Darkest Dungeon

Been playing this a lot for the past few days. Into the latter half of my first playthrough. Various thoughts:

Overall it’s great. The writing and atmosphere are very cool, and consistently build on the themes of exploring the mental effect of being in a gothic horror setting. Very immersive throughout.

The game systems are well setup to hit all the standard roguelike patterns for people who like XCOM et al. Making the choice not to have a failure state (no game over) is a very significant one. It enables a slightly different direction from a lot of similar games–more drawn-out pacing, and you do all your learning within one playthrough. And the gameplay is carefully set up to reinforce the themes: right as you start out, buildup of stress and other afflictions is too high to deal with efficiently, so you have to give in to treating your adventurers disposably.

Design-y criticisms:

1) It’s missing any sense of intermediate progression targets to keep you motivated. In XCOM, you were always looking forward to the next “new” thing you were getting right after the next mission or two–new guns, new research, new building, etc. A lot of times when a mission popped you didn’t exactly feel like doing it, because you were actually hoping time run on until your next little progression rush from finishing a project. But you’d focus and get into the mission, and would often enjoy playing with whatever your latest new toy was. Then you go back to look forward to the strategy progress aside. DD just doesn’t have those progression beats on the strategy side. You want to farm up deeds etc. so your character upgrades go to higher numbers and are cheaper. Once you get a sense of how it all works, it’s too transparent that you can just say “okay, after I get X hundred deeds I’ll have the strongest characters and can work on the endgame.” XCOM _never_ made me think (even though I theoretically could), “okay, I just need Y scientist-days of total research to fill out the whole tree, so let’s grind out missions until I get there.”

The outcome I predict is that I’m going to plow on enthusiastically with my initial rush of enjoyment of the game’s systems, and curiosity about the final dungeon. But once I start suffering setbacks that cause me to redo any grind, the motivation will evaporate and I’ll stop playing, feeling like I saw all the gameplay it had to offer.

2) There’s a really bad progression snarl based on the rule that high-level soldiers won’t go on low-level missions. I get what they’re trying to do, but it’s not extremely well thought-out. Your adventurers’ strength is based not just on their level, but on their gear and trinkets (in fact, the latter two are a lot more important). So even if you’re trying to build up your party strength by improving your Blacksmith before going into higher-level areas, you might have many adventurers who refuse to go to any areas other than the higher-level ones. At minimum, you wind up leaving lots of adventurers at, say, freshly L3, while you use your L1-2s to farm up materials to upgrade the L3 guys further. In the extreme, this can lead to the bizarre case of wanting to fire mid-level characters to make room for lower-level characters who have the “advantage” of actually being willing to do the missions you want to do. Again, I get they’re trying to prod you to keep pressing on, but this is a very messy and aggravating way of doing it. A simple rule like “the adventurers don’t gain XP from too-low-level missions” would accomplish the goal of making you get into the higher ones to keep progressing.

3) There’s too much fiddly setup before a mission. Here’s my general routine between finishing a dungeon and starting the next:
–Check if I have mats to upgrade any buildings
–Check if anyone on the Stagecoach I want to recruit (can be a complicated decision since I have to fire someone if so)
–Check what missions are available, and which one I want to do
–Pick a party for that mission (also complicated, but no complaint here since it’s the interesting part)
–Go back to the town screen, and drop each of those party members into the Blacksmith to see if they have any gear upgrades I want to get them
–Then the same for the Guild Hall for skill upgrades, and also checking if I want to swap any skills in/out on the characters based on that particular party lineup and mission
–Same at the Survivalist, although at least you really only need it once per character
–Of all the characters that aren’t going, who needs disease treatment, who needs stress relief–put them in those buildings.
–Scroll through all the remaining characters and pick a few negative traits to remove–put those in the Sanatarium.
–Check the trinket shop for anything interesting
–Open the trinket box. Unequip all trinkets, then pick some out for this mission to equip to the planned party.
–Pick provisions (usually formulaic based on mission area and length) and go.

There’s just too much hassle here. You probably didn’t even read all that. Looking again to XCOM, it’s really worth thinking about the ways that it avoids burying you in a nonstop slew of really minor decisions–this is one of its important successes. Most things on the overworld take time, so you really only have to actively make choices in 1 or 2 buildings at any given point. Characters of the same class are the same other than talents, and can’t be respecced. Accessory slots have some more standard basic options so you’re not totally re-picking from a large complex list every mission.

Darkest does have interesting systems around characters/skills and trinkets, that allow for good customization and formation of a plan. But the lesson from XCOM would be to be more parsimonous about the different things that the player can constantly rearrange and juggle, while figuring out how to distill out only the important strategic decisions.

The Guiding Hand of Dark Souls

The Guiding Hand of Dark Souls

Posted on  by Hamlet

Dark Souls is widely regarded as an outstanding game for a variety of reasons, such as the “hard but fair” ethos in which its only demand is that you prove your growth as a player at each step.  The world is similarly praised for its clever interconnectedness.  I want to illustrate the level of thought that went into the level design, using a series of highlights progressing through the game.  This is somewhat inspired by this analysis of Super Metroid, a game which itself is clearly an element of the Dark Souls pedigree.  Many of the concepts described in that article apply to this game as well, but Souls has to maintain a special focus on one particular element: quietly helping the player endure the intense difficulty that the game itself created.

One of my goals is to refine the image that both fans and non-fans have of Dark Souls, as being defined solely by difficulty.  Perhaps its best-hidden secret is that it does not actually leave the player lost in the wilderness with no direction and death awaiting at every turn, but rather guides and supports them in subtle ways.  The game fulfills two roles at once: presenting a seemingly crushing challenge, but also, behind the scenes, doing what it can to make it more likely that each player finds a way through.  The defining emotion of Dark Souls is the thrill of both facing the challenge and then eventually surpassing it, and the game succeeds because it does everything it can to provide both of those experiences.

This post will take us through ringing the Bells of Awakening.  Many of the themes of the level design are shown most strongly in this first segment, when it’s critical to give a player an understanding of how this game works, and most importantly, to give them the tools to succeed in what comes after it.

Undead Asylum

The game opens with one of the very few moments that I’m going to question.  Before its threadbare tutorial is even complete, it throws the player into a boss fight, the Asylum Demon, where the solution is not to fight, but to escape through an open door elsewhere in the room.  I’m not sure what this is intended to teach, since nowhere else in the game is there a boss fight that you can run away from; you’re always sealed in (there are occasional large non-boss enemies you can run from, such as the Bridge Hellkite or the Hydra, but they’re in open outdoor spaces that make it obvious anyway).  These opening minutes are precious for hooking the player and starting to teach them, and adding a potential sticking point with no teaching utility is not parsimonious.  Arguably, one hint is that the boss fight is completely unfair (you don’t have a real weapon yet, so no non-expert player has a chance), but it’s too early to rely on the player’s trust that Dark Souls is never unfair.  Building that trust is a key goal of the early levels.

The escape door is behind the pillars on the left.

After finishing the brief tutorial and getting their starting weapons, the player returns to fight the Asylum Demon for real, which is where they prove they’re ready to begin Dark Souls proper.  Making the player do this before they can begin exploring the world is a technique that will be used throughout the game: the gatekeeper encounter.  Asylum Demon is harder than the basic enemies you’ll fight for a little while (many new players will take some time to reach the next boss, Taurus Demon), so this seems like an atypical placement of a boss fight.  This is done to ensure that you’re not struggling to learn combat against the basic Hollows while you’re also starting to explore the world.  This is hallmark Souls pacing: move some of the struggle to the front to make sure the player’s skills are at least somewhat shipshape for the upcoming area.  This ensures they can explore the area without being hamstrung by an inability to kill basic enemies along the paths.  Asylum Demon is a special use of this technique because the “area” here is the entire game.

Asylum’s huge windups introduce the basic concepts of combat.

Firelink Shrine

Firelink will be your only home for the entire game, and the first hub in the complex network of the game’s central levels.  Once you arrive at Firelink by bird, you will not take any mode of transport again for a long time, not until the end of Sen’s Fortress (other than revisiting the Asylum if you want).  This is the first impressive feature of Dark Souls, that everyone probably notices quickly.  The entire first half of the game is one continuous “level”.  It has different named areas with different looks and feels, but there are no breaks.  This is used for more than merely the immersive effect of avoiding loading screens, but allows (most of) the game to be one large 3D space you can explore.  Most dramatically, it allows for impressive reveals where the game takes you back to a place you’ve been, along a new route, without your realizing it until you get there.  That effect would have no power if you hadn’t been traveling along an unbroken path the whole time.  I won’t repeat this point every time it comes up throughout the game, but the discussion wouldn’t be complete without mentioning it.

Giving you a home as the first stop as a way out of the tutorial area is not a trivial gesture.  Places where you can feel safe will be a rarity, and one that you can always come back to is an important comfort.  We’ll see how the game pushes you away from Firelink and then returns you to it a few times across the early acts.  On a smaller scale, each bonfire serves a similar role as a root for your exploration of the local area.


For purposes of our discussion, the most important features of Firelink are the paths out: up the hill into Undead Burg, into the Cemetery right next to the Shrine, and down the elevator into New Londo Ruins.  The Burg is the right one for a first-time player, and the other two demonstrate two different approaches to the important design goal of silently but firmly turning you away.  New Londo is the harsher and more obvious approach: you cannot fight the ghosts there because your weapons pass through them, immediately causing all but the most stubborn players to try something else.  The Cemetery is more deceptive: nothing is obviously wrong with fighting the skeletons, but they outmatch you numerically for now.  Unlike New Londo, players might vary in how long they take to give up here before trying the Burg.  The game and the player haven’t yet had a chance to come to a meeting of the minds on the right level of stubbornness to cultivate.  The difference between the Cemetary and the Burg paths are the first two data points: the former is more punishment than even Dark Souls desires to inflict on you, and the latter is business as usual.  Calibrating the player’s sense of what to expect, in much more detail than “I heard Dark Souls is hard” will be important going forward.

Seems worth a shot.

A formidable-looking set piece involving 6 Hollows guards the stairway to the Burg. Only trial and error will reveal this as the easiest path.

There is a fourth path: down the elevator and through the locked door into the Valley of Drakes, provided you took the Master Key as your starting gift.  This will be discussed below.

Undead Burg

The gatekeeper encounter uses mob placements to create a tricky spot where you’re “tanked” by the incoming melee Hollow on the stairs, right where the other one starts throwing firebombs at you.

Once you do make your way into the Burg, things are mostly linear for a bit.  Simple exploration will find the first Undead Merchant and a bonfire.  The only major diversion is a Black Knight at the end, but that doesn’t open new routes.  There are no explicit shortcuts that loop back yet, because the path to Taurus Demon from the bonfire is quite short anyway (the area’s length is probably calibrated to still account for frequent deaths to Hollow packs).  Even with no shortcut, it exhibits a standard property of Dark Souls levels: once you know the enemy placements, the path from the bonfire to the boss can be sprinted with no fighting and comparatively few complications.  Bosses are designed to take many attempts, and there’s no question that this can produce a pacing problem—newcomers frequently cite the frustration of reclearing to the boss after every death.  The fact that this can mostly be avoided simply by trying is a lesson that’s good to start early, in the hope that players will pick up on it sooner rather than later.

Undead Burg Bonfire in a room that will quickly be familiar. The broken stairway looks like a generic scenery element, and nothing particularly entices you to look up.

Soon after Taurus is the Hellkite bridge, which the player will start to walk down and then be slaughtered by fire from an unseen dragon, who only after this will take up his visible perch at the end of the bridge.  A near-unavoidable (without prior knowledge) death, especially immediately after giving the player a lot of souls from a boss, seens oddly harsh at first.  They may wonder what the lesson is here.  Still, the player is now adept at getting back to Taurus’s room, and retrieving souls from the bridge is likely.  The firm suggestion that the bridge may not be the right path will at least ensure the player meets Solaire on a nearby platform.  Eventually though, they must brave the bridge—it is the only path, and the church is visible behind it—with the notion of either killing the dragon or somehow getting past it.  It turns out that either of those options is possible, but a third will present itself: a route from the midpoint of the bridge, under the dragon, and into the Parish.  The purpose of the initial fiery death was to increase the tension surrounding the decision to attempt this.

A leap of faith: all you have to do to advance is prove that you’re willing to charge headlong onto this bridge. Once you do, the game is satisfied and doesn’t even require you to fight the dragon.  Anything other than a sprint to the the midpoint will usually get you  incinerated.

A huge reveal, showing the player what the Dark Souls level design is all about.  Just before entering Parish, a new connection opens to the room with the broken stairway, and the same Bonfire the player has been using ever since heading out of Firelink.

Undead Parish

Once again, at the start of a new area, the path is laid clearly ahead for a while, with the biggest change being some new enemy types.  There is a formidable gatekeeper encounter involving a number of Hollows and a Fang Boar, an advanced enough enemy that it’s more like a miniboss at this point.  The room contains a lot of geometry that can be exploited to remain safe from it.

Literal “gatekeeper” in that the portcullis at the back will close when you enter. Experienced players can charge past everything and get under it before it closes, shortcutting much of the area.

After some mostly-linear rooms, the player arrives at the chapel, which contains a large number of interesting things:

  • Across a short, quiet bridge from the hall, a safe area with another bonfire.  Only the second that’s been seen since leaving Firelink.
  • An elevator that provides a shortcut all the way back to Firelink, the second reveal that shows off the connectedness of the world.
  • Near the bonfire, Andre the blacksmith, likely the first one found.
  • Two more paths: the gate (currently closed) into Sen’s Fortress, and past an imposing Titanite Demon into Darkroot Garden.
  • The first Fire Keeper Soul, a very rare item.  Its dramatic placement on the altar indicates that it is something special.
  • The stairway up to the bell tower, with the boss and ultimate goal of this area.
  • Two more non-respawning enemies in the chapel hall (the large Berenike Knight near the altar, and the Channeler shooting from the balcony).

The Knight also has a Titanite Shard as a guaranteed drop, possibly the first one found. This is given immediately before encountering the blacksmith, as a nudge to try the weapon upgrading system.

In addition to the usual role of non-respawning enemies, providing tangible progress through the zone, these two have an added function.  Killing them makes the chapel hall into a safe place.  Since it’s directly connected to Firelink (a fact which already brings a huge sense of relief and comfort), it is now an extension of your home, stretching all the way to the blacksmith’s area.  Making you conquer your new space is much more rewarding than simply presenting it empty.

The boss encounters themselves are not the focus of this discussion, but the Gargoyle pair guarding the bell tower is by far the hardest yet.  The nearby connections to Firelink and a blacksmith provide easy opportunities for improvement and preparation if the fight poses a problem.  A thorough exploration of the area also will allow for two NPC helpers to summon if needed, which is unusual.  After defeating the Gargoyles, the ringing of the first Bell provides the first major success of the game.

The way forward will be into the Depths.  This requires going through Lower Undead Burg, whose entrance is a bit out of the way, but the Basement Key that was found in the Parish spells out the location in the item description.  I won’t say too much about Lower Undead Burg.  It’s a small and generally linear area whose most notable feature is a hard boss fight (Capra Demon) which is often a sticking point for new players.  It reinforces the themes seen so far by including shortcuts back to the Undead Burg bonfire and back to Firelink.  Killing Capra Demon gives the key to the nearby entrance to the Depths.

The Depths and the Curse

Descending into the Depths presents nothing unusual for short while, merely a few new enemy types.  The layout of the level is still interesting in that you have to either encounter the giant rat, or drop into the sewers (probably accidentally), but not necessarily both, to get to the end.  It does add a new twist with a bonfire behind a locked door, but the key is only separated from you by a few ordinary rats.

The key to Depths bonfire can be grabbed through the bars, before having to find your way to the end of the area.

A preview of an eventual shortcut from the bonfire. The immediate exit to this room is much less obvious (a downward ladder hidden behind crates), ensuring that the player tries the door.  This is important because the shortcut will be necessary for reasonable attempts at early Blighttown.

The big drama of the Depths is what will happen to most people shortly after their fall into the sewers:

The sudden appearance of a debuff that persists through death is shocking. It pulls the rug out from under the player by breaking the sanctity of the bonfire respawn for the first time.

They’re comfortable enough with you now to try a harsh presentation of a side quest: suddenly you have a big problem to solve before you can continue any further.  Of course, if you either never drop, or never get hit twice (probably) by the Basilisks’ breath attack, your reward is skipping this quest.  This is a case where there doesn’t have to be a backup plan to get the player to a certain path: never getting Cursed and simply continuing on is perfectly acceptable.

The reason I call decursing a side quest is that it will play as one for many players.  If you have a Purging Stone already, this won’t happen, but they are costed to make it unlikely that you will.  You could have bought one for 3,000 Souls from Oswald, whom you met after Gargoyles (when that was likely to be an entire level’s worth), or for 6,000 from the Undead Merchant near the Depths entrance (same).  Prudent types who made the investment regardless are rewarded by removing the Curse immediately, but this is set up so that a lot of players will have to make the trek out of the Depths with reduced HP, which is a significant task.  The player will also have 0 Souls after having been Cursed (safe retrieval from the sewer is unlikely), so they’ll have to farm up 6,000 and/or burn any soul items they saved.  After all this is done, the reward is having to go back to the Depths and make it through safely this time, with an intense fear of Basilisks, the only enemy so far that can inflict a fate worse than death.  Getting Cursed again would be crushing, but it’s unlikely as Basilisks are not too difficult to fight.  The fear of the Curse, more than the reality, is used to give tension to the level.  Notably, the eventual shortcut from the Depths bonfire (pictured above) bypasses the Basilisk tunnels.  The game is not mean enough to make you risk another Curse during your repeated runs to attempt Gaping Dragon or progress in early Blighttown.

The torch-wielding Hollows may not have seemed like much on the way in, but they can likely kill you in one combo if you’re trying to exit the Depths with half HP in search of a Curse solution.

The merchant is positioned so that you can’t miss her as you run back towards the Depths. This is important as Oswald and Ingward are more obscurely located, and the player could be stuck if they don’t find one of the three people who can help them with the Curse.

One concern about this otherwise well-executed diversion (and to be sure, a side quest whose only purpose is to instill fear with a painful lesson is not easy to pull off well) is that the reference to New Londo seems like a trap.  In theory the player could trek all the way to Ingward to be decursed for free, but that would require fighting through one of the hardest combat rooms in New Londo (a set piece where ghosts trap you in a room), at low level, with half HP.  Especially to a player who doesn’t remember on what merchant they saw a Purging Stone, finding the healer may well look like the intended path.  They know where New Londo is, and the curse lets them kill ghosts, reinforcing that this is the right idea.  The signals lead you this way, by far the least practical way of getting decursed, which is a misstep.

That the Curse allows you to damage ghosts is a clever interaction, but it doesn’t flow neatly into the game at this point.


Blighttown opens with another gatekeeper encounter that introduces important enemies—not the two club-wielding giants, but the bridge with two Infested Ghouls on it and a Blowdart Sniper firing on you from the darkness.  As always, this is designed to ensure you probably don’t advance without 1) seeing the Ghouls’ attacks, including a dangerous grab, and 2) figuring out what in the world keeps debuffing you with Toxin out of nowhere.  The game is being gentle by showing this to you in close range of the entrance.  The later and more devious placements of Snipers on difficult-to-reach ledges is fair because the possibility was demonstrated up front.  Springing that on you deep into a segment is the sort of “difficulty” you wouldn’t find here.

Once you notice that the tiny white darts are debuffing you with Toxin, you can locate the direction of the source. The first one isn’t too far off the main path.

As with any non-respawning enemy, the Snipers are also used to mediate progression throughout the difficult segments between bonfires in Blighttown.  Finally, players who both diligently explored the Depths and also read their item descriptions will have the poison-blocking Spider Shield as a nice reward to give them a leg up along the way.

The Snipers are fragile when you hunt them down, but enemy placements are designed to result in Ghoul ambushes just as you move in for the kill.

The bottom of Blighttown is a large area with a generally simple layout.  It’s a large swamp, but easy to get lost in due to the darkness.  Exploring well reveals a variety of items and potentially a well-hidden entrance to an optional side area (The Great Hollow).  Hopefully at some point the player happens to look up and realize that they can see all way back to the surface, and furthermore, that the giant stone pillars in the swamp are in fact the bases of the huge buttresses supporting the city they have been exploring.

The angled buttresses supporting the city, and the wooden structure you climbed down to descend into Blighttown.  It’s a little hard to recognize Firelink at the surface.

The same Blighttown swamp and pillars, as seen from Firelink Shrine.

It’s possible that, in exploring, the player will find the waterwheel and the exit route back to Firelink before finding the boss.  This is generally fine if those two things are done in reverse order.  If they leave Blighttown without ringing the second Bell, they know they’re not done and have to go back, and discovering the path out will still be useful.  The fact that the boss’s area is named as a new zone when you enter it (“Quelaag’s Domain”) may be minorly confusing for someone specifically hunting for the Bell, as an NPC back in town said that it was in “Blighttown”.

Return to Firelink and the Key to New Londo Ruins

The waterwheel climb is a difficult but generally linear path to the Valley of Drakes, which is a new area.  But right next to you is an entrance back to Firelink, in yet another major shortcut reveal (see below regarding what happens if the player tries heading along the Valley instead).

The arrival back at Firelink is a rare use of plot drama:

The triumphant return home is cut off when you arrive and find that the place was burglarized.

The Fire Keeper has been murdered.  Furthermore, this is a serious practical problem, since you need to touch a bonfire to consummate your escape from Blighttown.  After calming down a moment, you remember that the practical problem is fine.  Your home has a second bonfire now—you can take the elevator to the Parish bonfire and rest there, without having to fight a single enemy.  On the way, a bizarre new NPC, Kingseeker Frampt, tells you some of the plot, one of the few times that happens in the whole game.

Dark Souls’s characteristic storytelling is through items, not through plot that happens to your character. Anastacia’s set, found on her corpse, tells you her history.

There are three ways to know that the next leg of the game will be to head into Sen’s Fortress.  The cutscene after ringing the second bell shows you, if you remember what the gate looks like.  Frampt tells you if you talk to him.  But most importantly, the game has ensured that the only place you feel safe now is the Parish bonfire, which is directly in view of the Fortress gate.

A door whose key costs 20,000 souls is a message that exploring most of Darkroot Garden is probably not required, or even expected, yet.

With Firelink out of commission, the only candidate for a temporary home base is the Parish bonfire, which happens to look directly into the now-open mouth of Sen’s Fortress.

One item acquired on the way from Blighttown back to Firelink, whose significance cannot be overemphasized, is the Key to New Londo Ruins.  It opens the gate from Valley of Drakes (right near the Blighttown exit) to New Londo (right near the Firelink elevator).  Your home base is now expanded to not just the Firelink and Parish bonfire areas, but the Valley of Drakes, itself a small secondary hub that gives direct access to Blighttown and Darkroot Basin.  It shrinks the world by putting all currently important areas within a few minutes’ run from each other.  Later on the game, Dark Souls will give in and provide a form of fast travel, but only at an appropriate time, after there has been ample time to master the known areas.  For the first half of the game, the goal of maintaining practical travel between all the areas while growing the world, without offering any transportation, is accomplished solely through ingenuity in the level design.

The Key to New Londo becomes even more interesting when you ask: why does it exist? That is, why is it a key instead of a one-way shortcut door (an example of which was recently seen in the Depths)?  The key is in plain sight on the path back to the door, producing the same result as the standard shortcut mechanic.  The reason is that there is another way to open it, the Master Key.  The significance of this one junction that connects so much of the world together is that an advanced player using the Master Key can have access to all those areas from the very start (for example, starting the game and going directly to fight the Gargoyles and Quelaag without ever passing through the Burg, most of Parish, the Lower Burg, the Depths, or most of Blighttown).  The Master Key opens a few other doors, but this is the one that matters.  This one door is keyed so that the Master Key can exist as an option which, true to its name, unlocks the whole game.

The key chest is in plain view at the back right of this small platform, but a player who’s not looking might miss it as they turn left to the Blighttown exit.

The only loose end that must be tied up, given the choice to use a key here, is what happens if the player misses it.  Creating a sticking point just before the triumphant return to Firelink would be a problem.  But look at how cleverly this is handled: a player who arrives at the New Londo gate and finds it locked has only one new way to go, down the Valley of Drakes.  With lower New Londo still flooded, that exit from Valley is blocked, and the only exit is an elevator up to Darkroot Basin.  Basin has a few exits, but coming from this route you will arrive right next to the exit to Parish.  The other exits are much less accessible, being guarded by the Hydra and by Havel the Rock.  There is even a bonfire along the way so that, if you probe at other parts of the Basin first, your death doesn’t send you to back Blighttown.  There is only one path with little resistance along it (the biggest problems are a Black Knight and a Drake or two that you can simply run past), and it leads directly to the Parish bonfire, exactly where you’re meant to wind up.

The path from the Blighttown exit all the way to Basin and Parish has conveniently few significant combat obstacles.


The Key to New Londo example is a perfect demonstration of how Dark Souls‘s exploration-oriented gameplay works far more subtly than simply letting you roam randomly in a large, unfriendly world.  It carefully considers where the typical player is most likely to go next, and how to carefully cue them to (unaware of the game’s thumb on the scale) go in the direction that will be best.  The player is absolutely free to choose, and can choose differently, a sacrosanct axiom of Souls.  But this world was created by people with the staunch hope that your choices will always see you through to the way forward.

Everything about Dark Souls, from its reputation, to its cold use of player deaths/failures as a teaching tool, to the pervasive swagger of its fans, suggests hostility at every turn.  But make no mistake.  Underneath the facade, like a devoted parent, this game desires one thing above all: to see you succeed.

Hero Academy Mechanics Guide


In preparation for Hero Academy coming out on PC, I’m collecting and filling out the mechanics guides I used to post on the forum. If you’re 100% brand new to HA, you don’t need to read this first, just use the in-game tutorial and play a few turns to get a sense of what’s going on. But as you start playing, you’ll encounter units/mechanics/races that you haven’t seen before, and sometimes you just want to check what they do without having to use trial and error in the middle of a game. Even experienced players sometimes need to remember how some unit’s special attack works if a critical turn depends on it. The in-game tooltips are useful, but I wanted to provide something more detailed and comprehensive, and in one easy-to-reference location.

Without Further Ado:
General Mechanics
Dark Elves
The Tribe
Team Fortress
The Shaolin

General Mechanics


  • 5 AP per turn (3 AP on first turn of the game in competitive play).
  • 1 AP can be used to: deploy a unit to deploy square, move a unit, attack a unit/crystal (NB: not limited to once per turn), heal or revive a unit, use a unit special ability, equip a unit with an upgrade, use a one-time item, or throw a tile from your hand back into the bag to draw a random one.
  • A unit that reaches 0 HP is knocked out, but can be revived by an ally. It vanishes from the board permanently if any player moves a unit onto it (“stomps” it) or destroys it with certain spells (see below), or if its owner takes a full turn without reviving it.
  • A player loses upon having all of his crystals destroyed, or by having all of the units in his deck stomped or knocked out (the game will end even if he has an item left that can revive a KO’d unit).
  • Tiles are drawn in random order throughout the game, except that your starting hand always has at least 3 units.

Map features:

  • Crystal. Have varying amount of HP depending on map.
  • Deploy square. Enemy units cannot move onto it, even to stomp. An enemy unit adjacent to a deploy square can attack a corpse that’s on the square (which destroys it), without moving onto it. A Necromancer, Witch, or Inferno can also destroy a KO’d unit on a deploy square.
  • Attack square (red sword). A unit on the square has +100 attack power, applied before all %-based bonuses.
  • Defense square (blue shield). A unit on the square has +20% physical resist.
  • Magic Defense square (dark blue helm). A unit on the square has +20% magical resist.
  • Assault square (purple gem). While a friendly unit or Bamboo is standing on this square, all attacks against enemy crystals do an extra 300 damage. Multiple assault bonuses stack. The bonus is reduced accordingly if the crystal takes AoE or chain damage. The bonus damage has no type and is not affected by any resist.
  • Teleport square (light blue dot). A unit on the square can use 1 AP to teleport to the other teleport square, if it is open.
  • Speed square (light blue boots). A unit on this square has +2 move.
  • Barbed Crystal. Can use 1 AP to deal 300 P to an enemy within range 3. Only usable once per turn.
  • Train. When a unit moves (or is knocked or Ninja-swaps or Bamboo appears) onto a train switch, the train deals 400 P to every unit on the track and knocks them backwards one square. Friendly units will not be damaged below 100 HP. A unit that is knocked out by the train is immediately destroyed.


  • Range is counted orthogonally only, not diagonally (e.g. a unit one square diagonally from another is at range 2). In this guide I use “adjacent”, “range 1”, and “melee” interchangeably. “At range” means not in melee. “AoE range” (area of effect) is anything adjacent or immediately diagonal to the target (i.e. a 3×3 square centered on the target). A “knight’s move” refers to a chess knight (range 3 not in a straight line).
  • Each unit has an attack power, and a damage type (physical or magical). Ordinary attacks hit for damage equal to the unit’s attack power, but some multiply by certain percentage or add other bonuses.  The attack power is what’s shown in the unit’s info window.
  • Each target also has a physical and magical resist, which reduce incoming damage of the appropriate type by a fixed percentage (see Appendix for details). The only things that affect a crystal’s resists are Paladin auras and the Annihilator debuff.
  • Ranged attacks (but not ranged friendly actions) require line of sight to the target. Enemy units/crystals block line of sight, but not friendly ones. An attack to an immediately diagonal square can’t be blocked, but an attack to a square a knight’s move away can be blocked in either intervening square.
  • A unit that can act on a corpse (healer reviving a friendly corpse, or Necromancer/Witch/Wraith/Shadow destroying an enemy one) can’t choose to move and stomp the corpse instead–clicking on the corpse causes you to act on it.  However, if the unit does not have line of sight to the enemy corpse, it will move and stomp instead.
  • Knockbacks only occur if the target square is open, otherwise the unit simply doesn’t move. If a unit is KO’d by the knockback, it still moves. A unit can’t be knocked onto an enemy start tile, unless it’s also KO’d (in which case an enemy spawning there will stomp it).
  • A unit can wind up on an enemy deploy square if it’s KO’d onto it and then revived, or if it’s a Wraith that spawned from a corpse on that square. It will block the enemy from deploying units there so long as it remains.
  • A corpse on its own deploy square will be stomped automatically if a friendly unit deploys there.
  • All damage/healing is rounded to the nearest 5 (with 2.5 rounding down).
  • Units cannot target themselves with heals or other buffs.
  • Every race starts with 28 total tiles, except for Team Fortress who has 24. Team Fortress has 18 units (effectively 20 due to two respawns), the Tribe has 14, and other races have 13.
  • When you throw tiles back into the bag, you draw from what’s left in the bag (as long as there are enough). So you won’t get back the ones you’re putting in, but you might get other copies of the same thing.

Knight (3):
1000 HP
20% P resist
200 attack (P), range 1
Attack knocks target back one square.
Move 2

Archer (3):
800 HP
300 attack (P), range 3
50% damage in melee, 100% damage at range.
Move 2

Wizard (3):
800 HP
10% M resist
200 attack (M), range 2
Attack can chain twice if each target is in AoE range of the previous (see Appendix for details). Main target takes 100% damage, first jump does 75%, second jump 56%.
Move 2

Cleric (3):
800 HP
200 attack (M), range 2
Heals for 300% attack, revives for 200% attack, range 2
Move 2

Ninja (1):
800 HP
200 attack (P), range 2.
200% damage in melee, 100% damage at range.
Can use 1 AP to swap places with any friendly unit.
Move 3

Sword (3): +50% attack
Armor (3): +20% physical resist, +10% HP
Helm (3): +20% magical resist, +10% HP

Scroll (2): Targets one friendly. Unit has 300% attack power for one attack/heal.
Inferno (2): Targets a 3×3 area. KO’d enemies are destroyed, and other enemies take 350 M.
Potion (2): Targets one friendly. Heals for 1000. In addition, can revive a KO’d unit for 100.

Racial Passive: None

Void Monk (3):
800 HP
20% P resist, 20% M resist
200 attack (P), range 1
100% damage to target, splashes for 66% damage to any enemies adjacent (not diagonal) to target.
Move 3

Impaler (3):
800 HP
300 attack (P), range 2
If target is at range 2 in a straight line, pulled one square towards Impaler.
Move 2

Necromancer (3):
800 HP
200 attack (M), range 3
Can use 1 AP to turn any KO’d unit into a Phantom at range 3, including on an enemy start tile.
Move 2

Priestess (3):
800 HP
200 attack (M), range 2
Heals for 200% attack, revives for 50% attack, range 3
An unit damaged by the Priestess has 50% attack power for its next attack/heal (doesn’t affect Barbed Crystals).
Move 2

Wraith (1):
800 HP
250 attack (M), range 1.
Can use 1 AP to consume a KO’d unit at range 1. Increases current and max HP by 100 and attack power by 50 (bonus can only be obtained 3 times per game although you can continue to consume).
Can deploy onto any KO’d unit (stomping it) instead of a deploy tile, unless the KO’d unit is on an enemy deploy tile.
Move 3

100 HP
100 attack (M), range 1
Does not leave a corpse when killed.
If spawned on an enemy deploy tile, will be automatically destroyed if an enemy deploys there.
Move 3

Sword (3): +50% attack
Soulstone (3): Passive life leech increased to 66% (see below), +10% HP (bonus applied to base HP only)
Helm (3): +20% magical resist, +10% HP (bonus applied to base HP only)

Scroll (2): Targets one friendly. Unit has 300% attack power for one attack/heal.
Soul Harvest (2): Targets a 3×3 area. Enemies take 100 M. All friendly units in play gain D/(N+3) current and max HP (reviving KO’d units in the process), where D is the total damage dealt to non-Crystal/Bamboo units by the spell, and N is the number of friendly units in play.
Mana Vial (2): Targets one friendly. Heals for 1000 and increases max HP by 50.

Racial passive:
Units are healed for 33% of any damage they deal to enemy units (not crystals).

Paladin (3):
900 HP
10% P, 10% M resist
200 attack (P), range 1
Heals for 200% attack, revives for 50% attack, range 2
Self-heals for 50% of any healing done, not including overheal.
All friendly units/crystals in AoE range of Paladin gain 5% attack power, physical resist, and magical resist. Stacks from multiple Paladins.
Move 2

Grenadier (3):
800 HP
200 attack (M), range 3. Ignores LOS.
At range, 100% to target, AoE for 50%.
In melee, 50% damage, no AoE.
Move 2

Gunner (3):
800 HP
300 attack (P), range 2.
In melee, 100% damage, no splash.
At range, cone attack deals 66% damage to main target and up to two nearby targets (see Appendix).
Move 2

Engineer (3):
800 HP
200 attack (P), range 1.
Can shield a friendly target in range 3, canceling next damaging attack against that unit/crystal and any associated debuffs (Jarate can still be applied through it). Expires when it prevents any damage, when Engineer is knocked down, or when Engineer shields a different target.
Gets twice the increased benefit from special tiles as other Dwarves (see below). 140 attack, 28% resist, or 420 assault damage. 6 move when standing on speed tile.
Move 2

Annihilator (1):
650 HP
300 attack (M), range 3. 100% to target, AoE for 20%.
Target (unit/crystal/Bamboo) gets -50% P resist on the next physical attack that hits it.
AoE targets are knocked one square directly away from main target.
Move 2

Sword (3): +50% attack
Armor (3): +20% physical resist, +10% HP
Helm (3): +20% magical resist, +10% HP

Scroll (2): Targets one friendly. Unit has 300% attack power for one attack/heal.
Dwarven Brew (2): Targets one friendly. Heals for 1000 and adds 50% M and P resist lasting for one hit. Stacks with other resist bonuses. If target is also shielded, buff remains when shield is broken.
Pulverizer (2): Targets one enemy unit or crystal. Target takes 600 P. If target is a crystal, AoE for 33%, including any added damage from Assault tiles (the 200 P AoE is reduced by the target’s physical resist, but added splash from Assault tile damage is not). Splash occurs even if crystal was shielded. If target is a unit, its Armor, Soulstone, or Spike Armor (but not Helm) is destroyed.

Racial Passive:
+20% benefit to special tiles. Attack gives +120, Defense/Magic Defense gives +24%, Assault gives +360. Speed gives +3 move.

Warrior (3):
800 HP
20% P resist
200 attack (P), range 1
Infinite damage (auto-KO) if target is at or below 50% HP (25% crystal).
Move 2

Axe Thrower (3):
800 HP
200 attack (P), range 2
+200 damage after multiplicative bonuses, but before resistance, if target is above 50% HP (25% crystal).
Move 2

Witch (3):
800 HP
200 attack (M), range 3
Can use 1 AP to explode any KO’d unit in range 3, doing 50% damage (M) AoE around target. Range 3. Damage is increased proportionally by any HP upgrades the target has over its starting HP.
Move 2

Shaman (3):
800 HP
10% M resist
200 attack (M), range 2
Heals for 200% attack, revives for 50% attack, range 2
Heal/revival can chain twice if each target is in range 2 of the previous (see Appendix for details). First jump is 50% effective, second is 25% effective.
Can target full-HP units as long as at least one jump is to a damaged unit. Will never chain to caster.
Move 2

Chieftain (2):
1000 HP
200 attack (P). 2 attacks, Whirlwind (range 1) and Charge (range 4)
Whirlwind: Target takes 100%, AoE around Chieftain for 66%. Enemies 2 squares away from Chieftain (orthogonally, diagonally, or a knight’s move away) are pulled in by one square (see Appendix) . Pulled units take no damage, but are now in AoE range for next Whirlwind.
Charge: targets enemy unit or crystal, directly left or right only within range 4 and subject to LOS. Chieftain moves to square immediately in front of target (must be free), and deals 100% damage to target. No Whirlwind AoE or pull.
Move 2

Sword (3): +50% attack
Spike Armor (4): +10% HP, deals 33% of any damage taken (as the same damage type) back to the attacking unit or crystal. Reflected damage cannot take attacker below 100 HP. Reflected damage will consume one-hit buffs like Engineer shields and Beer. Reflected damage is applied simultaneously with Dark Elves’ Life Leech (often resulting in no net effect).

Haunch of Meat (4): Targets one friendly. Heals for 700 and gives 275% attack power for one attack/heal.
Typhoon (3): Targets a column. All enemy units/corpses get knocked 2 spaces away from the targeted column. Units in the targeted column are knocked back towards the enemy start, unless they’re in the back row, in which case they get knocked forward (so you cannot “split” enemies in the two back rows). Crystals and friendly units block movement.

Racial Passive:
When a friendly unit is stomped/destroyed by an enemy, all friendly units in play that are not KO’d gain Bloodlust (+50% attack power). Whenever any unit with the buff attacks or heals, the buff fades from all units.
Does not stack. Does not activate when a friendly corpse expires at the end of your turn.

Spy (2):
650 HP
200 attack (P), range 1
If attacking a unit from behind, 400% damage.
Cannot be targeted for attack by a unit at range (except as a corpse), except a Pyro. Can be targeted by a unit in melee, by a Pyro, by a spell/item, or by enemy attacks that target corpses. Takes AoE damage normally if something else is the target.
Move 2

Scout (2):
650 HP
100 attack (P), range 1
When hit, jumps one square away from attacker (diagonal if attacker is knight’s move away). When the attacker has its own push/pull effect, the Scout behaves like any other unit. When hit by Chieftain, the Scout jumps back and is then pulled, usually winding up where he started.
No AP cost to deploy (and can be deployed at 0 AP).
Move 4

Soldier (2):
800 HP
300 attack (M), range 3
At range, knocks back target away from Soldier (diagonal if target is knight’s move away).
Move 2

Sniper (2):
650 HP
100 attack (P), horizontally with unlimited range, or any direction in melee range
Kneels at end of turn, giving +250 damage (before resistance) to first shot of subsequent turn, unless Sniper takes damage or moves before shooting.
50% damage against crystals (applies to Assault bonus, but not to kneeling bonus).
Move 2

Medic (2):
800 HP
10% M resist
200 attack (M), range 2
Heals for 200% attack, revives for 50% attack, range 2
Self-heals for 50% of any healing done, not including overheal.
Link to full-health ally in range 2 to increase their attack by 50%. Link lasts until target and Medic are more than 3 apart, Medic links/heals/attacks a different target, Medic takes damage, or target is KO’d (it does not break if target takes damage or Medic heals target).
Two medics can link the same target, resulting in 225% normal damage.
Move 2

Engineer (2):
800 HP
200 attack (P), range 2
Can permanently upgrade an ally’s attack power by 30%, range 2.
Move 2

Heavy (2):
1100 HP
100 attack (P), range 3
At range, cone attack deals 100% damage to main target and up to three nearby targets (see Appendix).
In melee, 200% damage, no splash.
If consecutive actions in one turn are spent attacking, +50 attack power each shot, not modified by Engineer/Medic bonuses.
Move 2

Pyro (2):
800 HP
300 attack (M), range 2
If target in melee range, also hits square behind for 100%.
Move 2

Demoman (2):
800 HP
200 attack (M), range 3, ignores LOS
At range, 100% to target, AoE for 50%.
In melee, 100% to target, no AoE.
125% damage against crystal, if crystal is main target (including Assault bonus).
Move 2


Jarate (2): Targets one enemy unit. -175% to both resists for one attack.
Sandvich (2): Targets one friendly unit. 500 heal, +2 move for unit’s next move (buff no longer falls off if the unit is moved by any effect).
Respawn (2): Targets one KO’d friendly unit. Unit is returned to hand (ugprades are reset). No AP cost (and can be used at 0 AP).

Racial Passive:
1 AP refunded when an enemy unit is stomped or destroyed by a TF2 unit (but can’t move with 0 AP left). Doesn’t trigger from a train kill.


Monk (3):
1000 HP
200 attack (P), range 1
If target is unit, maximum HP reduced by 25% (computed from target unit’s base HP before upgrades). Does not stack. Expires if Monk attacks a different target, but not if Monk or target is KO’d. Due to a potential bug, debuff does not apply if Monk KO’s an un-debuffed target, but his previous debuff is still canceled.
Move 3


Windblade (3):
800 HP
200 attack (P), range 2
Attack jumps to up to 2 targets in AoE range of main target (see Appendix for details).  One jump does 50% damage and the other 25%.
Move 2


Poisoner (3):
650 HP
200 attack (M), range 3
If target is unit, gets -50% to both resists for one incoming attack.
Move 2


Taoist (3):
800 HP
200 attack (M), range 2
Heals for 200% attack, revives for 50% attack, range 2
Heal clears any debuffs from allies.
Attack clears any buffs from enemies. Against Dwarf Shield, breaks shield but does no damage and gives no combo point. Against Beer, clears Beer before doing damage. Cannot remove Soul Harvest or Wraith HP upgrades, but does remove Mana Vial.
Move 2


Shadow (1):
1000 HP
20% M resist
100 attack (M), range 3.
Can revive any corpse (friendly or enemy) with 100 HP, range 3, not on enemy deploy tile (also, cannot melee-stomp an enemy on a deploy tile).
Actually destroys the corpse and creates a new friendly unit of the same type, so buffs/debuffs/upgrades are not carried over (and a new Monk can debuff a second target without canceling the first).
Move 3

Revived Units:
Revived units keep the racial properties of their old race. In the case of Tribe, the Tribe unit alone will Bloodlust if any friendly is stomped/destroyed. In the case of TF2, you get 1 AP only if the TF2 unit stomps an enemy. Revived Shaolin units can combo with other units on your team, but units of other races don’t generate or use combo points.
You can give Shaolin equipment to units that couldn’t normally wear it. Tribe/TF2 units will not have art for Armor/Helm upgrades (opponents watch out for this). Dark Elf units with Armor will use the Soulstone art. TF2 units with Swords will use the Engineer upgrade art.
A Revived TF2 Engineer can give the 30% buff to Shaolin units, which stacks multiplicatively with Sword upgrades.

Sword (3): +50% attack
Armor (3): +20% physical resist, +10% HP
Helm (3): +20% magical resist, +10% HP

Combo Potion (2): Targets one friendly. 1000 Heal, and unit’s next attack adds one extra combo point to the target (added before the attack).  Against Dwarf Shield, attack does no damage but adds 1 combo point. If unit has already hit the target that turn, it still gets +200 damage and +1 combo point.
Combo effect does nothing on non-Shaolin units.
Dragon (2): Targets any square. Deals 600 M to target square, and and 450, 335, and 255 M to the three squares in front of it (towards Shaolin side).  Only damages enemy units/crystals or Bamboo.
Bamboo (2): Places a Bamboo on one empty square or corpse (stomping it, not allowed on enemy deploy tile).  Bamboo has 1200 HP, 0 resists, and can be attacked by players of either team (only blocks LOS for enemies).  Bamboo are not units for purposes of buff/debuff eligibility, Tribe damage bonuses, and preventing the end of the game, but they can activate Assault tiles.
Due to a current bug, a Taoist attacking a friendly Bamboo will remove any combo points before attacking.

Racial Passive:
When a Shaolin unit attacks a target, if it has not previously attacked that target that turn:
1) The attack does bonus damage if there are any combo points already on the target (combo points clear every turn). 1CP: 200, 2CP: 400, 3CP: 800, 4CP: 1600, 5CP: 3200.*
2) One combo point is added to the target.
The bonus damage is not affected by the attacker’s attack power, but it is affected by the target’s resist (it has the same type as the underlying attack).
If target is knocked down, no combo point is added, but this is of no importance.
Dwarf Shield prevents application of CP, except the 1 bonus CP from a Combo Potion.

*An attack with 6 CP is theoretically possible, but very unlikely in a real game (there isn’t even a graphic for a target with 6 CP on it). It does enough damage to destroy anything in the game.


Gunner/Heavy Splash: A Gunner/Heavy on “o” shooting a target at “x” will splash to targets at “1” first. It will splash to a target at “2” only if there’s nothing at “1” in front of it, and similarly for a target at “3”.

Range 2 (Gunner or Heavy)



Range 3 (Heavy only):

?_?  <–  Unclear what causes these to be only sometimes hit.

o1_   <–  It appears that this 1 is only hit if the 2 behind it is full.

(Note that the Heavy can hit 4 targets in this arrangement)

Wizard/Shaman/Windblade Chains


  1. The two jumps are computed sequentially–i.e. the first jump is determined without reference to what second jumps are possible. So the game checks all targets in AoE range of the main target, picks one using the priority sequence below, and then picks a third target in range of that one.
  2. In practice, this means that sometimes the first jump will go into a “dead end,” even though two jumps might have been possible along a different path.
  3. Shorter jumps are always preferred. Remember that adjacent jumps are shorter than diagonal ones.
  4. Amongst equal range jumps, the “straightest” is preferred (i.e. a jump in the same direction as the initial shot is favored over a 90-degree turn). Note: even if the first jump and initial shot were at different angles, the second jump is compared to the initial shot.
  5. Amongst equal-angled turns, a target further north will be favored.
  6. If all factors above are equal, a target further east will be favored*.

*NOTE: this is an asymmetry between the two players. The first player’s Wizard, shooting north or south, is more likely to have his shot turn toward the enemy backfield. The first player’s Shaman is slightly more likely to chain heal all way to a unit on the front line.


  1. As with the Wizard, the two jumps are computed sequentially (with range 2 instead of AoE range).
  2. The Shaman will always favor KO’d units over partial-HP units over full-HP units. It does not differentiate among different levels of partial HP. The Shaman will never chain to himself.
  3. Among targets with the same HP status, apply Wizard rules 3-6.
  4. If the Shaman’s chosen path consists only of full-HP units, the cast is not allowed. In particular, if the first jump would be to “dead end” full-HP unit, the cast is disallowed even if a different path would allow a heal or revive on the second jump.


Unlike for Wizard and Shaman, Windblade chains are both made from the main target.

  1. Compute first jump as though it were a Wizard. This target takes 50% damage.
  2. For second jump, repeat, ignoring first jump target (i.e. choose the target that would be second-higher priority under the Wizard rules). This target takes 25% damage.

Chieftain Whirlwind: When a Chieftain at “o” attacks:


Melee target takes 100% damage.
Other x and y take 66% damage.
1 pulled to x if x is free.
2 pulled to y if y is free (priority to whichever target is further west*, or further south).

*NOTE: this is another asymmetry between the players.

Order of Operations for Damage:

A) What is unit’s attack power?

  1. Base unit attack power (from above list).
  2. If standing on Attack tile, +100. If Wraith, +50 for each consume buff.
  3. All multiplicative modifiers, in any order (Sword, Scroll, Meat, Priestess debuff, Paladin aura, Medic buff, Engineer upgrade).
  4. If Heavy, +50 for each repeat-fire buff.

B) Damage done by attack

  1. Start with attack power (which can also be found in the unit’s tooltip window).
  2. If heal or revive, apply healer’s heal or revive coefficient and end. Otherwise continue:
  3. Range modifiers: Melee: 50% for Archer/Grenadier, 200% for Ninja/Heavy, 400% for Spy. Ranged: 67% for Gunner.
  4. Assault Tile bonus if target is crystal or Pulverizer AoE from crystal.
  5. All other multiplicative modifiers in any order: AoE/Chain reduction (Wizard, Void Monk, Grenadier, Annihilator, Pulverizer, Chieftain, Demoman, Windblade), Sniper crystal penalty, Demoman crystal bonus.
  6. Axe Thrower +200 or Sniper +250 bonus, or Shaolin combo bonus damage.
  7. Physical or magical resist of target (does not apply to Assault Tile damage).
  8. If Monk attack would leave target at more than max HP (due to Monk debuff), increase damage to leave unit at max HP.