The Geography of Fantasy

I thought I might write about the amusing conceit of a wealthy, powerful and famous scion of a chronically discriminatory industry hectoring the voiceless working poor about how they shouldn’t bully her clique, because it would generate the same. I’m sure it pleased the powers-that-be in Hollyweird and about 25% of the population enormously. Unfortunately they were 25% who liked her anyway and watched her films. Perhaps the Hollyweird grandees and her bank balance don’t care about the other 75%… Now. It was rather like New York Times promoting a blacklist of advertisers whose products appeared on Breitbart. It’s a weapon that works really well… if no-one replies in kind. Just as in publishing: virtue signaling to NYC publishing by attacking anyone outside their urban left-wing echo-chamber is still SOP. For now.

The worm turns: they don’t seem to have worked that out. I know it’s not the origin of the phrase, but I always have seen history as a sort of worm-drive, needing a full circle to shift on one gear tooth – but inexorable. These things will sweep our industry, change its landscape – and contrary to popular belief within it, the arts and entertainment world follow, rather than lead.

But then I thought even Cassandra needs a break, and I’d talk about the magic of geography in fantasy. No: Geography does not magically transform people… but it does shape people, and their fantasy. I was fascinated to be informed this week that C.S. Lewis’s fantasy Narnia was shaped by County Down, Northern Ireland, and its green landscapes. Being me, I of course wondered why there was not more rain in the books!

Geography has some of the disadvantages and advantages of the cliché and the stereotype character – You don’t have to ‘build’ too much of the landscape of your story in the head of your reader, if you use the standard geographical tropes. The reader may not ever have been in the mountains or deep woods, but they are ‘established’ in their heads.

Diana Wynne Jones in her ‘Tough guide to Fantasyland’

(The picture is a link)

(and used in her Dark Lord of Derkholm and its sequels) had a lot of fun with the sheer inane silliness of some of these tropes, which as often as not make no sense in real geography. Like the ‘stew’ which those on the quests eat, (despite stew being a rather slow, time consuming process, requiring ‘stewing’) on their nightly stops after their horses have covered amazing distances, these are often an escape clause for the lazy or ignorant writer.

What they should be – IMO, is a frame into which the writer builds. I must admit, however, that it is one of the few areas where ‘write about what you know’ can lead you into a lot of trouble, unless you’re careful.

You see, the reality is a lot of your audience… won’t know. Some of course will, and be very irritated if you get it wrong. But all of them will THINK they know – and will be just as irritated if they think you’ve got it wrong (even if you haven’t. It’s their idea, and their money – as a writer you have to appease the former or at least convince them you’re not wrong, if you want that money) Most of us have strong ideas about what geography is like, in our heads. Of course it would be simpler if it was all the same ideas… but even seemingly ordinary things like ‘mountains’ can mean very different thing to people who have been into the mountains – let alone the urbanite who has seen them on TV. For a simple example – the mountains of South Africa (where I was born), and the mountains of Australia, and the mountains of New Zealand are all very different in something really basic – slope. Australian mountains are old and to me quite rounded, South African Mountains middle-aged (with a little middle aged spread), and New Zealand mountains are slim and spiky youngsters with ridiculously (to me) steep sides. And all of the above are not the mountains of the Brooklyn Fantasy reader or writer.

Keeping this in mind we come back to using those ‘stereotypes’ landscapes without slipping into DWJ’s fantasyland stereotypes (It’s worth looking at if merely to know what to avoid), using the strengths of it, without the predictable tropes and boring the reader. This is a balance thing: I can’t make you instantly good at it – it’s one of the skills you learn as a writer. But I keep turning back to Tolkien and the concept of using those preconceived ideas as a frame, in which I shape the picture in the reader’s head with small touches of precise detail. Personally – because so many of these images are so visual in the reader’s head, I’ve found adding non-visual cues – the sounds and smells of a scene very evocative and powerful.

This was me adding to the image of a coastal storm in CHANGELING’S ISLAND

The wind flurry brought angry drops of rain hissing down the blue-gray wall of the surging swell. It roared up the ramp in a seething ravel of white water and rolling stones. The inky blackness across the water devoured the outer islands, and the horizon had vanished into the rain haze. Suddenly it was backlit by a tracery of jagged lightnings showing every black billow of the vast, stark roiling mountains of cloud above the whitecapped gray sea.


Filed under Uncategorized

54 responses to “The Geography of Fantasy

  1. Coincidentally I stumbled across some posts by Karen Myers about creating a fantasy geography using a software program to do all the hard bits

    Obviously this doesn’t apply to places which are closely modelled on places in this world but it might help for authors who need a more general world or part of a world

    • Fractal Terrain 3, the software she references in her blog appears to be using standard fractal approach – likely fractal Brownian motion (fBm) with a Perlin noise basis function – to generate the basic elevation map, then applying a color table to color the terrain at certain heights, with the sea being a blue applied to every spot on the map where the elevation is below a certain threshold. Judging from the blog post and the images, it looks like there’s also an erosion algorithm that can be applied. (You can look at 1989 paper “The Synthesis and Rendering of Eroded Fractal Terrains” if you’re interested in more on the underlying algorithms – there’s legitimate free copies of that article available online if you search for it.)

      Its a relatively powerful technique that is relatively simple to implement, but does not tend to generate geologically plausible topography on a global scale. It does produce pleasing shapes, and a nice approximations of land forms like Greece, the Balkans, etc. They tend to lack is large plains, well-defined ridge lines, parallel lines of mountains caused by folding, etc.

      Also apparently lacking from the software are rivers, though techniques for those are also known, such as from S. T. Teoh’s “River and Coastal Action in Automatic Terrain Generation”.

      • Draven

        You have to force it to make rivers, and it has no action for making the water flow to the sea naturally. It barely makes climate decisions (largely based on lattitude) and really, if you wanted to create , say, the desert southwest… you go paint in the desert.

        • Ugh. That’s marginally better than I thought, but not by much. Rivers and climate have been addressed in the comp. sci. literature and in online forums for nearly a decade; some solutions even date back a quarter century at this point.

          In recent years, just with rivers, there’s the 2008 Teoh paper I mentioned, there is Cordonnier et al. “Large Scale Terrain Generation from Uplift and Erosion” from 2016, 2013’s “Terrain generation using procedural models based on hydrology” by Genevaux et al, and the 2010 open-source online example in the form of Amit Patel’s Polygon Map Generation demo, which generates complex islands include rivers.

          Teoh and Patel both address to some extent some of those other issues you mention, such as wind and moisture levels and there climate impacts.

          • Draven

            The 3d software World Construction Set, which is primarily for animation, had a really good small-to-medium-scale terrain generator and an older version had a noce water flow system where yuo click a popint on the landscape and the water flows to the lowest point (i.e. the ocean) and would follow the natural path and you could even make it fill hollows, creating lakes and ponds.

      • snelson134

        Look up a program named “Campaign Cartographer”. It was a staple about 10 years ago which is the last time I was world building.

        • Or Fractal Mapper. Different company. They do have a fractal Terrain generator as well via their space program (Astrosynthesis) but it was intended as a ‘quick’ way to create sem-realistic looking planets when they needed to be done in bulk.

          • I’ve looked at both Fractal Mapper and Campaign Cartographer in the past, and they both seem like decent packages, but they do have their limitation. With Fractal Mapper, all the world-scale maps shown in their samples look obviously fractal, and IIRC a lot of the features like rivers require manually drawing. From what I recall of Campaign Cartographer, aside from the ability to import a base map, it can create the basic shapes for islands and continents with a little input from the user, but it doesn’t really use any geology to, just geometry.

            Don’t get me wrong, depending on the usage scenario, they both look they have the potential to be very useful tools. The GUI for them looks decent, too. But I don’t think they’re ideal for generating believable planets without investing a good bit of effort.

            • Not really. The package that comes with Astrosynthesis (Fractal World Explorer) is okay but not really all that great though my version’s a bit old. It was designed to make semi-believable-from-space style maps. I may need to see if they’ve updated it.

              I asked the company once a few years back if they planned a more detailed planet generator, and their basic response was “Not at this time” mostly because they didn’t have anyone on staff that familiar with the tectonics and geology and it would take a while to get there, and they weren’t sure they could simplify it down to something light weight enough to work inside Astrosynthesis. The complexity of geologic modeling is a problem likely to be hit just about anywhere that tries it. I’m not sure there’s a big enough market out there for someone to make a real stab at it.

              • There are a number of techniques that can be used to avoid full scale geological simulation, yet yield plausible results at the macro scale. Plausible renditions of mountain ranges, as well as river and coastal details (meandering, carving of valleys into terrain, beaches, headlands, etc.), can all be generated easily enough with knowledge of plate boundaries, whether each plate is continental or oceanic, force vectors for each plate, and global sea level.

                Alas, none I’ve seen will also know where it should generate oddities. For example, the extensional terrain of the Basin and Range Province, the rolling Sand Hills of Nebraska, the mostly-parallel ridges of the Appalachians, East Coast barrier islands, the Frisian islands, and sandy deserts with dunes. A patch of terrain that looks like the Himalaya, the Rockies, or the Alps where two continents collide, no problem! These other features… maybe a random placement would do? The key problem is knowing what needs to get placed where. There are techniques for generating just about any type of terrain, once you know where it needs to go. There’s even example-based techniques (e.g. “Terrain Synthesis from Digital Elevation Models”) that can generate a canyon like the Grand Canyon along a defined path, or mountains like the Grand Tetons.

                Another thought that just came to me: latitude-based ruggedness of continental coasts. Due to glacial effects, the more-polar regions tend to have more rugged coastlines with greater degree of inlets, islets, fjords, etc.
                That probably could be simulated by generating the basic land masses, including mountains, then subtracting from the elevation a value based upon Perlin noise, with a weighting factor based upon latitude that scales up from nothing near the tropics to a greater number near the poles. Maybe.

                So what’s really needed is to get the basic land forms right, then know how to add the right details in the right places.

                • Draven

                  well tbh, most of these types of terrain generators are based off of really old algorithms and were written to where a high res map would have likely taken them hours…. they really weren’t written for modern computation ability.

                  (Remember, your horribad slow computer is still 60x faster than what was used to do the first season of B5.)

                  • That was one of the major advantages to the fractal-based elevation, in that you don’t need to calculate and store very much initially, and can calculate more in local areas if you do zoom in. The downsides are poor hydrology models* and poor feature distribution. But even with modern techniques I mentioned above it is impractical to generate and store high-resolution data for entire worlds. I think compressed vector data representations of feature center lines (ridge lines, water courses, etc.) combined with a short set of parameters to allow procedural generation as required may be part of a possible solution, but I’ve not seen anything yet that handles that at a global scale. Global scale still generally means a fractal solution, on-demand procedural content generation**, or low resolution data.

                    * James Bardeen, physicist and son of two-time Nobel laureate John Bardeen, spent a lot of time and energy trying to get a fractal-compatible hydrology that sort of worked. From what I saw, it was at best OK. Others have small-scale techniques that work.

                    ** Reportedly the game “No Man’s Sky” took the on-demand procedural content generation route. I’m not clear how it deals with hydrology.

                    • Draven

                      yep. WCS used fractal subdivision on DEMs to artificially create more detail.

                      And NMS didn’t handle hydrology. I can’t recall ever seeing a river in someone’s playthrough that was going from a high point to a low point.

                  • Most are also based on a very limited understanding of plate tectonics and geology in general. I’ll be blunt. The algorithms are useless if you don’t understand what you’re modeling, worse if they don’t understand how the different pieces need to interact. Add to that a lot of these tools are being produced by companies that are a couple of guys and a registered DBA and unless someone has a fair amount of interest or pre-existing knowledge there will be rough spots. (Astrosynthesis as an example: They had someone very interested in the astronomy aspects so those are reasonably solid, but it is slanted towards ‘making a space RPG in a hurry’ so the world maps were ‘good enough for looking at from space’)

  2. Luke

    Ah, but as a rock climber, those ridiculously steep slopes and unweathered rock surely have a great deal of potential. 🙂

    OK, about the only (possibly) useful thing I have to say is that stereotypes exist for a reason. I don’t believe the problem is the stereotype per se, but the employment of the stereotype in a predictable and boring way. (Or in a way that’s nonsensical without being deliberately funny.)
    And sometimes geographical tropes are much better known from fiction than reality. The number of Americans who have experienced triple canopy jungle might number in the tens of thousands, but tens of millions have familiarity with the “dark woods” of Oz and Middle Earth. If you’re going to journey between the boles, drawing from the common well is a good idea. Just don’t leave it like that, because well, water is pretty dull.

    • Stereotypes let you borrow other people’s work. You can sketch in a side character in a couple of sentences. Everybody Knows that thieves know all the rooftops and can climb anything, because of the Grey Mouser.

      Not very many people are rock climbers, so not too many know just how hard it is to get past the eaves of a house, or climb wattle and daub construction without putting a foot through the wall. Or how easy it can be to get up a limestone cliff.

      • Luke

        The author of the Original Post is a rock climber. (Among many other things.) 😉 I was giving him a bit of cheerful teasing.

        But yes, someone with the background to do so convincingly could make a great deal of hay with those details, and the stereotype becomes a building block.

      • heh. for certain values of limestone cliff. I had actually not climbed limestone – until last year April, when my son and daughter took me to climb on Portland head (UK). It was… hard. What I really hadn’t realized before this trip was just how heat-leaching limestone. By 20 feet up my hands were numb. I don’t know the physics/geology of it, but we climbed granite, slate, and limestone. The granite just didn’t freeze your hands the way slate and worst, limestone did. I used to fancy slate floors and limestone walls before this!

    • ‘The number of Americans who have experienced triple canopy jungle might number in the tens of thousands, but tens of millions have familiarity with the “dark woods” of Oz and Middle Earth.’ – exactly. And if you’re one of the tens of thousands who know they can be lighter and easier to walk than one thinks – well make sure you do have them throw your book out, because reality doesn’t match their belief of reality.

  3. “Like the ‘stew’ which those on the quests eat…”

    My medieval campers got Tactical Bacon and Cross and Blackwell’s canned date nut bread. They correctly assumed that everyone in the Modern World lived like a king, but they didn’t believe that the poor were fat. Said they’d need to see that, it was just too much.

    Oat or millet porridge with jerky in it is a lot more likely. Assuming they had time to make venison jerky. Rabbits on a spit would be possible too.

    • Luke

      Hey, now.
      Stew is the perfect thing to eat in the tavern. It tastes pretty good, it won’t give you dysentery, and it’s hard to examine too closely.
      Just don’t drink the water. It’s got evil spirits in it.

    • If you’re realistic about the speed/distance/time your riding or draft beasts can do in a day, stew is perfectly doable. I was shocked–yeah me, the horse fanatic–the first time I researched the old cattle drives from ranch to livestock market at the rail head hundreds of miles away. Ten miles a day, on average. Those steers grazed and grew and fattened up all the way. The chuck wagon could go ahead and set up the next night’s camp by water and start cooking those beans.

      • I always used – as an example of the distance possible to travel across country without roads -Dick King’s ride to from Durban to Grahamstown – 600 miles across country largely only populated by African tribes and full of bush and rivers to ford in 10 days as ‘about how fast it is possible’. I’d doubt if much more than dried meat and some form of biscuit could have been used most of the time for that sort of trip. Yet 60 miles is a short trip now…

      • Yup. In the long-trail drives, the cattle moved fairly slowly and leisurely – the object was to deliver them to the cattle buyers at the other end fat, healthy and fine-looking. So they moved much slower than the chuck-wagon, which was usually a light spring-wagon, drawn by horses or mules over fairly level country. The chuck-wagon would move out ahead of the herd, set up at a location chosen by the trail boss for water, wood (if possible) and a good bedding ground for the night – and set to work on the evening meal, while the cattle herd caught up to them. Time to build up a good fire, get stuff started cooking – and it would all be ready for the drovers by the time they arrived with the herd and settled them for the night.
        One of the shortcuts was for the cook to set a pot of dried beans to soak overnight in a Dutch oven, in the embers of the cookfire. The next day, those beans would be near-to cooked, and ready for the next evening meal.

        • I was intrigued to hear about an Oregon Trail trick, which works if your milk cows are still giving. You get your little barrel of cream and lash it to one side of the wagon. The motion of driving all day churned it pretty well, and that night, you’d have butter.

          Most people who did the Oregon Trail walked the whole way if they didn’t have a horse. The wagon ride was not comfortable…

          • Yep, it was on foot, just about all the way, as riding in a wagon with no suspension was brutally uncomfortable. And besides – the wagons were almost always packed to overflowing with equipment and supplies.

            The people who took milk cows usually only had milk from them for about half the way. Some people managed to have eggs, too – packed in a barrel of sawdust or flour.

      • Mary

        cooking your own stew is impractical, often. OTOH, the inn’s probably got a pot of stew bubbling. You can throw anything in and dish it out whenever you have hungry travelers. Mind you, it’s not going to be something so recognizable as, say, beef stew.

  4. I’ve personally experienced a wide range of geography in terms of the US West Coast states as well as several further inland. And by “experienced”, I mean “actually tramped around on or camped in.” It’s a lot easier to describe something when you’ve experienced it yourself.

    Of course, I am not entirely joking when I say that when I do world-building, I start with plate tectonics. I grew up with a paleontology-obsessed parental unit, and geology came as a side effect, with the end result that I not only have a more-than-casual interest in geology and its effects, but a base of knowledge that gives me an instant “that doesn’t work” sense for certain configurations. However, the easiest way to get your geology right is to model it directly on something. You don’t have to be too specific, but if you know that thus-and-so mountain range is based on the Sierra Nevadas, you can get the basic geography and flora right, and then you remember to have a proper rain shadow.

    • I’m using France. Google Earth is really handy.

      Remove population, furnish with medieval ruins, add demons and stir briskly.

    • Yes. I set something on a temperate environment in the southern hemisphere of a planet, and had to remember that the weather arrives from the east*, not the west – most of the time. When it comes from the west, it means Not Good Things.

      * Assuming same rotational direction as Earth.

      • But that’s not a hemispheric thing. Prevailing westerlies in northern AND southern temperate zones, trade winds (easterlies) in tropics, easterlies in polar regions.

        • As for ‘assuming same rotational direction’: East is defined as the direction of sunrise. (The English/Germanic word derives from a root meaning ‘dawn’; the Latin oriens comes from a root that means ‘rising’.)

      • er. Our rain is principally Westerly :-). It’s also about latitude.

    • If you’re interested in worlds built from plate tectonic models, there are a couple of simulation programs out there that do this. I blog on some of this stuff on occasion, but for brevity here are two of my favorites.

      The Experilous Planet Generator is a JavaScript-based web application that generates a stochastic plate layout with plate motion vectors, and based upon sum of forces at plate boundaries generates mountains and islands. It can display both a globe and a map view, and generates other details like moisture, temperature, winds, etc. (It seems to draw inspiration from the Flash-based island generator created by Amit Patel of Red Blob Games, which generates islands.)

      The PlaTec simulator is a slower but more accurate simulation. Alas, it operates on a wrapping 2D grid basis rather than a spherical model. It is a traditional computer application. IIRC there isn’t an executable, and it was designed for a Unix/X environment; it took me about 2-3 hours to take the existing code and port it to run on Windows using the cross-platform FLTK GUI. (Come to think of it, perhaps I should polish that up and put it on GitHub when I have some spare time.)

    • Modeling things on the Rockies can get tricky. They are the mountain chain that had something of a sordid and checkered past.

      I tend to start most of my maps with tectonics in mind, though not necessarily starting there. I tend to have particular features I need for the story then reverse Engineer the geology.

      • Geologic history is also fun. You could pick a world of, say, 85 million years ago, and have everything all mapped out and ready to go, yet not recognizable. (Of all the issues in the movie Armageddon, that’s the only one that really annoyed me. Modeling the dinosaur-ending meteor impact with modern geography? And they thought people wouldn’t notice? Arrgh.)

  5. *mildly disappointed* I thought you’d made up a new name for that fantasy stuff that, as best can be told, is written by folks who hate the very concept of fantasy. Antasy.

    Example: bridge to teribithia.