Mapping a Fantasy World - Challenges

2018-09-25
,

When I finished my first Flanaess map a couple of years ago almost immediately decided to do it all over again. The main reason for this was the advancement in technology since I started my fantasy mapping career in the late 1990's. Using Bryce and an early versions of CorelDraw and CorelPaint, on my first generation Pentium equipped PC running Windows 95. A lot of the techniques and design decisions I made back then have been with my Flanaess project up until now and still affects much of  the looks and and features of the 2018 Edition. 

Back then I was happy just getting something on the screen that looked decent, and there where some many technical hurdles that I didn't know if if was even technically possible to create a map covering the whole Flanaess. I made a big gamble and hoped that Moore's Law would work well enough for me to keep going, and  I made it with capacity to spare! So back in the 90's it was RAM and CPU power that was the main limiting factors in trying to create a top down map using limited 3D terrain tools in the form of Bryce. What are the hurdles to overcome this time around twenty years later?

First let me quickly state what I'm trying to achieve. This time around I want to create a realistically textured 3D elevation model of the whole continent at a resolution comparable to what you see in google maps or can download from the USGS in the form of DEM's. The data needs to be geo-tagged so it can be integrated with other data for a whole range of possibilities. 

On the hardware side of things not much to worry about, CPU's GPU's, RAM and storage of today can handle this and with the new cool toys on the near horizon no clouds on the hardware skies. The metal and machinery to handle this taks are already near my desk or will be so in the very near future. When it comes to software this is where the clouds start to appear. Windows kinda cam of age when it got 64-bits to play with, and Windows 10-64bit works like a charm for me. 

Now we comes to the the applications needed to try and create the terrain model, and this is where the limitations start to darken the fantasy skies. I use the following applications for terrain creation: World Machine, Vue, Terragen. They are in my experience the best and in may ways the only tools that can really be called professional. There are a number of new ones on the horizon and I have tested a number of them, and the only ones of real interest for the kind of heavy high quality lifting I'm after are Gaea and Instant Terra. The guys and gals over at Wysilab seems to have shoot themselves in the foot though, pricing themselves out of the game. Instant Terra Professional subscription costs $1,560/year, and that caps out at 8K render resolution. I'm not even going to ask them how much the Enterprise cost. The beta version I have tested didn't impress me much except in the use of GPU based rendering which made things more or less instantaneous on my machines. So I leave Instant Terra to brood over in France for the foreseeable future. 

I've mapped many thousands of square miles using these tools the past five or so years, which have made me impressed with some of their features and deeply frustrated at some of the limitations. First it needs to be said that none of the tools, except for Gaea which I will come to later, where made with the intent of accurately modeling a large fantasy world in this level of detail. All the tools seem to have one of a few of the features needed but always seem to lack in at least one critical area.  I can go on about my frustrations with this for a very long time, but the Flanaess is waiting so I will try and nail down the key problems this undertaking will face going forward. Then in a series of blog posts for the Cartographer backers I will try and explain in detail how I intend to go about tackling these challenges. 

Spherical Worlds and Projections

This is in many ways the most fundamental of the challenges when trying to map even a modest bit of a planet. The Flanaess are big enough in itself to give you a lot of issues, but it is like Europe best describes as number of peninsulas belonging to a large continent. In the case of the Flanaess it is the super sized continent of Oerik, roughly the size of the Pacific ocean and then some spanning more than half the planet.

Both Vue and Terragen can work with whole planet, and they even have fantastic procedural tools to do so. The problem with bot of them is a lack of control, and when you have a fantasy world already decked out with features at certain spots both tools are inadequate for the job. World Machine does a decent job covering large areas and Gaea will soon improve on that even more with better tiling features but they still only work on a flat plane. 

The way around this problem is to use some of the same tricks as the real world land surveyors and cartographers use but in a virtual form. So it is time to start forge those survey markers and bring forth the calculator to find the right projections for the job.  

Too large to file

Terrain elevation models are best stored as images or text data in the form of DEM files, the difference is one is binary and the other is ascii, but both of them will grow too large when they cover more than a few tens of square miles. The Flanaess is somewhere in the vicinity of 6,000 miles across (depending on how far west you include). This means you need to split it up into hundreds of small areas to make the task manageable. 

To split it up is a good thing for many reasons. Information storage and handling of course, but a major benefit is to be able to work in parallel. Different areas can be worked on by different people and then the result can be merged afterwards, which will require a robust set of standards. File formats, resolution, naming, meta data etc, all need to be worked out in advance and adhered to. 

With careful planning you will be able to use different tools both for added variation, but also to not be stuck with the same tools for years. For consistency I stuck with Bryce 5.5 for a decade, newer versions didn't work well with the humongous file I had created and the results didn't match well enough. This time around I don't want to fall into the same trap, so data need to be stored in formats that can be useful by different tools now and in the future.

Shades of Grey

Elevation data are usually stored as grey-scaled images, so the number of different elevation levels are limited to how many shades of grey there are. This is measured in the numbers of bits for each pixel, 8-bits, 16-bits and 32-bits are the options here. 8-bit are not enough, unless you can live with a Mindcraft looking terrain and 32-bits tend to be too much for a lot of applications to handle, so 16-bits seems to be the most resonable option here. 

16-bits gives enough detail for terrains with an elevation span of up to around 10,000 ft, try and push it harder and things have a tendency to look stretch and featureless. I imagine fantasy worlds to have mountains (and various holes) to span much more than that and the Flanaess is no exception. If we intend to map the rest of Oerth even bigger things might be out there who knows!? So we need to be able to span elevations, up and down, that are many times that. I imagine the Crystalmists and the Yatils to be taller than the Himalayas and the Right Canyon will need some depth to play with indeed.

This will force us to look into things like vertical tiling, various base levels and other tricks to make this possible. My first Patreon Project the Crater was a test of the possibilities with an elevation span of way more than 10,000 ft. So it can be done, but it needs to be combined with projections and tiling and the final issue below. 

Give me a Creek

Water seems to be either very difficult to simulate or seen as a nuisance by the creators of terrain creation tools, especially small amounts of it. Oceans and large bodies of water are well catered for in both Vue and Terragen and they can both create spectacular looking water in the renders produce. The problems comes when water flows downhill and that is the norm even in most fantasy worlds. Vue and Terragen have no functions for rivers, none which is to me a mystery. When you spend years trying to to simulate nature, but skip trying to even implement the most basic forms of rivers large and small. 

World Machine have what I would best describe as an embryo for creating rivers, which is clunky and requires a lot of minute fiddling for at best mediocre results. It is only good for creating larger rivers and not the myriad of creeks needed to make wetter areas believable. I have worked on a way to create rivers using the Layout Generator in WM, and after about three years and too many iterations to try and remember it did a better job than the native river tool. But it is still too clunky and then there are those creeks....

Gaea is where I put my money when it comes to first come with a river tool that can effectively do the job. Eventually I'm sure all of the tools mentioned will come with robust river functions, including even those pesky little creeks, but that will take time. Gaea isn't ready for production yet and its river functions months away still. So an interim solution is needed that can do the job for the foreseeable future. 

My answer to this is to use the always trusty Photoshop, and paint the rivers. This is nothing new I've been doing it since the early days with Bryce and CorelPhotoPaint, but this time we need to do it in 3D! How to get those creeks and rivers to flow downhill in a natural way where we want them to will require some clever tricks, that I will try and present to the Cartographer backers in the not too distant future.

Onward... 

There are the four main challenges I see going forward with mapping the Flanaess (and potentially the whole Oerth) in detail. My hit and miss test runs the pass couple of years, combined with my Midgard work for Kobold Press have given me the experience (and stubbornness) to think that now I'm finally ready to start this journey all over again.

The next steps will be to sketch out the location of the Flanaess to get the big picture right, and also to work on Lendore Isle as a way to start get the details right. Thanks to your support I will very soon have pleasure to invest in a full set of professional GIS tools that integrate with Illustrator and Photoshop. This will both be a cornerstone in getting this in the right place, but also open up a whole new world of possibilities. 

This was a look into the future, next up will be a new Flanaess 2018 pre-release version, Photoshop videos and more Campaign Sketches, all coming in the next few weeks.   

Thank you again for your support! 🙂