Peering through the glass window in a studio in East London, you might mistake the over-excited yelling and arm-flailing that was occurring inside for a million-pound lottery jackpot win.
The overexcited people inside said meeting room was us; a tiny AR indie game studio called Combo, and we’d just released our debut game, Pigeon Panic.
Pigeon Panic was a very silly AR pigeon-scaring simulator for iOS that had seen relative success on the App Store. Launched just in time for the release of iOS 11, we’d received prime placement on Apple’s editorial pages, a ton of prolific press, a healthy download count; and, of course, some questionable reviews...
Alas, the excitement in the meeting room wasn’t actually about any of that. We were brainstorming a concept for a game to follow up with, and one of the team members had just brought some information to light; it was early October.
We were only a few weeks away from Halloween.
Perhaps that might be a fun theme to explore?
Almost immediately, “Let’s Play” YouTube videos of P.T. were thrown onto the meeting room television. The room buzzed with various people sharing their horror game experiences - from old-school flash game screamers, all the way to “that one bit in Resident Evil with the dogs”; it was certainly no mystery that we’d all had a good scare from a horror experience in one way or another.
We’d found our idea. An AR mystery thriller - set in the player’s own home.
Six weeks, four team-members and one missed deadline later, dARk - a paranormal AR thriller - hit the metaphorical App Store shelves.
As this was our team’s first shot at writing, designing and building a narrative-driven experience, we encountered a ton of challenges along the way.
In this post, I’ll be writing about some of the AR-specific challenges we faced in the hope that some of this could be taken away to improve other, future narrative experiences and push the medium forward.
Let’s start with gameplay. The key mechanic of the dARk experience is pretty straightforward; you hunt for clues placed around your real environment to uncover the mystery of what happened to your ex-colleague.
Much like VR, the perfect playing space to play through dARk is a large, empty room, free of obstacles (like, say, home furnishings). And therein lies the problem; almost nobody has a space like that.
It’s worth mentioning we’re based in London, where space is even more of a commodity. As a result, there's a whole bunch of friction attached to forcing players to find a suitably-sized room before they even fire the game up.
So the question; how do we make dARk playable in most people’s small, obstacle-ridden homes?
Generate the experience around it.
After entering the portal and arriving in their parallel universe, the game asks the player to move around the room, with the motivation being to search for clues.
The game keeps track of how far the player's moved, and as soon as they’ve created a path around 10-15 meters in length, we have enough space to begin spawning clues and the experience progresses.
This takes about 30 seconds to complete, so to keep players on their toes (and to alleviate the feeling of room-scanning), we created a few simple "what was that?!" moments whilst the player searches.
Through testing, we also encountered a big bonus that came with building this mechanic; since clues were spawning on a path that the player had walked on previously, it drastically minimised the chances of spawning clues, integral to advancing the story, inside walls and player's furniture.
For dARk, we designed the experience so that players would have to physically hunt for paper notes on the ground.
And so, just like hunting for pieces of A4 paper in real life, we made them tough to find. The notes were visible only when the player's torch passed over them.
The problem was that it was just too tough. People lost interest, thought that the game had bugged (which, to be fair, it had sometimes), or they just straight up couldn't be bothered to walk around the room any more, and asked us what they had to do next - which took them out of the experience.
To ask people to physically move around the room, phone in front of their face, is one thing. To ask them to look for something which is almost invisible to the human eye felt simply unfair.
We wanted to maintain a balance; let the player feel like they’d found something on their own accord, but also not make it so hard that they close the game and leave the experience unfinished.
Unfortunately for us, this was a tricky balance to find.
Quick fix? We added dimmed spotlights to each of the clues, then tested again. These were still difficult to see on first glance.
As soon as we did this, we noticed that players got stuck far less. People were lead to the clue's spot by the light, then found the note by their own accord in that spot. As a result, the percentage of people who finished the experience increased significantly.
Our biggest takeaway from this; don’t let the player ever feel lost or confused. Always make sure they feel like they’re progressing forward, or at least moving in the right direction with some reassurance. As well as adding the artificial lights to the notes, we also maintained a heavy amount of in-game dialogue to keep the player reassured that they were still on the right track.
Despite wanting to immerse the player in a realistic experience, it’s worth remembering that they’re one button away from going back to the home screen, and potentially never re-opening the game again. The possibility of making things too difficult in the name of realism is simply not worth the risk.
As the game progressed and more complex clues appeared, we noticed people snorting in amusement at the game's clues very obviously spawning directly in front of them. We decided that amusement ranks pretty low in the desired feelings we wanted someone to experience during our "horror" game, so something needed to be done about it.
Thankfully, with the game being set in a paranormal universe - anything can happen, right?
So, as a solution, as soon as a new clue needed to spawn in the world, we created a system which decided on a spawn point for the clue on the path created earlier - but only spawned it when they weren't looking at it. Simple.
The result was stellar. Testers would complete a section of the story, turn around, disoriented - and then, right where they were just looking, a single note would appear on the ground. This would often make the player jump in fright, and prompted the question “was that always there?!”
This added to the player's feeling that they were being "played", which in turn added even further to the chilling atmosphere we were trying to create.
In testing both Pigeon Panic and dARk, it was pretty obvious from the incessant “20% battery remaining” alerts that AR is a bit of a battery-sink. It's resource-heavy with just the camera feed and AR framework running, let alone the performance ramifications of a game/experience on top of that.
Keeping the experience short in aid of this worked out pretty well for us.
In testing, we found that holding a phone outstretched for 8-10 minutes wasn’t too bad on the ol’ arm muscles - but anything more was a little uncomfortable. With our ambition being to get as many people to the end of the story as possible, 10 minutes proved to be a great experience length.
After launching dARk, we returned to the aforementioned meeting room once again; this time, going through the 200+ reviews for dARk we’ve received so far.
And I'm delighted to report that the reaction was as ecstatic as the first visit.
Some of the reviews were tech bugs, others complaining that the game was too short; but most of them were asking for a second chapter.
Closure to the story.
An answer to the cliffhanger that we accidentally created at the end of the game.
To us, these reviews are a really encouraging sign that states that storytelling in mobile AR could be a strong use case for the medium, and as the technology continues to develop, we can tell bigger, longer, richer stories and create hard-hitting, emotional experiences.
AR still feels like the wild west to us - so any thoughts on how we could improve the experience is massively appreciated. Fire over any thoughts (seriously, anything at all) to email@example.com.