How Halo and classic anime inspired a Roguelike shooter with the brightest of enemies
Like any ambitious space mission, Galak-Z’s launch was fraught with peril and the potential for disaster. The game represented 17-Bit’s difficult second album following the success of Skulls Of The Shogun, and one that packed remarkable and potentially problematic enemy AI behind its atmospheric 2D visuals. It didn’t go smoothly. A litany of bugs and technical glitches blighted Galak-Z’s release, forcing 17-Bit’s small development team into three weeks of damage limitation as they tried to get their new vessel back on course.
“There was a bunch of stuff coming in that we’d never seen before and we were freaking out, trying to patch it immediately,” 17-Bit founder Jake Kazdal tells us. “So it was kind of a bumpy start. But we fixed it, and the reviews were fantastic. We made some really daring design choices, so it was gratifying to see that, for the most part, I think people really liked it.”
Those daring design choices had inauspicious, but fortuitous beginnings. Kazdal had finished his work on Skulls and was now focused on marketing. He’d previously been invited to a game design workshop following a talk he delivered at the DigiPen Institute Of Technology in Redmond, Washington, where five of the attendees had expressed interest in a summer internship. “I was like, ‘Sure, we’re close by – why don’t you guys come down and we’ll start noodling on the next project’s concept?” Kazdal says. “I’d been thinking about it for quite a while, and I gave them tons of notes, sketches and stuff like that. I just laid out the whole project for them and they started banging away on it, and we had a prototype very quickly.”
A second chance encounter saw Ed Fries, former Microsoft vice president of game publishing, introduce Kazdal to Cyntient, an artificial intelligence specialist. A startup, it was looking for a partner project as it finished working on its middleware while deciding on the direction the company would move in next.
“They were pretty focused on more social stuff, and strategy games,” Kazdal explains. “I said, ‘Well, this is going to be an action game, but I really love cutting-edge AI.’ My favourite game of all time is probably the original Halo on Legendary, where the combat is just so fluid, organic and different. Even if you’re playing the same level over and over, those bigger battles are original every time. So we talked a lot about that stuff and I was like, ‘That’s what I want to do – I want to do Halo in 2D. I want that same level of control, that same level of fidelity and great, organic, evolving combat.’”
With Cyntient happily onboard, Kazdal worked with one of Skulls’ animators, industry veteran and former PC Gamer journalist Colin Williamson, to design the look of the game. The aesthetic they settled on was heavily inspired by ’70s and ’80s anime, the final component – along with deep physics and AI- based gameplay – in what was rapidly shaping up to be Kazdal’s ideal adventure game.
“I’d been doing sci-fi stuff for years as a concept artist, and I wanted to do something that would stand out and have its own look,” Kazdal says. “I hate rendered explosions and going for realism in games. I remember watching the old Macrosses, and the really sexy explosions in the old animes. I started looking at some of that stuff, as a first reference, thinking about how we could stylise our explosions. But then I realised that [style] has sort of been forgotten. I thought it would be fun to doodle on all these things, sketch stuff and build this campy world perfect for a space shooting game.
“I didn’t really mean for it to be that era per se, but the more I explored it, the more it made sense. Using that era meant we could have all the stereotypical bad guys and all the explosions and mechs, and it just sort of fell into place from there. There’s a whole generation of players who grew up watching this stuff. The first time we showed it at PAX, people were walking by and before they’d even played the game were like, ‘Oh my god, this is amazing, when can I buy it?’ just because they were into that aesthetic. I thought, ‘OK, this is cool, this is working.’”
The PAX demo was a prototype whipped up by Kazdal’s small team of interns. They worked on the game all through the summer, two of them eventually taking full-time roles. Kazdal also made a handful of additional hires as Galak-Z’s prototype took shape. The build consisted of one giant map populated by basic enemy ships, and Kazdal hand-placed every asteroid and rock formation, spending time tweaking the twisting path through a labyrinthian cave network.
“We were a tiny team so we couldn’t do everything at once,” Kazdal says. “We were focusing first on the feeling of the game – the overall aesthetic and what the combat and controls would feel like, and how the basic flow of the game would work. That all came online really early, so we kept fine-tuning that and we kept tuning the level – you know, we would try different enemy combinations, and because they would be in different spots and you might run into them in different places, it really changed the feeling of the encounter. It might be a really tight, confined space, it might be two squads that you’re fighting against in the same space, or it’s just one squad and it’s a wide-open area. Every playthrough felt completely different, even though it was the same level every time.”
Although the game was coming together quickly, and showing obvious promise, 17-Bit was under additional pressure because it had to operate across an ocean. While Kazdal founded 17-Bit in Seattle, the draw of Japan – where earlier in his career he had worked at Sega, assisting Tetsuya Mizuguchi on Rez and the Space Channel 5 games – proved too great.
“I just really wanted to go back,” Kazdal tells us. “I thought, ‘What the hell? My wife is Japanese; we could just move there and have our company, but we could go to lunch in Kyoto every day. Wouldn’t that be awesome?’ So we did it. Kyoto is an unbelievable city to live in.”
Not that Kazdal is a fan of distributed development. “I do prefer having everyone in the same room, in the same time zone. It makes stuff so much easier, and being the art director and creative director it makes a lot of sense for me to be able to check in regularly as opposed to once a day. It was difficult, but we made it work.”
With Galak-Z’s basics in place, and transcontinental development headaches overcome, the team started to focus on building something more substantial. Kazdal had a highlevel story in mind, and plenty of ideas for levels, but continuing to hand-design and place bespoke components wasn’t sustainable. Kazdal was also aware of the extent to which the game was likely to evolve, meaning hand-designed levels created early could be rejected – and the work that went into them wasted – down the line. This need to streamline was a catalyst for the burgeoning space shooter to take a new direction.
“Going in and moving literally every single rock in every single one of these dungeons was just insanity; it was never going to work,” Kazdal says. “So I talked to the lead engineer, Zach [Aikman], about creating a tool for me so I could hand-draw the rooms, then put them together so we could reuse them and put them in different spots. So we built this editor and I created a handful of rooms. By that time I was really conscious of the fact the combat was the core of the game, and that exploration and not knowing where you were going was a key component of the whole experience. So I was like, ‘Can we just try shuffling these things up?’
“I was playing a lot of Spelunky and Rogue Legacy at the time and I was really getting into that format, just jumping in and having fun and not having to memorise stages. And I’ll never forget the night that, having played through our dungeon a thousand times, all of a sudden we had this shuffling thing that would give you a new experience every time. I mean, I was flattened by how awesome it was.”
Kazdal hesitantly told the rest of the team he thought the best structure for their formless prototype was that of a Roguelike. Expecting resistance, he was surprised to find they all felt the same. Now, as a game about procedurally generated encounters and exploration, Galak-Z had real direction. But while Kazdal had found plenty to appreciate in the work of Derek Yu and Cellar Door Games, he still had misgivings about the use of randomised levels over hand-tuned creations. Concerned, and wrestling with technology and algorithms that were new to them, the team devoured as many articles and videos on the making of Rogue Legacy, Spelunky and other procedurally generated contemporaries as they could. It quickly became apparent that the numbers could be controlled, refined and directed with purpose; chance didn’t have to factor into it.
“We had that one hand-designed dungeon that we played over and over, and had tuned the hell out of to be sort of a perfect stereotypical Galak-Z level. So we thought, ‘OK, let’s reverseengineer this thing.’ It took a while, but we were spending a lot of time on how to recreate that algorithmically so it would keep a lot of these key pacing decisions we’d come across naturally.
“Sometimes certain enemies are way too hard, so you go in and tune that algorithm. The same for when the levels are too short or too long. And so we started to go down the line and nip all the problems in the bud, keeping an artificially higher level of control behind the scenes that players don’t necessarily need to know about. So now I’m the biggest fan in the world of this stuff, because even as the designers of the game, we get to play something new every time.”
That’s all very well for human explorers, but it presented a unique challenge for Cyntient’s AI. After a brainstorming session in which 17-Bit talked with Cyntient about what it would mean to have enemy craft that could think for themselves, adhere to orders, retreat when threatened and even make predictions on what the player might do next, a basic system was plugged in. It proved immediately amusing and made for combatants that were capable of surprising – exactly what Kazdal had hoped for when he described Halo’s encounters. It proved so effective, in fact, that the team quickly realised they would have to dial down the ferocity of the opposition’s intelligence. Enemy pilots would dodge incoming shots and place their own with unnerving tactical nous.
“We were like, ‘All right, we’ve got to take these guys seriously,’” Kazdal explains. “They’re not necessarily faster than you, they’re just really good at their jobs. They’re these little terminators trying to kill you, and they’re creepily intelligent. I don’t want that AI going into drones or anything anytime soon because it’s terrifying.”
Getting Galak-Z into orbit may have been a fraught experience, but Kazdal is thoroughly satisfied with the results. 17-Bit created a conglomerate of classic genres that introduced new ideas to each component part, while at the same time offering up something that feels refreshingly unfamiliar to modern players.
“Some of the systems and tech that are going into these classic genres is so exciting to me, and that’s what I want to keep focusing on,” Kazdal says. “The next couple of games I’m thinking about are the same sort of idea – combining classic genres with modern tech and seeing how that shakes up the whole experience. And I think it really does. I mean, Galak-Z is a shooting game that feels unlike any other arcade cabinet, just because the AI is making such a difference. In that sense, each encounter is much more of a combat simulation than it is a videogame.”