Making VR games reminds Infinity Ward expats of PS1 game dev glory days

“If you’re going to start developing a VR game, you really have to think about performance from the start. It’s like we’re back in the PS1 era.”

So says James Chung, a former Call of Duty developer who’s spent a good bit of time trying to make a first-person multiplayer VR shooter that doesn’t make people sick.

Together with fellow Infinity Ward expat Taehoon Ho, he co-founded the VR-focused Reload Studios in L.A. last year with the goal of developing a multiplayer VR game in time to capitalize on the commercial debut of VR platforms. At the VRLA Summer Expo in Los Angeles last weekend, Chung laid out his perspective on the state of VR game design.

“You have to think about optimization from the start when you’re making VR games,” he told me, after demoing Reload’s upcoming VR shooter World War Toons. “Right now, I think the trend in game dev is that people just cram all this stuff into their games. They don’t make games like they used to, in the PS1 and PS2 eras. Back then you had big limits, so developers had to be extremely creative to make their games look good.”

His opinions echo those of many VR game developers I’ve spoken to, including Darknet developer E McNeill, who blogged a set of guidelines for GearVR game makers last year that advised developers to “start worrying about performance before you even begin coding.”

But Chung also had some advice for making VR games that I hadn’t heard before, bits and pieces of design wisdom the folks at Reload (now grown to a headcount of roughly 25, with at least 2 contracted artists) have figured out for themselves in the past year. What follows is an edited transcript of our conversation on the subject which touches on Reload’s VR design learnings, its attempts to make a multiplayer FPS work in VR, and Chung’s feelings on leaving Infinity Ward.

Reload has a lot of former Infinity Ward folks on staff with, presumably, a lot of experience making hyper-realistic military games. So why go for such a cartoony style with World War Toons? 

The main reason we wanted to go cartoony, is that the moment the users understand that in this world, there's no such thing as real-life physics, you can bend all the rules. The moment that people realize that nothing is real in this world, their brains become much more accepting of the simulation.

Ah, yeah. Valve's Yasser Malaika talked a bit about that at GDC Europe this month, about how the abstract trumps the real right now in VR game design.

Right, yeah! So for example, when we do the long jump in World War Toons, we slow down the jump a bit. That would look weird in a realistic game, right? But in cartoony VR it works really well, it's not an issue.

There are a lot of different issues in VR development, though. You can't have the usual HUD elements, for example, because they often make people sick; so we just got rid of it all. 

So we had to figure out how to replicate functionality, for things like the minimap -- that's really important in a first-person shooter. So if you get rid of that, how do you show people what's happening on the battlefield, at a glance?

That's why we have an angel animation; when an enemy dies, an angel appears and floats up to the sky. That way, if you see a bunch of angels floating through the sky out of the corner of your eye, then you know immediately that that's where the action is.

Even the control solution that we came up with is different. You can't rely on a traditional first-person control scheme in VR, because if the player starts moving the camera around with the right thumbstick, they'll immediately start to get sick.

So to some extent, your decision to make a VR game dictates the kind of game you can make.

Yeah, ultimately, our decision to develop a VR game had a lot to do with our decision to go with a more cartoony look. It also had a lot to do with our control scheme.

It affected our optimization process, too; performance-wise, we optimized the game from the start to be as smooth as possible. You can't do a lot of the shader tricks, a lot of the advanced shaders that devs on traditional games use to make their games look real. You can't do that on VR right now, so that's why we went with more stylized, hand-painted cartoons.

But also, a lot of it has to do with having the freedom to do the kinds of things you can't do because you're making Call Of Duty. Even something as simple as a death animation -- you can't do a lot of very expressive stuff if you're trying to be realistic. So that kind of stuff is not only fun to look at; it's just fun to do. It has a VR component, but it was also just fun to do. 

What bits weren't so fun? What ideas looked good on paper but didn't work out in VR?

VR development is interesting because we wind up sitting around and theorizing about a lot of stuff, and then when we try it...a lot of stuff that we thought would work, doesn't. There's also a ton of stuff that we assume won't work, that does actually work.

So for example, in our game, when you pick up a weapon you actually become the class [that carries that weapon.] And because they're cartoony, there's definintely a height difference between the different classes -- like 12 or 18 inches. On a flat screen, you really don't notice the shift in heights. But in VR, you really feel the shift in perspective. Stuff like that always surprises us. 

Did shifting perspectives like that, without warning, make people sick?

No, surprisingly, that stuff was okay.

What we found interesting, what we learned, is that if you have a lot of repeating textures, that makes people sick. So like, if you go up lots of stairs and stuff, if you go over the same texture for too long, that makes people sick. So we had to break up a lot of repeating elements in the game.

Side-to-side camera movement -- strafing -- is also one definite thing that makes most people sick really easily. That's why we took out strafing for this [VRLA demo build], but we're working on a more advanced control scheme for advanced players.

Yeah, I noticed there's not a lot of open space in the demo -- the map is cluttered with environmental objects, each with different textures.

Yeah, and that was actually really challenging because you have to make VR games extremely lightweight. Really lean. So like, our game is so lean right now that you can actually run it on mobile. 

You mean you have to keep the framerate high and consistent, or else people might puke.

Yeah. Imagine you're making Call of Duty -- or any high-end triple-A game right now -- if you can get it running in 1080p, full HD at 60 frames per second, that means you've done a fantastic job, right?

GearVR is already 1440 -- it's quad HD. And you have to render one instance for one eye, and another for the other eye. That's how you get that stereoscopic depth. And with a fast-moving game like ours, you have to hit a high framerate. Like on [Oculus'] DK2, right now, we're maxing out at 75 frames per second, and our goal is to hit 90 frames or higher. Performance has to be really, really good in order for players to not get sick.

On top of that, not a lot of people have $3,000 PCs to drive these games. So going with a cartoony style was important from that perspective, too; we wanted to make sure that our game runs well on medium-spec PCs. So all of those concerns are mingled into our design process.

If you could go back and give yourself one piece of advice about designing VR games before you started, what might it be?

It's all about performance, and that PS1-era development culture of optimizing games has been lost. 

Even mobile developers, we're sort of spoiled at this point. People can just make things beautiful without really thinking about performance or optimization. But if you're going to start developing a VR game, you really have to think about performance from the start. 

So like, if this is your goal [holds hands about three feet apart] of how you want your game to look, you really have to make the game to perform really well at this level [holds hands about two feet apart] and just start carefully adding in things until it runs well and looks the way you want it to.

Whereas most developers now are used to cramming in allll this stuff [holds hands about four feet apart] and trying to trim it down until it runs okay. You can't make VR games like that. You'll make people sick.