EDITOR'S NOTE: Guest author Brian Crecente founded video gaming site Kotaku and co-founded Polygon. He was also the video games editor for Rolling Stone and for Variety. He currently consults for publishers and the video game industry.
In this 'Next-Gen Gaming' article series, Brian interviews a variety of personalities from around the gaming industry to get an objective look at key topics related to the next generation of gaming spanning genres, platforms, team sizes, business models, and more.
Perched on technology’s razor edge—balanced between generational shifts in game consoles—makes discussing the current and future states of game development all the more interesting.
As we eagerly anticipate a new generation of consoles to come, we’re taking some time to reflect on the lessons learned from this most recent hardware generation. Creators and leaders from nearly a dozen studios shared a wide range of opinions about the biggest and best innovations to hit this generation, the technological limitations that sometimes held them back, sometimes inspired them to excel, and where they see game development headed in the near future.
Image courtesy of House House and Panic
“This argument may be valid at any point in time, but it's better now than it has ever been. Other than the usual suspects such as increasingly powerful hardware, this generation has seen the advent of the third-party engines such as Unreal and Unity,” Jure Ratković, engine and graphics programmer at 2X2 Games, which developed Unity of Command II. “Nowadays indie teams that build smaller games don't need to have a very high level of technical expertise that was historically a prerequisite for game development. This has certainly enabled more people to get into the industry.”
Nico Disseldorp, who helped create Untitled Goose Game at developer House House, calls the ubiquity of game-creation tools and their increasing ease of use the single biggest innovation of the generation.
“At the start of this console generation, no one at House House was a game-maker,” he said. “We were able to teach ourselves the tools we needed to know as we went, only because everything is so much easier to use and more approachable now.”
Roger Mendoza, one of the founders of Nomada, notes the last few years have been very generous to creators and Mike Bithell, whose Bithell Games released John Wick Hex most recently, calls this era a period of technical plenty.
Image courtesy of Bithell Games and Good Shepherd Entertainment
“There's a lot of power which AAA is obviously utilizing at the high end, but that allows us to make ambitious games with enough headroom for the inevitable inefficiencies of our work,” he said. “Put bluntly: We can afford to make games with the performance players expect, without having to optimize everything to within an inch of its life, which makes development more accessible.
“We've certainly benefited from this by being able to jump genres and scopes, without worrying about over-investing in any one set of tools or sunk costs. We've been able to diversify the games we make, which is hopefully exciting to our audience, and definitely exciting and commercially successful for us.”
Greg Street, Riot Games’ vice president of intellectual properties, said the generational leaps in development tools also change the way he talks to up-and-coming future developers.
“I try to spend a lot of time talking with the young people who want to be part of the industry,” he said. “Several years ago, I would have to say, ‘Pick up the StarCraft editor and see what you can do.’ Now I can steer them toward Unreal or Unity and challenge them to actually make a game, even if it’s a small one, with friends.”
Epic’s David Stelzer, who heads up the company’s Unreal Engine business development, also sees the growth of easy-to-use game engines and how it has lowered the entry point for game development as one of the major impacts of the generation.
“It’s come to a place now where I think game development is probably easier and more cost-effective then it’s ever been,” he said. “Off-the-shelf game engines, university and other vocational programs, YouTube videos et cetera has made it so that just about anyone can open their browser, pick up a book, do an online course and learn the very basics to start making games in some way, shape, or form.”
Image courtesy of Epic Games
Games as a service
Perhaps just as important, he notes, is what he calls a major bump in graphical fidelity that arrived this generation as well as a number of other major trends.
“I think you’ve seen in this generation, the real rise of games as a service and the ability for more companies to create service-based games without the need to have a massive publishing organization behind them.”
And with this sort of new games as a service model came the ability for developers to have a closer relationship with their customers and inevitably the ability to iterate faster and better on the sorts of things customers want in the games they play. Stelzer points to Fortnite as a good example of that, but notes it’s just one of many examples out there.
Riot’s Street also sees the increasing speed of innovation as a massive improvement in game making.“Today, you can start generating models and animation almost immediately, even if all you’re doing is proving out a concept,” he said. “This allows you to get to the iteration phase, which is really critical for game design, much faster.”
Aaron Cammarata, a software engineer and game designer at Stadia’s gameplay research and development group, has been making games professionally for 25 years, starting with the Timex Sinclair 1000. From his perspective, that increased development speed is the one thing that impacts game quality more than anything else. “The faster you build things, try them out and fix issues, the better your game. Period,” he said.
And, he noted, when he started developing games in the mid-90s, everything was bespoke. Studios, he said, had to not just make the games, but often also create the tools used to craft the games.“If you fast forward to today, there are a number of off-the-shelf solutions - not just engines, but whole ecosystems, with marketplaces, developer communities, plug-ins, and modules ... it means a creator can get to seeing their ideas on screen much quicker, which is exactly why we have such a high number of amazing games hitting the market each year,” he said. “Things like real-time editing mean the loop from idea or bug fix to implementation can be minutes or even seconds. What this all means is the days of the ‘daily build’ are numbered. So in terms of making the games that creators envision, this speed offers the freedom to explore and experiment, to fail, to go back, and to discover.”
Image courtesy of Gearbox Software and 2K Games
Gearbox CEO Randy Pitchford pointed to a confluence of powerful computers and inexpensive, easy-to-use tools that helped shape the current generation of developers and games. Not only does this mean that developers have more ready-made tools at their disposal, it also means they have the ability to pour more of their game budgets into what appears on the screen. “This is an exciting time for game developers and players,” he said.
This generation’s biggest improvements in the eyes of Feargus Urquhart, CEO of Obsidian Entertainment, includes things like cross-platform playing and saves, which have helped tear down the barriers between platforms and players. But he said that the more substantive changes are the ones that helped developers like Obsidian create larger, more dynamic worlds in which people can explore and play.
“I think about specific technologies and how it has changed how we approach our games,” he said. “So much of what we’re focused on is how do we create these words for people to go into? So for every new generation it’s about what is possible. Like how much can we do with our cinematics and how big of worlds can we make and how dense can those worlds be?”
Improvements in things like lighting, he noted, can have a substantial impact.“So much of what makes you feel like you’re in a world is how well we do the lightning. I think that a lot of the time people don’t recognize just how much energy is put into lighting of games so that they feel like you’re there and they have depth.”
Image courtesy of Obsidian Entertainment and Private Division
Urqhart points to the differences between Fallout New Vegas — which the studio released in 2010 — and Outer Worlds — which was released just last year — as an example of the impact improvements in a variety of technologies can have on a game.
Advances in the physics of a created world, the way a player can “touch the world,” is helped along by existing toolsets, freeing up game developers to spend more time on breathing life into their creations and less on building out the tools needed to do so.“Going back years and years, so much of what we used had to be written,” he said. “We would all have to create it all ourselves, but more and more of it is available to everybody which lets us focus more on the world itself.”
Ubisoft too sees the impact modern toolsets and improved graphic fidelity have had on games this generation.“The last generation of games grew in size and allowed creators to create bigger and more believable worlds,” said Pierre Fortin, the lead architect on the Scimitar/Anvil pipeline in Ubisoft’s Quebec City offices. “Graphic fidelity improved drastically, a testament to the skills of both the artists and the graphics programmers of the industry. Physically-based rendering and detailed global illumination were widely, if not universally, adopted by AAA games in the last years.”
But, Fortin added, while those were the most consumer-facing advancements, the way developers now build worlds saw an even bigger leap forward. “The way we build games has drastically changed in the last decade and some of the biggest innovations I’ve witnessed are those happening behind the scenes that enable creators to create those games. The advent of procedural generation and growing use of AI in our production pipeline allowed us to work much more intelligently, putting our efforts in creating higher-quality assets instead of quantity. This directly translates into an increase in production value. Whether it’s the way we create textures, synthesize facial animations, generate optimized objects (level of detail), and create worlds, automated synthesis of content has definitely been a game-changer.”
The advances delivered to game creators this generation — like the rising ubiquity of game engines, expansive power of new graphical processes, and the ability to iterate more quickly than ever before — certainly helped shape the games people play today, but so did the limitations that developers ran into while trying to turn dreams into code.
Image courtesy of Ubisoft Montreal and Ubisoft
Ubisoft’s Fortin said he believes that today’s tools and technologies have reached their limit and are starting to “crack under the pressure of the amount of data we are pushing through them.” Game worlds, he said, are getting more complex, realistic, organic, and that means that the systems used to animate characters are also growing more complex as they try to navigate and interact with these growing worlds. The result is that the amount of hand-crafted data required to keep things working is getting to a point that simply isn’t sustainable anymore, he said.
Fortin points to Ubisoft’s Assassin’s Creed Origins and Odyssey as examples of how this has played out this generation.
“These worlds are incredibly large while consistently improving the amount of details they exhibit,” he said. “On the Assassin’s Creed brand, where before we would handcraft all of those worlds, meticulously placing props, buildings, landscape features, etc., we now rely on human-driven procedural generation to output a similar result. This allowed us to significantly increase the size of the worlds, giving our teams the time to focus on quality over quantity. However, the result of this procedural generation is a static, fully optimized world which limits the changes that can be done at runtime. Since there’s no realistic way to regenerate all of this data on current console hardware, we are limited when it comes to physically changing the world, for example through adding new cities and modifying the landscape.”
Image courtesy of Ubisoft Montreal and Ubisoft
To make matters, if not worse, at least more complex, this generation of big AAA titles haven’t just grown in size and complexity, their lifespans have increased. Where entries in big franchises like Assassin’s Creed were once considered to be one-and-done single-player masterpieces, there is an increasing expectation that they live on beyond their release story and play. And that runs into the most basic of problems: space.
“One of the often-overlooked technical problems is the packaging as well as the deployment of updates and patches of such games,” Fortin said. “On titles such as Assassin’s Creed Odyssey, we are using the entire Blu-ray disc’s capacity in order to maximize the runtime asset streaming performance. In order to do so, our build packaging process involves complex post-processes that splice and merge the data in ways that optimize seek times on traditional HDDs. Thus, it is not as simple to replace a single asset file and/or property and it is even harder to replace a big chunk of data with lots of dependencies in a pre-baked world in an update down the road. On Assassin’s Creed Odyssey, we had a lengthy post-launch plan filled with regular content drops that were produced after the disk had been manufactured. Compound that with a strong desire to have the smallest update size possible to make it easier on our players, and you get yourself in a rather complex situation that we managed to tackle successfully, but not without limitations, on the post-launch content we could deliver.”
Despite technological advances, perhaps in some ways because of them, Urquhart believes that one of the biggest limitations developers today face is one of balance. “It’s time, resources, and fidelity,” he said. “There’s so much you can do with the current hardware and with the current engines that you have to pick your battles. You could render the most beautiful polygon you've ever seen in the world, with 972 materials, it's like the perfect wall and it refracts the light and its shadow is perfect, but that's not a world, right? So we’re always thinking, ‘How do we optimize that? How do we optimize that experience for players?’ It’s the fidelity versus frame rate question and that’s the wall.”
Put another way, Urquhart calls it the scientist-versus-the-engineer problem. The scientists invent amazing things that don’t necessarily have to worry about the cost of implementation, and then the engineers have to figure out how to put it to use in a real world with real costs.
“We have access to a lot of cool technology, but as the technology gets more complicated, how do we then actually take that down to something that'll actually run games? So those are the two walls of technology as the games get more realistic,” he said.
Image courtesy of Google
Lack of story
While Stadia’s Cammarata bemoans the lack of technical innovation in support of dynamic storytelling, he sees the roadblock causing that innovation stall tied more to hardware than software. “I’ve wanted to create dynamic storytelling for decades, and it’s still not there yet,” he said. “The day is coming where we have one game for our entire life and a single ‘storyteller’ that crafts narrative and content for us to consume, customized to our preferences, mood, time available, and platform. The challenge is that such an overarching ‘director’ needs a deep understanding of the human condition. I work very close to the forefront of where machine learning and AI are being applied to games, and we’re still far off, in my opinion. To craft a great story, you need to understand how to ‘be’ in the world, how to be afraid, or feel proud, or feel comfort. With current technology and the direction we’re going, I’m not sure I see a path that gets us there quickly. ”The issue, he said, is that improvements in hardware can only happen so fast and that is what dictates the ability to innovate in fields like machine learning and artificial intelligence.
Omar Skarsvåg, technical art director at Stadia Lab, who has specialized in animation technologies, character development, and pipelines for two decades, points to the same sort of issue that Urquhart notes: It all comes down to balance, but as with Street, he believes that’s because developers are “hardware bound.”
“If your ‘console’ scales to whatever the software needs to run at realtime, we can do so much more,” Skarsvåg said. That hardware issue is exacerbated by hardware lead times, Cammarata said. “The last three console generations have lasted seven years. From a hardware perspective, that means you’re working on hardware that’s at it’s best on day one, and on average you’re working on a spec that’s three and a half years old,” he said. “That’s costly - but what it gets you is a stable, solid platform that your game will run on, the same, for every player. PCs have always been more of a continuum - but that leads to a proliferation of configurations you have to support, which becomes a nightmare.”
He points to Stadia’s approach, which is working to remove the limitations of a console by moving everything to the cloud and refreshing the hardware automatically, as an approach that will allow creators to better realize their increasingly demanding visions.
Lack of standardization
A lack of standardization and closed platforms also were noted as major limitations for developers this generation. 2X2’s Ratković said he’s surprised that there still exists a lack of robust cross-platform rendering, something he said he hoped would have been completely solved by now.
“I'm a bit of a raytracing nut, so I'm always excited to see innovation there,” he said. “My biggest worry is that currently, ray tracing has a bit of proprietary feel to it. For example, amazing things can be achieved with NVIDIA's RTX and OptiX, but trying to do cross-hardware and cross-platform ray tracing can incur a huge performance penalty. I'd really like to see better cross-platform ray tracing. A standardized ray tracing API would go a long way.”
“I feel like the visual side is about to go through an interesting transition,” Bithell said. “AAA is about to leap headfirst into ray tracing and a leap in graphical quality. That brings with it a need for more expertise, which might be trickier at the indie end. Generally, we do fine by sidestepping this with stylization and bolder art choices.”
Because of the typical size of an indie team, these smaller development teams also need to be more selective on their focus and how they expend their energy.
“There are lots of things that are fun to play with but aren’t always usable for a team like ours,” said House House’s Disseldorp. “Of course, more things are possible to a motivated and technically focused team. But adding specialized technical features to a game places serious restrictions on what else you can do at the same time, both because it eats up development time and potentially your computer’s performance.”
But he, like many of the indies we spoke with, saw this issue as less of an impediment and more of a way of shaping their games.
“Our goals are always to work with the constraints of the tools we have at hand,” he said. “Constraints are useful, and interesting constraints produce interesting games. If we decided to only make games with 2019 technology for the next decade, I’m sure we could still find something interesting to do. And lots of game makers put that idea into practice by making things in the Quake engine or with the restrictions of the NES even today.”
Image courtesy of Counterplay Games and Gearbox Publishing
Gearbox’s Pitchford said that while technical limitations will always exist, they can serve as a powerful tool for stimulating innovation. “I remember when we were helping to port Half-Life to the Sega: Dreamcast and the challenge was that the Dreamcast’s VMU (the device that would store save-game information) had a total of 128k of memory and was expected to store programs and save games for a number of different games,” he said. “Meanwhile, the average save game file for Half-Life tended to come in at around 2MB. There was a deadline approaching and usual approaches, such as optimizing the engine’s save game system, was just not something reasonable that could be accomplished in the hand-full of days available between when we identified the task and when the problem had to be solved.
“The solution was a sort of dictionary approach our Tech Director conceived of – basically using the fact that the game was going to be on an optical disc with tons of extra space on it to store a library of possible save files and only record the tiny difference between that data and the save file on the VMU. He effectively compressed what should’ve been a 2MB save file into tens of kilobytes each.”
As famed composer Leonard Bernstein once said, Pitchford noted, “To achieve great things, two things are needed: a plan and not enough time.”
That’s also how Ubisoft’s Fortin views limitations: just another opportunity for innovation. “I am often surprised to see the clever innovations and optimizations made by my colleagues, whether at Ubisoft or in the industry in general,” he said.
It’s clear that as games become bigger, more complex creations with longer lives, one of the most pressing issues game developers are running up against is the sheer amount of data that needs to be generated, combined, and processed.
Fortin’s hope is that major advances in machine learning and artificial intelligence will help with some of that heavy lifting for the coming generation of game development.
“Some of the most promising prototypes I’ve seen lately all use machine learning to synthesize data of impressive quality,” he said. “I see this trend as a natural extension of the work on procedural generation. Traditional animation work isn’t going away anytime soon, but the tedious animation work, such as creating filler animations, will be the work of machine learning algorithms in the near future. This will allow animators to put effort where their expertise is really valuable and where human input is needed. This is just one example of what I would best describe as AI-assisted, human-driven, procedural data generation.”
He also believes that machine learning has a role in areas as diverse as game balancing, exploit detection, automated game testing, and even runtime game AI.
“There are a lot of possibilities of leveraging machine learning both in production and at runtime,” he said. “It may not be the silver bullet to all problems, but it is nonetheless an area worth investing heavily in.”
That’s in part because as the industry rapidly approaches solving the uncanny valley with photo-realistic games, the gap between graphical quality and things like animation and AI quality will likely grow; essentially creating a new sort of uncanny valley.
“However great and photo-realistic your graphics are, if your NPCs look and behave like robots, the illusion will be lost,” he said.
Image courtesy of Riot Games
Cammarata hopes that this next generation of game-making tools will become even easier to use, demystifying the process of making games and broadening the pool of people interested and able to make games.
“The more people who can get past the arcane magic of creating a game, the more unique viewpoints we can get, the more cultural and content diversity we’ll see. I want to see games based on new and different concepts, like Hindu mythology, or the Australian outback and Dreamtime, or the bacteria that thrive in the 200℉ water around deep-sea vents ... any one of the crazy and exciting things that pop into the minds of the most creative people on the planet, whether they are traditional game designers or not,” he said.
Urquhart points to some basic improvements in engines that could go a long way to help both those new voices and established developers so they can spend less time “reinventing the wheel” with every game.
“I know it sounds dumb,” he said. “But every time we have to recreate how people sit in chairs, we’re not focusing on other things. When developers make RPGs, most of those games have inventories and so that means everybody is always creating a new inventory system and a new inventory UI. Or take a door for instance. We all recreate that door. You know why only some of us put horses in games? It’s because horses are super complicated and hard and even when you do them well, they’re janky. So literally there’s always a conversation when we’re starting a new game, ‘Are we going to put horses in it?’ I know this is very, I don’t know, non-sexy technology, but I think that the more we can start with as baselines in these things, the more it lets us focus on the things that we as developers might be awesome at.”
For Riot’s Street, it’s called the rock problem.
“Developers spend a lot of time and money generating assets for games that don’t need to be solved over and over again,” he said. “Any open-world game is going to need grass and rocks and trees. I don’t want to see the same commoditized rock asset in every game, but at the same time, it doesn’t feel like we should have to have the ‘How many rocks do we need?’ discussion every time."
He also said the company spends a lot of time essentially recreating the same tools. So what they’re hoping for is more focused on a better set of baseline tools for account connection, social features, and player behavior tracking.
“It would be great if those sorts of features were well-supported by commercial tools,” Street said.
Others hope to see improvements in a variety of things including 2D animation, AI-supported development to help streamline production, and even a reduction in console-power consumption.
In some ways, Pitchford said he’s most excited about the fact that these generational shifts in innovation are more about iteration then reinvention.
“I’m always thrilled when I notice Moore’s Law plugging along as we can always benefit from more power,” he said. “I’m also excited that as the traditional games industry has sort of coalesced around a fundamental architecture for CPU’s we’ve been able to get more and more gains through iteration on our software foundations (rather than starting over every generation). When our foundations are more stable, our abstractions can become more robust and this translates to less of our game budgets going towards foundations and more of our game budgets going towards the kinds of things that players can see and feel and experience directly. I want to put every dollar we invest on the screen because I want our customers to feel the best value for their money when playing games from Gearbox. I’m also really excited about how further progress with distributed compute and concurrent compute will allow us to continue to reduce friction for customers that interact with other games in different places or on different platforms or when gamers want to move from one platform to the next or be able to upgrade hardware without necessarily having to rebuild their gaming catalog.”
Image courtesy of Gearbox Software and 2K Games
Pitchford believes that the coming generation of game development will be in part defined by emerging approaches to game logic, though he cautions that new tools can add layers of abstraction that might make debugging harder down the line.
“Unreal Engine’s ‘Blueprint’ is an incredibly powerful visual scripting system that allows cross-discipline creators (i.e. – non-coders) to build and deploy game logic very quickly and very easily,” he said. “Blueprint may be one of the greatest game logic prototyping tools ever engineered in the history of game development. It effectively allows anyone to make a game.
“The most disciplined development would respect that the way we build prototypes is not necessarily with the most robust engineering and can be more prone to logic flaws (bugs) or inefficiency (low performance) and anticipates that with every successfully prototyped system or feature there must be time accounted for in the schedule to rebuild the prototyped system or feature into something that is sufficiently optimal and sound for production and maintenance over time.”
Pitchford added that the natural conflict between real world issues and best practices can make even the greatest video game prototyping tool quickly become a project’s biggest Achilles Heel, if not approached correctly.
Complexity of tools
The improvement of tools becomes a necessity as the products they create become more complex, Fortin notes.
“If the current trend in AAA games is any indication, I think we will be seeing bigger, open, persistent and connected worlds – or universes even – which evolve in a player-centric fashion,” he said. “The last games of this generation stretched the limits of what is sustainable through traditional development. The biggest challenges won’t be how to load, package, or even render those games, but how we produce their content both from a quantity and quality perspective. To illustrate this: You may be able to build a shed using a traditional hammer, but good luck building an office building without a nail gun. Thus, I expect to see quite a lot of new paradigm shifts in data production.”
Cammarata also believes that the use of machine learning in game development will become a defining facet of the coming generation.
“We’ll see a hybrid approach of ML and game development - game developers will find themselves able to generate orders of magnitude more high-quality content for less work and lower cost,” he said. “The task of creating content will become learning to work with new tools that you guide and shape, rather than crafting assets directly. Developers who can think parametrically, and who understand procedural content will lead the way. Game teams will reorganize around curation of the blooming pile of content, rather than racing to finish the asset list and burn down the bug count. We’ll find we have too much content, rather than too little, and we’ll have to figure out how to carve out whatever doesn’t fit, rather than scramble to build enough to realize our full vision.”
“We’re just on the cusp of seeing machine learning reach production - first, we’ll see it in game tools, then we’ll see it in games. ML-based pipelines are a quantum shift in terms of how content will get created.”
Image courtesy of Ubisoft Montreal and Ubisoft
Deep learning is Mendoza’s prediction for the coming generation of game development. “I think deep learning will play a big role in the future,” he said. “It's already being used for anti-aliasing techniques and starting to be researched for character animation, and it's only going to get better from there. I hope deep learning plus AI behaviors could define the next generation. AI behavior systems have pretty much remained the same for the last 10 years. I feel it's one of the least improved fields, and if we can bring those two together to create more intelligent/believable NPCs, that would be a huge win.”
Driving all of that innovation in video game creation is a technology that constantly advances, turning each year into a new golden age of game development.
The ubiquity of easy-to-use development tools and increased access to gaming platforms have exploded the reach of games and the opportunity for people to make them. Jumps in both the sheer power of graphics hardware and prowess of the software have so fundamentally improved the visuals of games that the surging technology is on the verge of creating new sorts of uncanny valleys. Technology has even changed the way players play, creating new sorts of interactions that give birth to not just new genres, but new forms of multiplayer experiences.
Ubisoft’s Fortin notes that the conjunction of advancements in both hardware and software are driving forward gaming as a whole as a self-feeding loop. But he also argues that the opposite is true.
“How important are games for advancements in technology?” he said. “Whenever I attend developer conferences, I’m struck by the amount of businesses and industries looking at what we, as an industry, are developing. Whether it’s advancements in real-time rendering making their ways to the movie and the CAD industry, to augmented and virtual reality (AR/VR) used to improve worker safety in heavy industries, to our own development tools such as Clever Commit being used by other companies to spot code defects early on or, simply the demand for ever more CPU from games pushing the technical boundaries leading to mobile devices with better hardware for general purpose work, I think it’s undeniable that technological innovations through game developments have impacts which transcend the games themselves.
“This, to me, is an argument that games are driving technological developments as much as they are driven by them.”
In our next article, we’ll speak to great thinkers in the realms of design, research, development, and play about their thoughts on how views and approaches to video games, development, and gameplay will evolve over the coming years.
Get Unreal Engine today!
Get the world’s most open and advanced creation tool.
With every feature and full source code access included, Unreal Engine comes fully loaded out of the box.