, triggering the AI civs to do the same. I then retire and take a screenshot of the replay, croping it down to just the map.

Q: What is the easiest way to get screenshots for each start?
Firstly, you need to enable the "cheat mode", which is more accurately described as "debug mode". Find the .INI in your BTS directory and add CheatCode = chipotle. You can now use ALT+Z to cycle through civs while playing singleplayer.

Q: What are some interesting start gimmicks?
In PB49, Miguelito had cattle on a tile that wasn't accessible from the start.
In BP60, each start had stone visible on an island.
In PB67, I had a lake clam that the capital couldn't build a workboat for.
Most of my later maps feature bananas on a forest, a tile which doesn't occurr naturally, as 3f1h is a great balancing tile with some interesting options for how to use later in the game.

Q: How does lakes work?
Civ has three types of water, depending on size (unaffected by presence of ocean):
1-9 tiles of water: Freshwater lake; +1f but can't build lighthouse or ships
10-19 tiles of water: Saltwater lake; can build lighthouse but not ships
20+ tiles: saltwater ocean
Workboat are an exception that can be built if (and only if) there is a resource in the water accesible from the city.

Q: When is circumnavigation available?
The circumnavigation bonus is available on any map where atleast 33% of the tiles are water.

A minor point worth remembering is a resource on the tile S of the capital will obscure the city-plate when resource bubbles are on, so I try to avoid that for more readable screenshots.

[Tarkeel]

Addendum: Resources

• Charriu's mapmaking tutorial
  
• Sevenspirits' Torusland mapscript
  
• Sevenspirits' Mirrorland mapscript
  
• Novice's maptool
  

[Qgqqqqq]

Great thread! Makes me want to craft another map. Note that Novices tool can copy, rotate and mirror sections of the map.