06 February 2018

Rationale behind Methodology of my Thesis

This post is an extended reply to Thorfinn Tait’s recent comments and the discussion that has taken place in the Mystara Cartographic Society Facebook group.

First off, I want to say that I haven’t completely abandoned the project; it’s just on the back burner right now. I wasn’t satisfied with how the DEM for Karameikos turned out; there was too much irregularity with the dimpling and I didn’t like how the mountains plateaued, but my biggest issue is interpreting the b10 map. I can’t decide which lines should be ridges or valleys. Should it resemble something like the Appalachians or the Sierras? So, I’m just giving it time to stew in the back of my head and been working on other projects (which admittedly have not made it to this blog)

Many of  Thorfinn’s thoughts bear similar threads, so I’m going to go through talk and about general topics instead of addressing each comment directly.

I want to start by talking about assumptions and providing a basis for creating an accurate map. When making any map the first object of the cartographer is to find Data. This may come in the form of a spreadsheet or descriptive text, or old paper maps that have to be scanned, or electronic data designed for a GIS. However old or obscure the data is it has some relationship with a real geographic location on earth that can be given latitudinal and longitudinal coordinates. When working with most fantasy or science fiction worlds this isn’t true. A location on Tattooine cannot be geographically mapped because it is not on earth. Geography literally means writing or descriptions about Earth. Geography technically applies only to Earth; the geography of Mars for example is called areography. This is what I mean when I say that previous maps of Mystara are geographically inaccurate. Maps of Mystara can only be geographically accurate if using earth as a basis.

This basic premise that Mystara is Earth underlies my whole project. I have read all of Thorfinn’s work that I could find and love his maps and his whole Atlas of Mystara project. My own meager efforts would be impossible without all the work he has already done. The reason I didn’t incorporate more of his ideas and those of other members of the Mystara community was because I was creating a project for university and had decided to stick with TSR published sources as much as possible. I understand why Thorfinn has approached the Atlas the way he has, but decided to start from a different premise precisely because I have seen how he has developed his maps over the years.

As far as false assumptions, I probably could have made my point with a little more tact. My main issue with all the maps of Mystara has been the assumption of a specific projection without any evidence. This includes the weird “Robinson” projection provided in the hollow earth set and the equirectangular grid on the maps of Thorfinn and a few others. I remember reading on Dragonsfoot somewhere Frank Mentzer was asked what projection the original maps were in and he didn't remember but guessed it was equirectangular. But that didn’t definitively confirm anything. There's no reason to assume an equirectangular projection. If anything, the Mercator should be assumed as it is much more common in general purpose maps despite its inherent distortions. Without knowing what the masters set was based on, there is no reason to assume one projection over another.

I want to address a few things Thorfinn brings up to argue for an assumption of an equirectangular map; these include the compass on the master set, a rectangular border and the assumption that north is at the top of the map.

The presence of a Compass or North Arrow never implies a specific projection. Many maps will have a North Arrow even though they are not using a Mercator, equirectangular, or similar projection. I have even seen some globes with a north arrow, which is completely unnecessary. The common assumption that north is at the top of the map is the main reason for a common principle in modern cartography; namely, that a north arrow should not be on the map unless north is not the top of the map.

A rectangular border is only present in the Masters set because the page is rectangular and doesn't really say anything about the map itself. Every map I made for school had a neat line, it differentiates between the map and anything else on the page such as art work or text. The border on the Masters map serves the same purpose.

Let’s talk about projections. I am completely certain that the original Jurassic map, and therefore the known world of the master set, was in the Mollweide projection for 2 main reasons. First, all paleogeographic maps of a significant age that I have seen are in the Mollweide projection. Only maps made recently have had alternate projections, such as Plate Caree or Mercator. Second, I tried to match the Master's set to a reprojection of the Jurassic Map in an equirectangular projection. It doesn't match up at all. If Mentzer did a trace of a map, it was one that was already in the Mollweide projection. Also, the Jurassic map I used is labeled as being in the Mollweide projection, any good map will have such a label somewhere.

The lack of a scale or coordinate system on the master set map is irrelevant because it will assume those of the Jurassic map. This means that the scale of the hex maps must be subsumed by the Jurassic map. This results in each side of each hex being a different length. As much as I want to stay true to the first maps of BX and x1 they give no coordinates and so the georeferenced master set map takes precedence. Hex maps provide a basis for understanding the terrain and places but can't be taken at their given scale.

This doesn't mean the hex maps are “wrong”, only that the scale is off. They still provide plenty of geographical data.  Because the Masters map was traced from a map already in the Mollweide projection the hexes can be laid directly on top of the Masters map without distortion. Each hex will still contain the same square mileage as every other hex because Mollweide is an equal area projection. The hex scale will be off mainly because the scale of the Jurassic map wasn’t accounted for in the first place.

“Very important question here: is the map you're working from here the map upon which Mystara is based?”

It's impossible to know without direct confirmation from Francois or Frank, but if it isn't then it is one almost identical. As shown in my first post Frank Mentzer and Francois Froideval were at least considering the idea of using a map of Earth in the Jurassic period. Therefore, they had to have been using some reference map published as early as 82. The specific map I used can be found here. Christopher Scotese (one of the authors) published several different papers around the same time period that used an identical map. I was unable to find anything else published around the same time without Scotese's name attached; which doesn't mean there wasn't anything else that may have existed at some point, but considering how well the master's set lines up with what I used, I think one of Scotese's maps was the basis the Known World map. I am of the opinion that Mentzer did make some changes to the map as he stated on Dragonsfoot, specifically to fit the map as presented in the Isle of Dread into the Known World. I think Cook created his map without any plans to expand beyond what was presented, and when Mentzer took over he shoehorned Cook's map into a map of the Jurassic period which became the Known World.

As put so succinctly by Thorfinn, the key difference between our approaches is my assumption that Mystara is ancient earth. the only info we have for a basis of real coordinates of the earliest map is the statement in the books and by Frank that the known world is earth in “the age of magic”. I do hold the Masters map as having more validity as a model for what was originally intended for the known world. I may have some bias as I didn't discover the Mystara fan community until high school and only had the BECMI box sets and a few modules growing up, so I always knew it as the known world and not Mystara.

Regarding intents of the various designers and how the maps have been interpreted over the years: I am specifically trying to stay true to the intents of the original designers; those who created the setting and not later designers. Though I will use the work of later designers if it doesn't contradict the original vision of the setting. interpretations are just that, only interpretations, I feel no need to be beholden to those. giving the readers interpretations equal value to the original designers is like saying we should take into account the interpretations of the readers that LOTR is about the world wars when Tolkien repeatedly said it wasn't, though he of course may have been influenced by the time in which he lived. I am trying to account for intent of original designers, not readers who later became designers and interpreted the works of the original designers and threw so much of their work out the window(the first episode of the Voyage of Princess Ark).

I don't really strictly follow either camp concerning what is or isn't Canon. I don't adhere to the idea that later works should supersede earlier, or that less information is better and shouldn't be expanded. I actually prefer greater detail. As far as sticking to older sources, that has to do with the history of development of Mystara. With real world maps, newer generally trumps old because we have more accurate data derived from better methods of measurement and more refined Cartographic techniques. However, with Mystara we have the original maps made by cook, which provide no info for placement besides climate/culture; it was probably meant to span an area from Denmark to Italy regardless of what the scale of the hexes were in the end. Then Mentzer gives us a revised expert map and later the continental and world maps of BECMI. He really only added to the earlier work of cook. But then Bruce Heard comes along with the gazetteer series and voyage of princess ark and outright states that all of Mentzer’s work is wrong and inserts a new reality to the game. I appreciate Bruce's immense contribution and work and creative genius, but I'm not going to put his work above the original work of cook and Mentzer. So, my methods are to include everything from cook all the way to the 2e stuff as long as it doesn't contradict something published earlier or by an original designer. so, for example when gaz 1 changes the pop of specularum to 50000, I stick with the earlier number of 5000; and I give precedence to anything in cm4 as opposed to cm1-3(I’m not sure if there’s any actual contradiction between those products).

My methods of precedence of large vs small scale vary based on context. given that the map upon which everything is based is one of the entire world, the coordinates for the placement of any large-scale maps are predetermined. However, for specifics, such as differences in the coastline, I generally side with larger scale maps. Though I may make exceptions, giving the smaller scale maps precedence based on publication date.

Let's talk about scaling and overlaying and georeferencing. so, I did 2 instances of georeferencing, one with many control points requiring a 2nd or 3rd order polynomial transformation resulting in warping. this inevitably leads to the hex maps not matching the world or continent map even though there is greater fidelity to the Jurassic map. the method I used in the end was with two control point using a 1st order polynomial. I did this to maintain the integrity of the image as a rectangle even if it was rotated to some degree in order to best fit the Jurassic map. the reason larger scale map might not align perfectly with the smaller scale lines on the continental map is due to the variance in exact placement of control points. which pixel on a hex should match up with which pixel in the borders provided by the continental map? there is always a degree of error when georeferencing any raster, the error was only compounded by the fact that there was no vector data to match up with the raster, and instead the raster had to fit into another raster image. the difference in scale itself and how much area each pixel covers creates the error rather than any warping from georeferencing. In order to have each reference map a different layer they had to be individually georeferenced, I couldn't overlay them in illustrator and then georeferenced it because that results in single layer and single image.

As Thorfinn says the maps should overlay perfectly. Which they do as long as you only take one generation of maps at once, e.g. x1 x4-6 and cm1, but if you start matching up x10 or 11 or the dawn of the emperor's maps to earlier maps then there are problems. This is why Thorfinn has been working on his Lining up Mystara series for so long, not all of the hex maps line up like they should. Placing the early hex maps directly on to the continental map, everything fits fine as long as you accept that the hexes don't directly reflect the projection of the world map. Georeferencing itself won't cause warping unless higher order polynomials are used.

 In establishing coordinates for the hex maps, it is necessary to place the companion map because it is the only map which provides specific guidelines for lining up the hex maps onto the world map, without it I would be basing it on coastlines which vary between the companion and master sets, and  don't match up with the coastlines of the hex maps. using the map of Brun requires a slight rotation, meaning the hex maps get rotated with it and are not directly aligned along latitudinal lines. this rotation of the companion set is unavoidable. It doesn't line up with the Masters set without rotation. I've tried with illustrator, gimp and through georeferencing with control points, and in no case does it line up without rotation (If anyone can get it to line up perfectly I’d love to see the results).

With the coordinates for each map established and the hex maps lined up it is important to keep in mind that the hexes won’t keep their original scale. around Karameikos one 6-mile hex is generally only 4 miles north to south. In my mind this just confirms my theory that Mentzer shoehorned the isle of dread map into the world map and didn't account for scale. Also, the location of features in any given hex may not be in the center of the hex, but this is hard to quantify if we are only shown one symbol per hex it is impossible to know exactly where the location is in the hex without reference to a different map with larger scale hexes or to a description in the text.

“Would you be willing to share your DEM files”

 yes, the file is a geotiff. Here are links to both the base dem and a hill shade derived from it. As stated in the beginning of this post I will probably be changing it to better reflect how I think the landscape should flow in Karameikos.

I’ve been contemplating using the google api going forward for the web map in order to give the option of viewing from a global perspective, and because most people are familiar with the google earth interface already.

It’s been my plan since I first tried my hand at this a few years ago before I began my university thesis to create separate versions of Mystara for each publishing era: BX, BECMI, gaz, and post woi.

P.S. Since Thorf found the Jurassic map in that was mostly the map Mystara was based on (though we won’t know for sure until either Frank or Francois confirm it), I will base my future work on that and realign everything to it, there shouldn’t be any big changes, just slight adjustments.