partnered with

Killing Time

Up To Speed With SUPERHOT

verything to do with SUPERHOT, the shooter where time only moves when you do, seems to occur in moments. It’s the overnight sensation you can sell with just a sentence. An ongoing Kickstarter campaign to turn its prototype into a complete package blasted through its funding goal in a day. The pitch itself is all bullet points, sound-bites and explosive animated GIFs. Bang, bang, bang. Now stop. For all its momentum, SUPERHOT has also been at something of a standstill this whole time, hacking at a single corridor sequence with slice after vertical slice. Its creators are full of amazing ideas, as you’ll discover, but the front-loaded nature of crowd-funded development has meant it’s polishing ahead of schedule, rewinding and refining that single hallway. Needless to say, the game’s minimalist facade is deceptive. No worries, though. The following interview with creative director Piotr Iwanicki and art director Marcin Surma reveals developers just as comfortable with full game development as with the seven-day game jam or the gymnastics of a Kickstarter campaign. With experience spanning experimental Flash puzzlers, large-scale R&D projects, comic books and 3D art, Team SUPERHOT seems to have all the mental agility it needs.

How will ‘moments’ like a car bomb wiping out enemy bystanders fuse with the chess game of SUPERHOT’s combat? With the game’s very timescale linked to your actions, it’s hard to imagine they won’t. Perhaps the bigger question is how emergent the spectacle can be.

How much does the nature of a Kickstarter campaign affect the development of a game like SUPERHOT? Piotr Iwanicki: We’re coming from the game jam, obviously, so the game was a complete something that you could show and talk about. It gained popularity very fast just because we put it on the internet. Kickstarter was more of a deliberate plan to make content for the video, and that’s a big development problem. When you’re making a game spontaneously in a game jam, you’re experimenting, just having fun, doing what works immediately. When you’re going with Kickstarter, it’s difficult to retain all that because you always want to have something finished and demonstrable. Marcin Surma: And it’s only a small fragment. This is very tiring because doing a videogame, especially one with many different levels, should start with the levels. At the moment we’re still in the very first level we did: a corridor – three enemies in the back, you start at the front. It’s the level we did two months ago for the Kickstarter, the level we did a month before the Kickstarter, and it’s the level we’re doing now for another demo. It’s one level and it’s going to drastically change when we have the other levels; when we know how the game’s going to react to light and other situations that aren’t simple corridors. Iwanicki: Kickstarter is a wonderful idea, but I’m not sure if a lot of people realise that it also changes the way you make the game. It’s like a never-ending promotion. Now we’re hoping to concentrate on good old-fashioned hard work, just experimenting with the game. It needs to bloom. DET: Does spontaneity mean you base the game on instinct more than any explicit inspirations? Iwanicki: When we were working on the storyline in the beginning, we had this fear of ripping something off too much. It was too obvious: too much like Eyes Wide Shut; too much like The Invisibles, the comic book by Grant Morrison. Now the story has had some time on the shelf, we think that’s just the way inspiration works. You rip something off and at first it might seem like total abuse, but if you’re mixing things your own way then you find your own voice. I love firstperson shooters. I did a lot of levels for Quake, the first one, even when there was a third one: it was fun and my computer was old. There was this trend in the mapping community of doing maps without textures – geometry competitions. I started building those and found it was something that works. Textures take something away from mapping, a freedom. You have to switch programmes, choose textures, and the geometry is constrained by your texture set. When you resign from that then you gain total freedom. That prototype feel really worked with SUPERHOT. Most of the guys on the team believed differently: that a game can’t work well without specific textures. They’d ask: ‘Okay, okay, but when are going to texture it? It doesn’t look like a real game.’ But I saw those Quake levels and knew that it worked. It’s not literal, which is a great feeling. Most 3D graphics tend to be very realistic, or stylised realistic, which doesn’t leave much for the player’s imagination. In the SUPERHOT prototype you’re dropped into those scenes at the middle without the whole story. There’s not enough of this in games generally. DET: Yet you still have to nudge the imagination in the desired direction. Grasshopper’s games are great at that: a label on a whiskey bottle or a single item of furniture can almost set an entire scene. Iwanicki: That’s the question we often ask when building a scene for SUPERHOT: how can you tell the place through just the minimum amount of assets or elements? What constitutes a bar, for instance? You don’t need a lot for a place to become something in the player’s mind. Surma: What makes a rooftop? It’s just a cube, but when you have other cubes in the distance then it’s a city. Then you can have a high-definition helicopter and that’s it, that’s all you need. Iwanicki: One of the very first inspirations for SUPERHOT was to make it more like theatre. We knew we had limited resources, so that was the best reference. The modern artistic theatre is very strange; once movies were invented it had to go a different, more symbolic way. So, if you have a helicopter in a scene, you don’t bring the helicopter into it because you can’t afford to. You have a shadow, you have a sound, even shooting from the helicopter – but it’s told with subtle cues. It’s like the shark in Jaws. Surma: Though sometimes you do have to show it. There’s one level with the helicopter where it’s not seen, but then this is a game about stopping time. The helicopter has a rotor that can move and then stop when you stop time – how could we not have that? You stand still and you see the blades, the shadow of it crawling through the geometry of the level – awesome stuff. Some sort of modelling is necessary there.
Bullet spin was introduced much earlier than the prototype art style suggested, smooth bullet geometry making it almost impossible to see. Size is the more important factor, though, the bullets now twice as big as they’d be in real life.
How much can you afford to have moving on-screen at once? Can the colour scheme help? Surma: Bullets? Sure. Rotating fans? Of course. Highways? Airliners landing? Everything that moves and looks awesome when stopped mid-move is welcome here. Besides that we have particles, like the glittering crystals enemies shatter into. If I had to remove all particles but one, those would stay because it’s amazingly satisfying watching them slowly rotate, catching reflections and glittering, all while this shattered crystal body falls to the ground. If I had to remove all but two, I’d leave the shattered glass. Debris in SUPERHOT is tricky, though. Since the player is able to see everything in very, very slow motion, we either have to create an absolutely hit-reactive environment that gets destroyed in a nigh-impossibly fluent way (and still not allow the walls to crumble), or handwave it and do it subtly. Handwaving is currently our way to go. The thing I’m looking forward to the most are the dust particles in light volumes – that’s going to rock in slow motion. It’ll probably be the last particle effect for now, just one step before ‘too much’. Colour is a tricky question. We had two ways to go after the prototype. One: making it look like a real space with geometric enemies and lights that cast shadows. On the other hand we could have a limited palette of sharp blacks, whites, reds – something that people could still call ‘Quentin Tarantino’s version of the Mad Men intro’. When I came to the project, it was all monochrome but with red bullet trails and enemies. While it looked quite interesting, it wasn’t pleasing to the eye. The first thing I did was to push the color in, to light all the levels with coloured lights: not red but yellow, blue… The next thing was to push the crystals up to eleven – and that meant removing the blood and making the enemies bleed polished gems. It also meant creating an environment where enemies, props and set-pieces all have distinct, sharp, highly reflective surfaces while the simplified environments are matte white in contrast. They look three inches high, like a model, something you’d see at a flat developer’s studio. In this white space we put bright red crystalline enemies and some nice shiny structures; low-poly but with the facet visible, all the lines shining and glittering. It clashes amazingly well with this environment. The contrast is something we’re pushing to the limits: crystalline bullets, set-pieces… Imagine the helicopter as a sculpture – it’s awesome. DET: How much iteration has already gone into designing those bullets, and how much is to come? Surma: One part of my crystallising frenzy was removing smoothing from the original bullet models and recalculating the mesh by force. While the result was far from perfect, it did its job: it both made the bullet rotation noticeable, which is very appealing when time slows down, and made it glitter during that rotation. At the moment I’m testing dedicated crystalline meshes, both regular and irregular. It’s quite fun, to be honest. Iwanicki: We experimented with the bumpiness of a bullet, adding a texture and so on, but the crystal look solved all those problems at once. The tricky part of a bullet in SUPERHOT is visibility. The most important thing is the red trail that explains the position and direction of a bullet at a glance. That’s the part where some iterations are still ahead of us – but in the basic form it was there even before making the first version of the game.

Sharing the in-late-out-early philosophy of a (good) movie, SUPERHOT likes to drop you into scenes such as this as they’re unfolding, leaving you guessing. Movement of realtime light and shadow is key to recent concept art.

The animation of the katana slicing the bullet works because of the FOV change and camera jiggle that goes with it. How do you discover and finesse those little touches? Iwanicki: Those little camera shakes, with the little glow on the moment the bullet’s been cut: on the one hand it’s like montage, just masking the moment where we’re switching the models of a bullet into two slices; but on the other it’s just this really intense moment, like in the movies. The craft of making games is often about making those little interactions work. A little flash when you pick up an item, a proper sound. It’s something you do a lot so it’s important to have this tactile, kinaesthetic feeling. It has to be pleasant. The GIF doesn’t sell the sound of cutting a bullet, though, which is the other half of the equation. But we’re thinking about those things and it’s good that people are noticing. DET: One of the great things about being a small team is that you can see through the whole process of hitting a problem, then discovering something that goes beyond merely solving the problem – something artistic. Iwanicki: Absolutely. In the prototype there was this idea of enemies just falling apart in these disappearing stripes. It was quite a fun, memorable moment, and it worked really great. Surma: Especially when you time-shifted. Iwanicki: But it’s a bit counterintuitive; in the beginning we though those baddies should stay. You’d come to those empty, clean, white places, and when you come out it’s full of red bodies. It’s an idea, yes? But this idea clashed with reality because, with the timescale of the game always changing, the physics totally explodes. You want this baddie to be left behind but it doesn’t want to stay in one place, it just flies all around the room. That could have been solved by putting a week or a month into working around those physics, but I’m sure that if we did that, there’d still be those tiny glitches that kill the effect. So, the solution is actually to not solve the problem, it’s to go around it and make that into a style. This is like the core of developing SUPERHOT: making your weaknesses into your strengths. It’s the same thing we had with finding an art style. We were iterating on something a bit more realistic, a textured environment that looks a bit more like a real place, but it was always obvious it was done by a small team. That we’re not Ubisoft. Surma: We were trying too hard and it still wasn’t enough. Iwanicki: And it never would be. So, the solution wasn’t to hire more people or just partner up with a bigger studio, it was to not tackle the problem and take it in another direction. DET: Great use of type is common to both SUPERHOT and one of your previous games, Rektagon. It’s something Flash games do a lot and 3D games really don’t. Iwanicki: It’s really fun and easy to do animated type in Flash. It’s just this way of letting your tools shape what you are doing. In Flash it was easy but in Unity, the engine we’re using for SUPERHOT, it’s not. But Rektagon proved that it could work and is an awesome effect, and I wanted to use it in a bigger game that would reach more people. The idea of full-screen type, it just looks— The Blurred Lines video [by Robin Thicke] with the big fullscreen hashtags looks awesome, like the screen is talking to you directly. It’s like the computer becomes this mind-controlling machine – like the ‘OBEY’ signs from They Live. I can’t imagine why it hasn’t been done much before. Device 6 has this awesome title sequence, but it’s surprising that it still is surprising. You’d think it should be commonplace for a game to have that. Surma: But then Device 6 borrows a lot from the ‘60s, and at the moment in cinema there’s not a lot of title sequences. There’s James Bond, The Girl With The Dragon Tattoo, and that’s about it. The focus changed. You’re straight into the action now. Iwanicki: And when it comes to games, it’s always this nightmare of you making a title sequence and people just skipping it. Surma: Tomb Raider did a cool title sequence: fifteen minutes of gameplay and then Lara Croft just standing on a cliff on the right, a shipwreck on the left, and simply the title. It was fresh.
The cars have evolved – or indeed devolved – since this concept stage, made even more primitive to fit the game’s new ‘art installation’ look.
If the game’s impression of a tunnel and street look abstract, wait till you see its trucks. Actually, don’t, it’s the thing at the bottom.
How much is SUPERHOT knowingly shaped by constraints? Iwanicki: We decided at one point to use a very specific tool, ProBuilder, for creating levels. I like it because it’s like mapping for Quake. Surma: Or building with Lego blocks. Iwanicki: It’s very constrained geometry. You could argue that we need a better tool, or that we should import models from 3D software and things like that. But if you decide that this is the tool you’re using, you come to a moment where you allow the tool to shape the work you’re doing. With ProBuilder you can try to make arches and curved surfaces, but it’ll always look a bit silly. You could try and do it but why? It just doesn’t work well. So you embrace the constraints of the tool you have chosen and, well, it worked for us. Surma: And it made us very happy to use 3D modelling for the set-pieces. As I said, it’s a brilliant contrast. You have this blocky architecture and considerably less blocky props. It makes it look like an art exhibition where the space is just there to block it from the outside – to not be visible but complement the art, the things you came to see. We’re building levels with blocks so we can complement helicopters, enemies, and the exploding cars that, right now in development, are these blocks of crystal with no windows, no tyres, and no moving parts. If someone had to sculpt a car from crystal and set it in a gallery, this is what they’d go for. DET: Have you had to invent a formal process for finding things that look good when slowed down? Surma: We didn’t really sit down to write anything. Sometimes I’d just have an idea in the middle of the night and send a text message to Piotr: ‘THIS IS GOING TO BE NEXT.’ What was the last thing? Ah, yes. A shootout. Three enemies, two cars – limousines – under a bridge. Nothing more. It might be a drug deal gone wrong, or an execution, but that’s for the text to describe. Iwanicki: So, there’s those two limousines. You start near the trunk, the trunk is open, and you have text explaining you just escaped from the trunk. You build the story. Most of those little ideas come to us when we’re trying to sleep. It’s wonderful. Surma: One particular text message went: ‘PIOTR, DON’T YOU DARE STOP ME FROM DOING A HELICOPTER WITH ITS BLADES ROTATING.’ He was pushing his idea of doing it without seeing it, but we had to have it. But that’s the great thing about SUPERHOT: we can have one level where you do see the helicopter and one where you don’t. Those two levels don’t have to be connected in either story or assets. We can have one level where the chairs are simply cubes, and we know they’re chairs because of how they’re placed in the environment; then we have another level where the chair is modelled as part of a set-piece. We have cars that explode, cars in a garage, cars that are further away which are simply black boxes. We don’t even need to have consistent assets or levels of detail. Whatever we want to focus on in the level gets more detail. DET: Have you considered things like having light move in slow motion? Iwanicki: We already did it, and it’s totally awesome. Surma: A subway level. In the dark. The player is in a subway car, and the only light coming inside is the tunnel lights that breeze through the level. They’re very slow when you’re still, but when you move they come to life, shining through the car and the enemies which cast those amazing shadows. It’s something you could not do in Quake 3, I believe. Iwanicki: More Doom 3. Surma: Well, not quite, because we’re having no normal maps. This is the thing I decided early on when I came here: if we have polygons, the specular must make every facet shine and glisten. It’s memorable and looks great in slow motion, especially when you have moving lights. Iwanicki: We hope to use them more because the whole level comes alive when you move. Enemies’ pistols produce those intense muzzle flashes that light up the whole place and make all the tiny particles glitter. It’s great feedback, and that’s important during the early levels where you have to explain how the mechanisms work. That’s when all the details come into play. We don’t have a meter showing how time is moving, it’s all shown through the environment.
What other effects are you adding, and what rules apply to them? Iwanicki: I’m thinking in terms of places, emotions and story-like moments. Effects are just effects in the long-run. It’s cool to see rain in slow motion, with individual droplets merging as time moves faster, then dividing again as you slow down; but it’s all about context. For instance, an explosion is a fantastic effect to watch in slow motion, but starting a level with an exploding car bomb, the car slowing flying towards you and crushing your enemies – that’s a moment. Those are more important to us, and it’s all about staging. One effect that would be great and allow for some mind-bending moments is introducing elements that move independently of the gameplay time scale. One idea I toyed with was pigeon-glitches flying away from you. It could add a lot of flavour but it should stay flavour, not a core gameplay mechanic. DET: You mention hiring a second animator to help solve some of the game’s problems. Examples? Iwanicki: It’s hard to talk about animation without getting very technical. In SUPERHOT you can see the details of enemy behaviour and watch them in slow motion. What we wanted to avoid were sudden jumps in the animation. When the enemy is aiming at you, you can see where his gun is pointing and move just a bit in the opposite direction, which is awesome in close combat. But this fluid approach hinders the behaviour of an artificial intelligence. Enemies can only choose from the behaviours that have animations, and the AI system needs to wait for the proper moment to start blending them. This may go unnoticed in games that use game time as-is, but in SUPERHOT these delays become an important part of the gameplay – which is a pain to tweak and iterate on. We’re now getting ready for a little revolution in our animation system, according to the same principle as we did with the general art: simplify it. Embrace the glitches as part of a style. Make the iterations easier and faster, deliver more interesting enemy behaviours without necessarily showing everything literally. I can’t wait to start working on it. SUPERHOT is currently being funded via Kickstarter and will be released during 2015. The September 2013 prototype can be played here and requires the Unity web plugin. [ssba]
Dead End Thrills Features are produced in partnership with international art dealership Cook & Becker.