There’s a special kind of anarchy in the fights you experience in Divinity: Original Sin II.
This computer RPG, released last year by Larian Studios, encapsulates the freeform promise of the genre, allowing you to tackle its quests and face its world’s threats in wildly varying ways. Nowhere is that principle better expressed than when you’re in combat. In any fight, half the battlefield can end up on fire and the other drenched in acid. The air might be thick with electrified clouds, and summoned characters and resurrected corpses wander free.
Victory often feels as if it’s plucked from the jaws of death – or from chaos – and yet DOS2’s combat design is founded on establishing predictability for players, so they can make and execute plans, tight pacing, and also a sense of a story within the battle. As systems designer Nick Pechenin says, “Fights are basically performances, and you want some kind of plot in them.”
The trouble with armor
DOS2’s combat design is a close evolution from 2014’s Divinity: Original Sin, but Larian Studios knew the original had some issues. The team liked the depth of its combat, but felt that it tipped the balance too far towards chaos. The problem was with its armor system.
“We see the best tactics when the player realizes a fight that’s going OK goes for the worst.”
Armor had the chance of blocking status effects, meaning that if you planned to knock a bunch of enemies out with a stun attack, you didn’t know for sure it’d work in every case. “The good part about this was that every encounter felt different, so when you started a fight it felt fresh. Things went wrong and right in very different ways,” says Pechenin. “But at the same time it really prevented long-term planning, because you didn’t know how many people you’d stun, so you couldn’t predict what you’d do next turn, and because of this you just wouldn’t think about the next turn.”
So one of the big changes to DOS2’s combat design was to its armor system. Rather than absorbing a proportion of incoming damage, armor completely negates it. There are two armor types: physical and magic, which negates any magical attack, including negative status effects. But as these values take damage they’re whittled down, and once gone, the character is left open to losing HP and vulnerable to status effects.
So far, so deterministic, but Larian wanted attacks to retain a ‘spicy’ feeling. The solution was a small variability in incoming damage which may entirely knock armor out, or it may not. “So there’s still some RNG there and you don’t know exactly how things will turn out, but you have a high chance that things will go as you want them to,” says Pechenin. “But at other times the game will throw a curve ball at you and make you scramble to find a new plan.”
Pacing a battle
The next challenge was to set the pacing of battles. Larian wanted each to last an ideal number of turns. They wanted the time it took to destroy the armor on an enemy to feel good, as well as the number of turns that it’d take to stun an enemy, to destroy the armor on a player character, or to kill them.
It was not easy, since DOS2 features so many variables. Larian’s combat designers never know how many characters the player will be fielding in an encounter, since one or more of them can be off exploring an entirely different part of the map.
The characters who are in the fight will be equipped with very different armor and weapons, which might be very powerful because they’ve explored every inch of the maps, or they might be very weak because they’ve only played through the main campaign. They may be high level for the area, or low. Players might have unlocked many different spells and abilities, or very few. They may not know how to use them well, and they may simply forget to use them. They may have large stocks of consumables such as grenades and potions, or they might be hoarding them. In short, the dynamic range of the potential power a player fields in any given encounter is very wide.
Larian’s approach to balancing enemies’ armor and HP values was to create a curve to the way HP increases as characters level up, and then to use that a baseline value from which enemies’ stats would be calculated.
“Getting that curve nailed down was quite a challenge, just because of how much extra content we have,” says Pechenin. Some players might have discovered an amazing sword that allows them to one-shot enemies, which effectively reduced the challenge to nothing.
Embracing OP design
But rather than balance out these extremes, Larian embraced them. “Our usual philosophy is for player to be as OP as they want to be,” says Pechenin. But to mitigate the effects of a player finding an amazing sword, they also steepened the HP curve so that in a few hours that sword will be next to useless, returning the character to the baseline – unless they’ve found an excellent replacement.
“For AI it always makes sense to pile on one person and just murder them completely, but for the player it really just sucks, because the damage isn’t spread over their characters. They want to feel threat piling up, not having their characters one-shotted without being able to respond.”
In truth, he admits they went a little far with the steepness, because players complained about their super weapons getting superseded too soon, and so they patched in a slightly gentler curve. “This is completely valid, but in general the curve allowed us to give something very impactful to the player but still present them challenges even after 50-60 hours of playtime.”
And beyond just placing powerful swords around the maps, Larian is also comfortable with players exploiting its complex systems. If a player figures out a way of teleporting lava into a fight and drops it on a troll’s head, that’s a good thing, providing a good player story and fulfilling a lot of the reasons why many people play CRPGs. But as a player, you should have to work for it, whether creatively or effortfully. “And once you’ve used an exploit like this, it shouldn’t be universal, it shouldn’t carry you to the end of the game,” says Pechenin. “That would be no fun, and kind of boring.”
One of the ways Larian discourages exploits – and players favoring certain tactics too much – is in DOS2’s combat design. In Act III of the game, many of the encounters are specifically set up to flummox certain powerful tactics. So, for example, in one fight the player faces enemies with the Fortify ability, which prevents them from being teleported by the player. If they’ve been playing so far by teleporting enemies into killzones, they’ll need to scramble to come up with a new approach.
Making turn-based fights feel desperate
Still, whether you have a good strategy or not, DOS2’s battles have the knack of making you feel you’re hanging on by your fingernails. “We see the best tactics when the player realizes a fight that’s going OK goes for the worst,” Pechenin says. If you see a chunk wiped off your mage’s physical armor it can often seem if it’s about to become dangerously vulnerable, even if across the party you have suite of fantastic powers that will see you victorious.
One of the ways the game conjures this feeling is by managing armor and HP values in relation to the number of hits Larian wants it to take for them to be eliminated. So, if they want a player’s character to ideally be killed in five hits, they have enemies’ damage output kill them in 4.5 hits. The character still dies in five hits, but their HP bar will look more depleted and have just a sliver left before they receive the final blow.
“Just seeing this bar being very short will feel a lot more threatening,” says Pechenin. “You don’t know where it’s going to go, and you’ll be pushed to focus on this guy.”
Moreover, Larian’s careful to ensure DOS2’s AI picks its targets in the right way. They don’t want them to be merciless, always focusing on the weakest player character. “Of course for AI it always makes sense to pile on one person and just murder them completely, but for the player it really just sucks, because the damage isn’t spread over their characters,” says Pechenin. “They want to feel threat piling up, not having their characters one-shotted without being able to respond.” Larian knows that a good fight is not about fighting in the most brutally efficient way but the most dramatic, with pacing that allows players to face threat and then have have a chance to react to it, before the AI mounts the threat.
The trouble with armor (redux)
Balancing DOS2 was a major challenge, one which has continued after its release in September of last year. The process has led to various surprising observations about the way players approach kitting out their party. Pechenin says that, overwhelmingly and regardless of skill, players buy skillbooks over any other item from shops. Then they’ll invest in upgrading their weapons. But even good players tend to skip buying armor.
That’s particularly true when they’ve experienced a period of being overpowered, and it’s only countered when they’ve felt threatened across several successive battles, after which they tend to blame the game for having a difficulty spike. But it was their gear that was the issue.
Armor continues to pose problems in combat itself. To put it simply, players hate to hit armor. Pechenin says that if, for example, you have two enemies next to each other, one with 100 HP and the other with 50 HP and 50 armor, the player will almost always go for the unarmored enemy first. “Just for the pure psychological joy of digging into HP,” says Pechenin.
But it’s the wrong choice: since armor blocks such status effects as stuns, it’s more tactically sound to clear it before hitting HP. ”It’s kind of counterintuitive; as a systems designer you don’t always think about this stuff.”
To help counter this, Larian tried to make hits look good to the eye. “It’s not a trivial matter, because when you hit something you want their bar to go down in a very visible manner, a good chunk of it gone,” Pechenin concludes.
“This kind of pacing is separate from challenge; it’s hard to nail down, especially in a game where you can have four party members with wildly varying power levels. At the end I think we got something close to feeling good.”