Real-Time Strategy “Level Design”

As well as development updates, I’ve decided I want to start writing more pieces about interesting mechanics or things I admire in other games outside roguelikes. This is a long and detailed blog entry about level design in a genre nobody really talks about level design in – real-time strategy games. When conversations like this happen it’s generally about multiplayer balance, typically for Starcraft multiplayer maps that come with their own strategies for each of the three sides. However, here I want to talk about level design for singleplayer missions in RTS games, specifically the original Command & Conquer, one of the most well-balanced and well-designed games I’ve ever played. There are three missions in particular which really show off the thought and effort put into making these missions not just “maps”, but actual levels that demand more thought from the player than build tanks -> move tanks -> win. If people are interested in more entries like these as well as game development ones, let me know.

So, this is our first example. At the start of the mission your forces need to reach the green cross, but they are blocked by a pair of enemy vehicles on the bridge. These are Mammoth Tanks, the strongest units in the game – a single one of them would destroy your entire force by itself, let alone two. You begin with six units. The largest is an MCV (“Mobile Construction Vehicle”), a unit required to construct a base, and therefore vital. You also possess a single light tank with an anti-armour cannon, two “attack bikes” (at the front of your force) with anti-armour rockets, whilst the two “buggies” either side of your tank can only damage troops. Your tank and two attack bikes, whilst both strong against armoured units, would be pulverized in direct combat with the Mammoth Tanks. In this case, the game requires you to utilize the design of the level to survive, and if you wish to play optimally, to take advantage of – and, perhaps, even exploit – the game’s AI.

Strat1

As above, you have been given two buggies to start the game with. These are flimsy and can deal no real damage to armoured units, but you have been given them for a reason. Your attack bikes are effectively minor glass cannons – a Mammoth Tank would destroy them in one or two salvos, but their rockets deal significant damage. Thus, your first step is to send a buggy down the white line, get the attention of one Mammoth, then have it lead it away. At this point the player has two options – one is arguably the optimal play, whilst another is simpler to carry out and requires less simultaneous movement of units, but leaves the player more exposed after it. This less-optimal play would involve you having a second unit go in, distract the second Mammoth, and have them go around in that loop whilst your other units get across the bridge and set up a base. Your tank and attack bikes will then be able to protect your fledgling base, though you will need to keep your buggies moving. However, unless you are willing to keep commanding your buggies in an endless circle which will distract you from the rest of the mission, you need to find a permanent solution to the Mammoth Tanks.

Strat2

Instead, the optimal play is to have your tank & bikes move up behind one of the Mammoths chasing a buggy, and attack it from behind. Sometimes the AI will turn its attention to the new attacker, but sometimes it won’t. I’ve never worked out what the exact trigger is for this – maybe it’s random, but I doubt it, as C&C is a very deterministic RTS with few RNG factors. I think it has something to do with proximity – if the Mammoth Tank feels it is sufficiently close to its target (the endlessly-fleeing buggy) it will ignore those shooting at its rear armour. Thus, you need to keep your buggies moving in circles. This must be done slowly, so they don’t advance too far on the slow Mammoths and get the Mammoths to turn around and destroy your useful units, but still stay out of range of the Mammoths’ weapons. This must be done whilst having your armoured vehicles attack the Mammoths, but not getting too close, and also moving your MCV (that large vehicle, one that lets you construct bases) across the bridge to build a base. This is a demanding task with several parts, and the map is specifically built so that you have to use the units you’ve been given very carefully, but at a cost of leaving your base undefended whilst you deal with the initial Mammoths. This could not be done without a map design of this sort; it looks impossible at first if you attempt open combat, but ensures the player can find a single very specific solution to master the situation. You are also under a constant time pressure in all late-game missions in the game because you know the map’s very finite resources are being mined by your foes every second that passes; you could wait until you’ve killed the Mammoth Tanks before building a base with your surviving units, but you are instead encouraged by this additional pressure to send your MCV on unguarded whilst your units stay back to handle the tanks. Although your base will be initially undefended, against the tank and the two attack bikes the non-retaliating Mammoths will fall reasonably quickly, and certainly before the first waves of enemies – primarily troops, against which the tanks wouldn’t be that much use anyway – come at you from the enemy base.

In this second example, you start the level with a single unit, a “Stealth Tank” – it is invisible to all units unless it strays within a tile of an enemy unit. They are flimsy, and would die in open combat to any of the units or defensive buildings visible here. They are also revealed if they stray within a single tile of an enemy unit. You are tasked with retrieving the red unit – an MCV, and the only one you get in the level to build your base with. To do this you must find a way into the base and then find a way to construct your own base after that, without having the ability to extract your MCV and build safely outside the walls of your foe.

Strat3

Here, the player is unable to use traditional RTS strategies to retrieve the MCV. You obviously cannot build up a force to enter the base (because you have no production buildings), and nor do you start with a force sufficient to break through even the weakest of the three gates surrounding this base. The first part of this mission is more like a puzzle game where you must find a particular route through the level whilst avoiding enemies that behave in specific ways. There is a single path to enter the base and retrieve your mission-essential unit:

Start99

Having done this, your secondary objective is to destroy all the enemy units and buildings on the map. You cannot get your MCV safely out of the enemy base, and thus the player must build their base within the enemy’s. The tall yellow buildings with blue/white tops are “Advanced Guard Towers” (AGTs) – these are powerful defenses with a significant range, as (approximately) shown by the red circles below. You cannot build within those circles, nor safely move your units within them, as they will come under fire you cannot resist at this point in the mission. However, AGTs require power, and there are three power stations your Stealth Tank can safely aim at by uncloaking (they must de-cloak to fire) just north of your MCV. Killing two of these power plants is not sufficient, but all three is. Once your stealth tank is positioned safely, they can destroy those power plants, disable many of the base defenses (though not the smaller yellow towers, nor the tanks) and thus start to build a base.

Strat5

Subsequently, the optimal strategy is arguably to build walls outwards to block enemy vehicles, and although this is too complex to show in a diagram, many of the enemy buildings and enemies are positioned in such a way as to encourage the player to slowly advance across the base and taking it over in a very specific order. So, on this map the player is tasked with finding a route through obstacles – a very un-RTS principle – before thinking outside the box and finding a way to take out a base from the inside whilst not triggering its external defenders, rather than attacking from the outside. Were any of the buildings placed differently – for example, a second AGT on the left side of the top gate – this map would become impossible. As it is, there are just enough power plants within safe firing range for your fragile Stealth Tank, and just the right locations for defenses, to enable a very specific and unique strategy and get the player thinking carefully about the power/defense/enemy layouts the mission has put in front of you.

In this third example, you start with two groups of units split off from one another. The units at the bottom need to navigate the lower half of the map to retrieve a non-essential cache of money in the village at the end of the green arrow; the units at the north of the map need to enter the base in the centre-left, capture its buildings, and start building a base from there. In this case, you have a lot of units and the base is only guarded by a pair of standard “Guard Towers” (blue circles) – these are weak against tanks, but tear through soldiers. You are only given soldiers, but the makeup of your soldiers, and using both your forces effectively, are the keys to making the opening stages of this map far from trivial.

Gunboat

Your unit composition on the top is very cleverly thought through. You have a dozen normal soldiers (blue); one commando (the light green), who can one-shot any troop at range, and instantly kill any building if in an adjacent tile; and three engineers (dark green), who can capture buildings (note the three yellow buildings within the western base – these are what you are expected to seize). Your troops are not sufficient to overcome the two anti-troop guard towers; your commando, being infantry, cannot possibly get close enough to destroy them (using C4 in close proximity). The engineers are essential. The only solution is to use the standard troops as fodder; they must run ahead of the commando into the guard towers to allow him to C4 them both. Guard towers target enemies closest to them, so the player needs to also ensure the commando hits the guard tower in the one-second window between its bullets without giving it time to ID the commando as the “closest enemy” and open fire. Commandos briefly overlap with the a tile a building is built upon during the placing of the C4 – ensuring they will be considered the closest enemy – so meeting this timing window is essential.

Additionally, there is a “Gunboat” patrolling (the white circle) left-to-right-to-left on the river. This gunboat will destroy your troops if it comes into range; a second use of your lower units is to keep an eye on the gunboat, and ensure that your northern units only attack the guard towers within the safe window whilst the gunboat is on the other side of the map. Much like the first level discussed in this post, this adds a secondary time pressure – the player’s sacrifice-then-C4 attack on the western base must be carried out before the gunboat concludes a single loop, and you should ideally command the southern force of soldiers to keep an eye on the gunboat and avoiding sacrificing your key northern troops before you get the chance to then sacrifice them to the guard towers.

On this level, then, the defenses of the base, much like the above second example, are specifically designed and placed so that your unit composition must be used in a particular way to overcome them. This sacrificial solution is also particularly fitting because the army you are playing as for this mission, a strange kind of terrorist-militia-irregular-army-paramilitary-cult-populist organization, are willing to sacrifice unimportant soldiers (within the plot), and this mission nicely reinforces that from a thematic standpoint.

Finale

Ultimately, all three of these missions ask far more of the player than to merely build a base, build up an army, and conquer the enemy. In all three cases this is certainly a later objective, but rather than simply starting the player off with an MCV, some resources and some escort units, they each ask the player to think about the tactical situation in more interesting ways. In the first the player has to utilize the map to their advantage to overcome an apparently impossible battle, and take advantage of, or even exploit, the AI. In the second the player has to navigate through a “puzzle” base before cautiously developing their own base within the limits imposed by the enemy’s fortifications. In the third, possibly the most interesting, the player needs to control two different groups of units; use the southern unit group to assist the northern unit group in timing their attacks to avoid the gunboat; and then to launch an attack in a narrow time window where you need to manage both your meat-shield units and your crucial unit very carefully with one-second time windows for destroying the buildings in question. Rather than multiplayer maps that demand perfect balance (or at least strive towards it), or maps like those offered by most RTS games which are merely “somewhere to fight” (looking at you, Supreme Commander), the layouts of these maps and the specific and few units given to the player provide unique and interesting challenges, and also show the player particular tactics or strategies that could potentially be deployed in later missions even without the restrictions these missions have. Whilst other C&C games certainly have some interesting levels, the original still stands out to me as a cut above the rest.

Be Sociable, Share!

23 thoughts on “Real-Time Strategy “Level Design”

  1. Saw your post on /r/RealTimeStrategy. I quite enjoyed reading through this, especially nice to find while having a slow day of work. Your style of writing comes off as clean and professional but remains enjoyable to read. I hope to see more from you.

    I remember only ever getting to play the original C&C on the N64, if that was even the same game. When I did have a PC to play I went straight to Red Alert and Tiberium Sun, bypassing this game entirely. Definitely looks like I missed out on some fun.

    Its level designs like these that I miss, where you have to figure out a very specific and predetermined way to win. Don’t get me wrong, open sandbox is great, but guided levels like these, when well designed, are amazing.

    I preordered Company of Heroes 2 and to use your wording here, mostly build blob-> move blob -> win, there is one level I’d like to mention in particular.

    Mission 8 Panzer/Tiger Hunting. This sees you controlling a group of conscripts, basic infantry, following a allied AI tank convoy. The convoy is ambushed by a Tiger Tank and quickly destroyed. The game then tasks you through radio orders to deal with that tank. With basic infantry? Impossible.

    But don’t worry, command is sending you additional reinforcements. Sadly these are simply more infantry! A sniper, some engineers (if I remember correctly) and a couple squads of anti-tank rifles. Turns out the anti-tank rifles are almost useless against this behemoth, the sniper is only able to provide additional line of sight, the engineers can set AT mines and remotely triggered explosives and the additional conscripts have highly ineffective anti tank grenades.

    On top of all this, all the aforementioned abilities require the requisition resource, which is also in short supply. You wont be provided any more either, and have no way to gather it through traditional means. However, if you scout the map, which is a small village, you will find supply crates that do indeed yield additional requisition. If you look hard enough you can also find additional weapons to equip your pathetic forces. There is a light AT-Gun, which unless your hitting the rear or side of the Tiger is not helpful, and more of the under powered AT Rifles.

    I’m sure by now you know what must be done. The Tiger Tank is patrolling the village in a seemingly random fashion, and you’ve got to lure it into traps and ambushes. Eventually you’ll manage to wear it down enough so that it retreats to another portion of the map, a derelict air strip. This creates a new set of challenges with the wide open runway. I personally devolved into mob tactics, and managed to take down the Tiger.

    From here you have to defend the disabled (not destroyed) tank wherever it falls until you repair it, through additional reinforcements, to a driveable state and bring it to the triggered victory. Of course this is through enemy roadblocks and aggressive hostiles.

    Thanks for this,

    -Shain

  2. Market forces I suppose. Not many people play them, and they’re expensive to produce. There’s that famous Supreme Commander statistic that most played skirmishes with AI rather than campaign missions or multiplayer.

    Planetary Annihilation isn’t even going to have a true single player campaign, rather a procedurally generated series of missions called Galactic War.

  3. @ Shain, many thanks for the kind words. Most of this blog is updates or discussion around the game I’m developing or roguelikes more generally, but RTS remains a genre I’m very keen on (and actually influences the design the game somewhat) and there are going to be more non-game-development posts every now and then. The N64 C&C was “the same game” on some levels, but for all intents and purposes it was its own beast, and suffered on console for all the reasons that RTS games remain almost exclusively PC titles. Also, I really like the example you posted – that’s exactly the kind of puzzle-solving in non-puzzle games I enjoy (and something I’m hoping to add to my game, though it does have “actual puzzles” too!). Do check back – I’ll probably do a follow-up to this piece, as the response here, on Reddit and elsewhere has been really positive.

    @ Herisson – true, on both counts. Planetary Annihilation… I have to admit, it doesn’t interest me that much. It feels like a game similar to SC2 in my mind, or the more modern games that laughably use the C&C label – games where units are effective in just too many ways. Sure, units might have weaknesses, but in general they have only one or two weaknesses rather than one or two *strengths*. And that seems the best form of RTS balance to me – “Every unit is terrible, except in one or two situations”. Take C&C 95 as an example where most units had only one or two very specific situations in which they shined. Few RTS games seem to master that these days. Nevertheless, I’m keeping an eye on it…

  4. Your non-URR-posts are definitely something to look forward to and from time to time have the effect of making me view specific gaming related issues in a different light. So, would I want to see more of such posts? The answer is an emphatic ‘Yes’.

  5. @ Arth, Vincent & Yog – many thanks all! Very glad you liked it. I’m pleased to announce a few other similar ones are in the works, to be interspersed with URR updates (like today’s), so another one will probably appear in a couple of weeks.

  6. Great article. C&C 1 was great – creating puzzle singleplayer maps and balanced asymetric multiplayer / teamed multiplayer maps was super fun back in the time …
    The tiberium mecahnic was one of the best things in RTS – heritage of Dune2: vulnerable units have to collect a limited (execept for the tiberium-trees) resource -> area control is necessary -> forces you to use map layout. The place of tiberium collection can vary to speed up production, gather at more vulnerable places first to optimize area borders, gather at more protected area while doing an offensive strike, etc. Rock,Paper, Scissors was cited often when describing the battle mechanics, but its not as simple – there is always more than one scissor to a paper and even enough stones will do. While the large scale battles of TA and descendants have their appeal, I always liked the strike group / special forces feel of C&C1.

  7. @ Manu – Glad you liked the piece (there is a potential follow-up in the works). I agree about the placement of Tiberium fields – in some maps they were very clearly placed to be difficult to defend (in singleplayer missions) or serve as choke points (in multiplayer). Obviously every decent RTS does that these days, but I’d never seen a similar kind of consideration around resource placement before the original C&C. The third example here is in fact a good example as the main field to the east is such a long journey from your base and very difficult to defend, though you are given a small patch within the confines of your base. I totally agree large-scale battles have their place, but I agree that the feel of taking a few units and doing a lot with them always appealed to me more, and certainly forces the player to play more intelligently than gathering up a swarm of units and just hurling them at your foe. Anyway, thanks for the reply! The follow-up should appear before too long…

  8. “Sometimes the AI will turn its attention to the new attacker, but sometimes it won’t. I’ve never worked out what the exact trigger is for this – maybe it’s random, but I doubt it, as C&C is a very deterministic RTS with few RNG factors. I think it has something to do with proximity – if the Mammoth Tank feels it is sufficiently close to its target (the endlessly-fleeing buggy) it will ignore those shooting at its rear armour.”
    – Actually, it’s more or less the reverse. If the Mammoth Tank manages to fire shots at the unit it is chasing, it will also consider other targets. As long as the unit you used to lure it with will stay out of its range, though, it will singlemindedly pursue that one unit and ignore everything else until it managed to attack it at least once. This is especially noticeable in units on Area Guard mode, who act the same way and pursue units outside their original guard range, but once they manage to fire at it once, they will lose interest and go back if they are outside that original guard range.

    “You cannot get your MCV safely out of the enemy base, and thus the player must build their base within the enemy’s”
    – You can, actually, and the mission specifically has a script that allows you to lure away the two mammoth tanks guarding the front of the base by attacking the village on the map, which will make GDI send in a squad to check out the disturbance. If you do that, then attack and destroy the guard towers at the base entrance (which can be done while staying out of the AGTs’ range), and then rush out your MCV, you can get it outside alive and deploy it lower on the map. This leads me to believe that strangely, the method of deploying the MCV inside the base may not actually have been intended by the developers.

    “who can one-shot any troop at range, and instantly kill any building if in an adjacent tile”
    – Not quite accurate; they actually have to move INTO the actual building to bomb it. In fact, as seen on larger buildings like the Temple of Nod (which is advisable in the “Blackout” mission since it’s the only way to avoid a nuclear strike from it), they go all the way to the center of the building before they place the bomb.

    “Your troops are not sufficient to overcome the two anti-troop guard towers”
    – This is wrong. C&C infantry uses a system of five sub-cells on a game map cell, spread out in X shape. Bazooka soldiers can wreak havoc on these towers if placed in the correct sub-cells. The trick is to only use two bazooka soldiers, so that when they enter the cell from which they fire, they stay on the sub-cells the furthest away from the tower. (You can do this on two cells vertically to get at least three bazookas firing at it simultaneously). Add a third on the same cell, however, and it would enter the middle of the five sub-cells, where the guard tower can reach it, and the resulting barrage’s spread-damage will take out all three of them. But get them placed correctly, and they will make short work of the guard tower. So, you CAN get into the base without using them as cannon fodder. The real nuisance in this level is the combination of the harvester and the gunboat, which foul up the approach of the guard towers, and give you a very limited time window. A possible way around this is to make the artillery at the bottom force-fire on the gunboat’s path to take it out, but this takes quite some takes time.

    • That’s very interesting about the AI. That explains a lot about the AI behaviour I’ve seen, though I’ve never been able to isolate the logic like that. Have you modded the game or made custom missions or similar, or does your knowledge of the AI just come from having played it a lot? Re: moving the MCV out, I had no idea the mission had such a script! I’ve played it many times but never discovered that. How curious. I can’t imagine they intended you to move the MCV out as it seems so well-made for a “build within their base” strat… and yet, as you say, there is such a trigger. Weird.

      Ah, yes, that’s just a typo re: the building; I meant they need to be adjacent in order to move into the building and plant the bomb. I know of course about the X-shaped subcells, but I wasn’t aware guard towers prioritize based on their proximity; I thought they fired randomly into the tile. That’s very interesting! I’ll try a non-sacrificial strategy next time I replay C&C 95.

      • I’ve made missions for it, I’ve modded it, I’ve hacked it, and I’ve patched it. I’ve done about everything you possibly can with the game. My knowledge of the unit behaviour is mostly from experience in mission making, though. These kinds of things are hard to figure out from disassembled code.

        I checked the mission triggers for the specific situation in the Nod #12 mission, and it seems slightly different than I thought, though. I assumed they were simply on Area Guard, which would give the behaviour I described. But instead, it seems that once one of the Mammoth Tanks is attacked, the AI combines the two into a team with as command “attack units”. This is one of those special logics specifically for giving to attacking forces, and is mostly for making the difference with the other group-attack command, “attack base”. I have to admit, I’m not entirely sure about the specifics of this logic. It indeed seems to be less singleminded than the Area Guard one.

        On the sub-cells, it’s not a prioritizing issue; it’s simply a matter of weapon range. C&C’s range is not per-cell; it is an actual distance. The range of the guard tower weapon is exactly 4.0 cells (I got all the C&C internal stats; internally, it saves the range as a number that is range*100h, so it is technically accurate to 1/256th of a cell), but you have to realize this is exactly 4 cells, as seen from the CENTER of the guard tower. This means that in reality, it can fire up to 3.5 cells away from itself. Thus, in that fourth cell away from itself, it can’t target the two sub-cells the furthest away from it. They’re simply out of range.

        If you’re interested in looking at these internal stats from C&C1, I’ve extracted them all from the exe file and converted them to ini format. You can see them here:
        http://nyerguds.arsaneus-design.com/cnc95upd/inirules/
        Note that some things, like colors.ini and the ColorScheme settings in houses.ini, are data added by my 1.06 patch, and not part of the original game. Some other things, like music and warheads, have also been slightly modified to fix things (for example, in warheads, 255/256 was seen as “100%”, which gave rounding errors on larger damage values). My goal is to eventually hack the game to such an extent that it can read all these ini files and use them instead of the hardcoded data. A proof-of-concept of that is already in the latest release of my patch, which reads the music data from the themes.ini file.

      • Oh, by the way… the Bazooka’s rocket weapon also has a range of “4.0”, but they have the advantage that a building is in their range from the moment any of the building’s cells are in their range.
        (I assume the same will probably hold true for vehicles, who also take a full cell according to the game. Infantry are just an exception to all that)

        Strangely, their projectiles will still fly to, and do their damage at, the center of the building, though; something I once tested after accidentally changing a building’s perceived size to something along the lines of 40×40 (I was basically overflowing an enum, so it took the random data behind it 😉 ). It was especially fun to see with a flamethrower.

        • Hmm, interesting. I modded Red Alert 2 extensively and I often found the exact behaviors/stances of enemy units one of the trickiest things to puzzle out. That’s fascinating about the Guard Tower – so it cannot ever hit the units on the far right of the cell. I’ll have to try that out next time I play the game through again. I’ve often noticed that weapons aim at the centre (TS/RA2 had something similar, though they would rarely hit another point instead) – I can imagine re: the flamethrower! Thanks again for the really interesting comments, it has been great to hear some thoughts from an expert on the game.

      • By the way…. someone recently pointed out to me that on the split river mission, the artillery from the south part can actually reach these guard towers from across the river, once you reach the end.

        It honestly never occurred to me, but indeed, when timed right to avoid the gunboat they can just wipe them out with no problems at all, giving a really simple way to access the base. This also makes the mission a logical chain of events, since there is suddenly a purpose to having these specific units in the southern group.

  9. Pingback: Real-Time Strategy "Level Design" II | Ultima Ratio Regum: The Roguelike

  10. An interesting thing I’ve heard recently about the mission with the split forces, in response to me comparing it with the simplified Nintendo 64 version of the mission, is that apparently, the devs designed it so you can use the artillery from the bottom part to shoot over the river to take out the guard towers without any risk. All you need to do then is time it so the gunboat isn’t around.

    In the Nintendo 64 version, the gunboat is absent, and the two parts are connected, making that strategy far more obvious:
    http://nyerguds.arsaneus-design.com/cncstuff/mappics/campaign-n64-nod/scb11ea.png
    (well, this is the second version of the mission, but the principle still applies)

    • Whaaaaaa?!?! I HAD NO IDEA! My mind has been entirely blown. What an interesting map, I had no idea that was something possible. That actually makes a degree of sense; I’d always (as I wrote) just assumed that the lower forces were for pursuing the cash cache…

Leave a Reply

Your email address will not be published. Required fields are marked *