Game Design Deep Dive is a new series from Gamasutra, with the goal of shedding light on specific design features or mechanics within a video game, in order to show how seemingly simple, fundamental design decisions aren't really that simple at all. See last week's installment here.
This explanation pertains to the sanity meter feature in Amnesia: The Dark Descent. But before delving into the meter itself, I need to explain the darkness mechanic as well, as these two components are very much connected.
Early on in the development of Amnesia, around the time when we were applying for funding from Nordic Games, we thought a lot about ways to develop a game without having to spend too much money and resources. An idea popped up: "Why not get rid of enemies altogether?" Instead of having physical monsters present, we'd just have the dark as an enemy. You wouldn't be running from monsters, but running from this pitch blackness. This would help us cope with our limited resources: When we started out the project, there were only three people, including a sole artist at the time.
Stages of sanity - final version
The issue was now to figure out a way in which the player could interact with the darkness. We began by designing, on paper, a mechanic in which players could manipulate things like windows, candles, and mirrors in order to control the dispersion of light and dark. Being in the dark too long meant your health would suffer and the player had manipulate the environments in a way that kept them away from the shadows. With this as our foundation, we then started to design some maps.
Turned out it didn't work at all!
The problem is that once you start mixing something that's very aesthetically meaningful -- like darkness and light in a horror game -- with a game mechanic, you ruin the aesthetics and the mood. That's unacceptable in a horror game. The lighting was all off, and the environments made no sense at all. In order to get the proper gameplay going, rooms had to feature countless mirrors, awkwardly-sized furniture, and nonsensical layouts. It was just weird.
This was the point when we first came up with the sanity mechanic. Instead of having the darkness being directly linked to player's health, it made a lot more sense if it affected their mental well-being. This also made us decide to add actual monsters and other kinds of (non-darkness-related) puzzles as well.
Partly inspired by games like Call of Cthulhu and Eternal Darkness, and partly by a continuation of features from our Penumbra games, the sanity system was born. We found that it was a lot easier to fit the concept of insanity into our game. The player would lose sanity from seeing monsters, from witnessing horrific events, and of course from being in the dark. So what started out as our core mechanic -- the fear of the dark -- had been demoted to an aspect of the of the overarching sanity system.
"We made the decision that it was time to go back to what we knew best: making adventure-style games with less focus on mechanics, and more focus on narrative and ambiance."
This system worked a lot better narrative-wise, too. The scenarios got more interesting, the environments more believable, and having the protagonist fear the dark worked perfectly, narrative-wise. Most importantly, though, was that it allowed us to compress the environments. Good horror often comes from claustrophobic spaces and thrives on low visibility. The original darkness system required the player to survey large areas in order to form proper strategies, whereas the sanity system worked well in cramped spaces. This also meant a lot of player attributes, such as jump height and speed, got better suited for the mood we wanted to portray.
Side Note: The impact that basic mechanics have on space and character is quite interesting and worth thinking about. Early in development, we had the idea of basing the game a lot on movement, such as climbing, extending bridges, and jumping on platforms. But as with the darkness mechanic, we found that it did not fit with the story. However, the recently-released Among the Sleep actually features just this type of gameplay and it works just fine, because both the world and protagonist are quite different.
Despite having been demoted, the darkness was still a prominent feature, as is evident in the items that we added soon after settling on the sanity system. To avoid darkness, the player needed to pick up lamp oil in order to keep the lantern going, or pick up tinderboxes to light various in-game light sources (e.g. stoves and candles).
There were also potions that replenished player sanity, and of course special potions for increasing health. This gave us pretty good dynamics in the game. The player had to manage items for keeping in the light, and keeping sanity high. At this point in the game, low sanity also decreased the player's health, connecting the health potions to the sanity dynamics. This item looting and management system became a foundational part of the gameplay at this point.
In all, we found that it was a pretty tense system, and having the player trying to stay sane (and alive) offered a surprising amount of gameplay. At this point in development, we were hoping to make a game that is less like our adventure game series Penumbra, and more of a mechanics and level-based game. In fact, Super Mario Bros. was a big inspiration. Amnesia was based on a hub system like in Super Mario 64, and the sanity and health levels were confined to a level. This made it easy for us to tweak and base gameplay around the dynamics.
There were tons of very “gamey” aspects to this sanity system. The current health and sanity levels were printed in actual numbers, messages popped up when certain amount of items had been picked up, and so forth. We were explicit with all the numbers and the gameplay was very system-focused.
But around August of 2009, we decided that we were going the wrong direction; money was running out and the game didn't have the right feel yet. We made the decision that it was time to go back to what we knew best: making adventure-style games with less focus on mechanics, and more focus on narrative and ambiance.
Sanity could no longer be this abstract metric. Something had to happen when sanity got low, as an indicator for players, but without showing it in direct numbers. We solved this in two ways.
First, we made the current level of sanity less explicit by just relaying it using an image and text that was only seen in the inventory screen. This is similar to how it is done in the early Resident Evil and Silent Hill games. It gives the player an informational gap regarding their current situation, making it harder to feel safe.
The final sanity meter is more ambiguous than originally conceived
In a more action-oriented game, getting a good idea of the current status of a stat can be good both for getting good gameplay dynamics and for upping the tension (think how tense a fighting game can be when one of the player's health is close to zero). But horror is slower-paced. It is less about making tactics based on information and more about flailing in the dark, having panicky reactions as things happen. This type of meter provides feedback that accommodates this type of play. The visuals of a brain slowly rotting, combined with a textual description of the protagonist's current state, is enough for the player to get invested in the current state of affairs -- but not enough for them to be able to craft any advanced strategies from it.
Second, we needed some sort of in-game representation of the state. This served two purposes. One was to get a general feel for the descent into madness. The other was to have an effect annoying enough to the player so that it wasn't a state they wanted to be in, but not so annoying that it made them quit.
Striking that visual balance of annoying-but-not-too-annoying was difficult. I sat for a couple of days just tuning the visuals for this -- often having to take longer pauses because of extreme nausea -- before I got something that felt right. In the end, though, we decided to pull back the effect a bit more than what had felt best. This was just to make sure players did not feel physically ill.
While you can inflict a certain level of physical discomfort in a horror game, too much will break the player's immersion. After all, horror is built upon the imagination; if you get to experience the things for real, it stops being entertainment. (Although I guess a few people would argue horror games are not very entertaining in the first place...) Horror should feel real, not actually be real.
Additionally, we also had to have some indicator when sanity was being drained. When sanity is high, the visual effects are quite subtle, meaning you do not notice it dropping. This meant we needed something that made it clear to the player that they where in a really bad spot. The goal was to give rise to a “I need to get out of this!” feeling.
Because the sanity drops often only lasted for shorter bursts, we could push the effect further in terms of unpleasantness than what we could with the insanity on. When the sanity drops, it should really get to the player. We ended up implementing an audio cue that was created by cracking egg shells in front of a microphone. This was played along with some subtle screen pulsing. It turned out to be a really a really creepy effect, and has such an emotional impact on players that there are Amnesia mods specifically made to get rid of it. It was a really, really nasty sound that gets worse and worse the longer the player's sanity was drained.
Another thing that we added to the sanity systems was sudden events that could make the sanity suddenly take a deep plunge. For instance, early on in the game a door swings open unexpectedly, and this is accompanied by a pulsing effect and a specific sound that signifies an “insanity drop event.”
We were really unsure about how these would work, as we were basically telling the player, “Boo! Now is the time when you're scared.” However, they served an important purpose, as they spiced up the situation, in terms of insanity, for the player. We felt it helped the dynamics a lot, and that was our reason for keeping them in. Playtests showed that few were bothered, which sealed the deal. These made it to the final game.
The sanity system was shaping up, and testing showed that it affected the player in just the way we wanted. However, as we started doing test runs with longer playtime, an issue propped up: The balance was way off.
When the game changed direction, it went from having disconnected Super Mario-like levels to having an interconnected game world. The levels were now set up just as in our previous games and titles like Silent Hill and Resident Evil. This meant that we could not longer easily reset the sanity status between levels. Half a year or so before release, it became apparent that this caused very serious balance issues.
"Eventually, we removed all challenge from the sanity system and it just became a feature with few gameplay implications."
Some players would have excellent experiences, fighting to stay sane and just barely making it -- just the way we wanted them to. The problem was, these were a minority of players. Others played through most of the game without noticing the sanity at all. Others still would be stuck with low sanity, an annoying state to be in, for large portions of the game. This meant that for most players, the sanity system was not working at all. Worse, for a big chunk, it made the game almost unplayable.
We first thought that fixing this was just a case of some quick tuning, but quickly learned that it would not be that easy. The way that the player's sanity changes throughout the game depends on a multitude of factors: How carefully they searched the environment, how good they were at navigating, how many tinderboxes they used, and so on. Even minor differences in monitor brightness factored in. It dawned on us that trying to get this right for a system that spanned hours of gameplay would be a close to impossible task.
In order to combat this we tried to reset the sanity at regular intervals somehow. This was done by giving the player a boost in sanity whenever they completed a puzzle. The addition worked nicely as it gave some sort of reset at regular intervals, but it did not solve the basic issue. Players could still get into annoyances between puzzles and we could not completely reset the sanity either as that did not give the right kind of feel to the system.
Eventually, we removed all challenge from the sanity system and it just became a feature with few gameplay implications. If someone went through the whole game not paying attention to their sanity level, even letting it drop all the way down, they could do that. Instead, we made extra sure the audio-visual cues were designed such a way that people were compelled to manage their sanity levels properly.
This worked a lot better than we had first expected. It made it possible for us to mess a lot more with the various variables involved. For instance, we could change the amount of lamp oil found (remember, darkness is deeply connected to the sanity system), depending on the player's past and current state. This would not have been right if it had been a competitive system, as it would upset the player's planning. But now that we had removed that, we could put all focus on expecting a certain sanity level, with a frequency of ups and downs throughout the game. It greatly enhanced the player's experience. Nobody was bothered by it anymore, and many feared going insane much more now than before. Because of the vague and random nature of the system, players made up their own theories of how it worked, and the lack of information made them more frightened about it.
A final note on the design history of the sanity system: Up until the last four months or so of development, the game featured special sanity potions. In May, we released a special preview build of the game, and a great deal of the previews complained about these potions. Previewers felt that they didn't fit and made little sense, and we agreed with this sentiment. We had thought about removing them for some time, but it was such an old feature that we were relucant to let it go. The feedback from the preview was the last nail in the coffin, though, and we decided to drop sanity potions. It was kind of scary to do this, as it required us to tune the dynamics of the sanity system all over again, but in the end it turned out great and was worth it.
The sanity system had a quite a journey throughout development. It started as a substitute for monsters -- due to team size constraints -- then became a system heavily inspired by Mario's trinket collection system, then became a game system, then an ambient background system that keeps the player on edge.
Around the time we were questioning the design of the sanity system, it started to become more and more clear to me that not everything in a game has to be competitive in nature. Sometimes, making something game-y just destroys the feeling that you're after.
I had written some blog posts on this subject before, for instance about how focusing on “fun” can often destroy certain aspects of a game. But, what really made it all click was a this post by Tale of Tales' Michael Samyn. Reading it now, it does not seem like anything special, but at the time it just clicked with me. It clarified some thoughts that had been brewing in my mind. It unleashed an idea of games that would dismiss common gameplay elements, and focus on other things instead. This is quite common now, but four years back, that was not the case. The blog post from Michael encouraged me to express my thoughts as a blog post too, and you can view it here. Some terminology and conclusions are off from what I think now, but it neatly summarizes my thoughts at the time.
"Sometimes, making something game-y just destroys the feeling that you're after."
This thinking greatly influenced a lot of the design issues made during the last half year of Amnesia's development, the sanity system being one of them. Had the zeitgeist of “notgames” not started at that time , the design would have most likely taken a totally different direction.
Looking back at it, it seems like a really obvious decision to make, and it felt natural at the time. But had the climate just been slightly different, I doubt it would had been made. I think it shows just how much a game's design direction is greatly shaped by the surrounding development environment. I also think it shows how important it is to be open to new concepts and not be to stuck in the trappings of older games. The focus should always be on creating a certain experience, and it is fine to break ingrained rules in order to accomplish this.
In the end I think I think it was successful -- the system ended up working better than we thought it would, especially considering the history behind the feature. It's actually surprising we got it to work at all!
Looking back at it all I think there are three problems I would have liked to get fixed:
We want to try something like the sanity system for our next game, Soma. But what we're trying to figure out is if we can make a sanity/darkness mechanic without having an abstract background system powering it all. The idea is instead to design the levels in such a way that a state of mind, similar to the sanity system, arises in the mind of the player from mostly narrative aspects. The idea is that when you no longer have a strict and simplistic system driving it all, the player's imagination is free to roam, and the experience can become much more powerful. However, trying to set hours of varied gameplay up in a way that fosters a particular state of mind is a hard task indeed.
We are still working on the idea, and while it is not where we want it yet, we are getting closer. We'll see how it works out!
Are you a developer interested in contributing to Gamasutra's Game Design Deep Dive series? Email editor-in-chief Kris Graft: email@example.com