Home Store Products Research Design Strategy Support News

5 May, 2008

 Products The Guns of Gettysburg Design Diary What a Relief

The Guns of Gettysburg Design Diary: What a Relief!

As previously mentioned, the design diary will be taking up map design as the topic over the next several entries. The mapboard, both conceptually and in the actual chronology of its creation, is built up in layers: relief (elevation), hyrdrology, vegetation, roads, buildings, regulatory graphics (which from a rules perspective is the only part of the map that is actually meaningful), and play aids. The general plan for the diary is to devote an entry to each, although it is possible that I will find that I need more than one entry for some and less than one for others.

This entry will concern itself with the first and lowest layer: relief.

From a research point of view, relief is generally the easiest layer to deal with. Reasonably detailed relief maps exist for most of the world. To be sure, most of these are for the twentieth century, but relief generally changes only very slowly (indeed, at a pace that is,well, geological), so a century or two difference (or five or ten for that matter) between the source material and the event usually makes little difference. Barring the odd volcanic eruption or open-pit mine (in the case of Gettysburg, the latter is an actual problem as a rock quarry started at some point in the last half-century puts a literal hole in the usefulness of the newest maps), the designer can be indifferent to chronological problems and can use new maps with confidence. In fact, modern maps are generally better than older maps (even where smaller-scale) as modern techniques for measuring relief are superior to those used in the past. Here are some maps for comparison: a 1:12,000 1868 map, a 1:50,000 1904 map, a 1:25,000 1946 map, and a 1:25,000 1999 map. Notice that the 1999 and 1946 map are in general agreement on the shape of the 500 foot line to the west of Willoughby Run, which the 1868 and 1908 map don’t quite catch, even though the 1868 map shows the most contour lines of any of the maps:

1868 (1:12,000)
1908 (1:50,000)
1946 (1:25,000)
1999 (1:25,000)

While it is tempting to get buried in the relief differences between different sources like this (differences that become even more visible when they are overlaid on top of each other as opposed to being laid side-by-side) the fact is that all of the above maps are high-quality sources and although some of the differences might have borderline military significance in reality, the differences are all below what wargames can easily catch. The fact that in the above illustrations you can see differences between where the 500 foot line runs is in part because the terrain is quite flat there; what one map is showing as 500 feet, the others might in fact be showing as 498 feet or 502 feet, and wargames in practice have trouble making use of even one-tenth of that resolution.

The reason for this becomes more apparent if we consider the main uses that wargames make of elevation information: movement and line of sight effects.

Movement effects are generally measured in wargames by assessing different movement costs for different types of terrain: a clear terrain hex, for example, might have a movement cost of “1”, while more difficult terrain would have costs like “2” or “3”. Because wargames almost never assess costs in anything other than an integer multiple of the easiest terrain for movement, terrain that might slow units down by 5% or 10% or 25% simply can’t be handled: no hexes have a movement cost of “1.17”. And so, with regard to movement, sloped terrain can halve movement or cut it by two-thirds, but nothing finer grained than that is possible, making concerns about an elevation difference of 2 feet over a distance of 100 yards a completely irrelevant question.

Line of sight effects are generally just as crudely handled as movement in most wargames. Prior to the mid-seventies, wargames generally made no attempt to make use of actual elevations at all, contenting themselves with at most classifying some hexes as “hill” and others “level ground” with perhaps slope hexes separating them. This practice by no mean stopped in the mid-seventies either, nor is it confined to simple games. Even high-complexity modern games like Advanced Squad Leader series, which have never seen a detail about some piece of hardware that they didn’t feel obliged to capture in some rule or table, can get very shy about line of sight and elevation issues and include no more elevation information than many of the earliest and simplest wargames did.

While it may seem that I am being critical of games that keep elevation very simple, it actually can be a very prudent choice. Rules that force players to calculate whether one unit at 480 meters elevation in trying to see another 1560 meters away at 420 meter elevation is blocked by a 435 meter elevation at 1060 meters distance are not by any means doing the players a favor. After all, while there are many players who enjoy learning obscure facts about military equipment (such as the field of fire of some tank’s hull machine gun), the number who enjoy doing complex line of sight calculations or enjoy waiting while their opponent does them is vanishingly small (in fact, it is entirely possible that it is zero). Thus, while from a simulation perspective the extremely fine grain of hardware detail in game like ASL and extremely coarse grain of elevation detail may be odd, from a game perspective it is perfectly reasonable.

The crude elevation rules common among even high-complexity simulations is interestingly paralleled by the state of cartography prior to the late nineteenth century. Before that time, detailed elevation information was very seldom included on maps. Rather than indicating elevation with contour lines tied to specific elevations, sloped ground was indicated by stroked lines called hachures, which were drawn in parallel to the slope, from high ground to low. This map-making convention was and is well-suited to wargames that have very little elevation information, and in fact was seen on the very first Avalon Hill wargame, Tactics II. By the mid-seventies Redmond Simonsen had created a variant of this for SPI that appeared in literally scores of their games. When well executed, hachures can be very effective graphically even today, as can be seen in the striking maps Rick Barber has produced for games like La Bataille d'Orthez.

1850 (from Alison’s History of Europe)
1958 (Avalon Hill’s Tactics II)
Quarry Map
1975 (SPI’s Marengo)
Quarry Photo
1999 (COA’s La Bataille d'Orthez)

It is hard to know for sure which wargame first made use of actual elevation information in the game board, but the first that I know of was Frank Davis’ Wellington’s Victory. Davis was developing his game at the same time that Richard Berg was working on Terrible Swift Sword. Berg was working on a more sophisticated version of the hachure method in use at SPI, with two different types of hachures: slopes and ridges, represented by different colors. Davis was considering using the same system but found it difficult to understand (a complaint that many players would echo after TSS was published) and so decided instead to go with actual elevations indicated with relief shading (relief shading is a contour variant in which the different elevations are color-coded to make them easier to see). Sections of the maps for TSS and WV are shown below:

Terrible Swift Sword
Terrible Swift Sword
Wellington’s Victory
Wellington’s Victory

While it is probably true that the relief shading method used in the WV map above is easier to visually comprehend than the hachure method used in TSS, the actual line of sight rules for WV were quite complex and a full page of the rules was devoted to explaining them, and it isn’t actually clear whether there was a gain in either playability or realism compared to the method used in TSS. Nevertheless, in the years since the publication of WV relief shading has become a common method of representing elevation in wargames, sometimes combined with hachures (the maps of later editions of TSS in fact were re-done from the original version to use such a mix) and sometimes not.

There is, however a substantial problem with relief shading, and that is color space congestion. Even though the image of the Wellington’s Victory map above shows only five distinct elevations, the highest is quite dark. Five different elevation levels is not a lot; while the battlefield at Waterloo was quite flat and so five levels worked tolerably well, five levels of elevation could give only a very coarse representation of battlefields with substantially more variation in elevation. The difference between the highest and lowest points on the battlefield of Gettysburg, for example, is about 400 feet and five elevations would require a step size of about 80 feet, and at that limited resolution many of the battlefield’s important features would simply disappear.

Of course there are things a designer can do to increase the number of steps: he could, for example, decrease the shading differences between the elevations in order to increase the number that can be represented. This technique, however, has a major built-in limitation in that as the colors are less well-differentiated, it is harder for players to reliably tell them apart. Human color perception is actually extremely complex, and perceived colors are greatly affected by surrounding colors, so the same color used in two different parts of the game board can easily appear to be different colors depending on the colors around it (there are in fact a large number of interesting optical illusions based on this effect, some of which can be seen here). As a result, unless the color differences between the different elevations are sufficiently large, they will be very prone to being mis-read, and even with as few steps as WV uses, this sometimes happens.

In attempting to increase the number of distinct elevations the game can graphically represent, it is of course possible to use chromatic variation (WV is mono-chromatic in that all the elevations are of the same hue, with the steps varying the saturation); for example, the lower elevations could be done in shades of green while the higher elevations would be done in shades of brown, with intermediate elevations of intermediate hue. This technique can and does help, but it has some built-in limitations of its own. First, not all hues look right when used for elevation (those who saw the pink used for one elevation in a later edition of TSS will not soon forget it). Second, color is extremely useful and is wanted for other map elements besides elevation, and because elevation is a background color, a color used for elevation is not available for any other use (it would not do, for example, for roads to become invisible at one elevation because very similar colors were used for both). Finally, there is the problem that the eye does not naturally assign an ordering among hues as it does to different levels of saturation of a single hue, so the eye has trouble sorting out among hues which is supposed to be higher and which lower, making the map look like a patchwork of random colors rather than a representation of high and low ground.

Current games that use relief shading generally use a mix of earth and vegetation colors of various shades to denote elevation, which produces a greater number of distinct elevations than the monochromatic method used in WV. Three Days at Gettysburg, the successor to TSS, for example, uses various shades of green and brown to produce a total of nine relief colors, with a step size of 40 feet. From a line of sight point of view, this isn’t ideal, but in any case the previously noted limitations on the tolerance gamers have for complexity in line of sight rules probably wouldn’t make it very useful to have substantially more steps anyway; adding more steps with rules that can’t take advantage of them really serves no functional purpose.

And speaking of no functional purpose, readers may recall how elevation was presented in Napoleon’s Triumph. That game actually had ten different elevation steps, presented in a purely monochromatic scheme. Of course it wasn’t possible for players to reliably identify which color represented which elevation, but in practice they didn’t have to; the game had no rules that depended on the presentation of elevation, which in the strictest sense was purely decorative. Game play depended only on the size and shape of the locales and not at all on how the elevations were rendered. Elevation coloring, however, actually did serve a purpose like the rest of the decorative artwork in that it made it much easier for the players to tell at a glance what they were looking at. Compare the illustrations below. One shows a “purely functional” section of the NT board with just the locales and roads, without all the decorative artwork, and the other shows the full map with artwork. In addition to being much uglier, the “purely functional” version is actually less functional because it is much harder to scan. Although it has no technical significance in the game rules, the decorative art in fact provides critical visual cues that allows the player to understand what he is looking at; experienced players will even generally know what the locale symbols will be based solely on the decorative artwork (that this is so should come as no surprise when you consider that the locales are in fact a representation of the same terrain that the decorative artwork represents, just in a different way).

Without Decoration
NT Without Art
With Decoration
NT With Art

For The Guns of Gettysburg (I bet you were wondering when, if ever, I was going to get to this, weren’t you) I went with the same basic graphical approach to elevation I had used in NT. The major difference was a large increase in the number of distinct elevations: rather than the 10 different elevation steps of NT, GoG has 21. Since GoG has no more dependency on people being able to determine which color represents which elevation than NT, the increase has no effect on actual play, but it does bring up another, purely visual, problem that comes up when the visual step size is so small: the edges of the colors can all but disappear, making the map look like a smooth gradient of continuously changing color rather than a series of discrete steps. While there isn’t anything inherently bad about continous color changes in general, it does make it quite difficult to see the elevation changes, and because those changes really are important visual cues to reading the map, the result is impaired map readability if they can’t be seen. The answer to this is a technique called sharpening or edge enhancement. Basically, this technique is an algorithm that when it finds a change in a numeric value, increases the value on the edge of the high side and decreases the value on the edge of the low side to make the change from low to high greater. In graphical terms, it makes it more visible. Generally it is considered a bad thing to crank this up past a certain point, because while the lightening/darkening on the edges of objects is not directly perceived when it is done subtly, large changes make this effect visible. As it happens, I rather like the effect and so crank it up pretty high. Below is a sample section of the map for GoG, with and without sharpening, to show you the difference:

Without Sharpening
NT Without Art
With Sharpening
NT With Art

Well, that about covers this subject, and it is, I suppose, high time for the first reveal in the mapboard strip-tease, so here is the relief layer of The Guns of Gettysburg mapboard (click to open a large version in its own window, which is really necessary to see much of anything beyond a soft green blur):

The Guns of Gettysburg Relief Map

Gettysburg Relief

Click on the image above to open in its own window

Oh, one last thing, just to satisfy any curiousity you might have with regard to that quarry I mentioned near Gettysburg, here is a 1946 map showing the original relief prior to the creation of the quarry, and the quarry itself as shown both on a 1999 USGS map and on contemporary aerial imagery:

1946 (pre-quarry)
Quarry Map
2005 (?)
Quarry Photo