Epic released a demo for the latest iteration of their game engine, Unreal Engine 5 (UE5). The demo looks magnificent and has a bit where the roof opens up and lights a room with 500 detailed statues. Once I looked at Twitter I was informed that the demo, apparently, wasn’t all that good. The objections ranged from skepticism (“it doesn’t do what it says”) to pure absurdity (“it took 2000 people to make this.” Though the person I saw making that claim later withdrew it). One broader theme that emerged was that the demo’s claims were overstated and that it is very important not to disappoint The Gamers™ because of how they might act in response. The whole experience left me feeling like this is why we can’t have nice things. None of the wet blankets are the usual suspects for depressing gaming discourse, and yet the after image of the demo had not yet left my eyes before I was being told that the thing I liked was actually a bad thing. A proper analysis of the demo requires someone thoughtful and experienced, so I will instead write about why we seem to be so determined to be miserable about a really cool video and try to put it in context.
The first reason we might want to be a killjoy about the demo is a skepticism of marketing campaigns. Epic wants us to think that UE5 is the greatest thing ever (or at least better than our next best alternative) and will show what it can to create that impression. What is odd about this kind of skepticism is that relatively few of us will be in a position to license a game engine. An engine may be an important component for a product you like, but I would be willing to guess that you have not put any thought into the compressor your refrigerator uses, and that you did not wonder if the last movie or TV show you watched was shot with an Arri, Panavision, Sony, Panasonic, Red, or Thompson Viper Filmstream camera. For most people talking about the tech demo, there is no product for them to be skeptical about.
However, gaming is unlike fridges and movie cameras in that being a fan of games can mean following at least some parts of development closely and even modifying the finished product. Epic has built a brand around high performance and visual fidelity in a way that a company like Unity (whose tech demos also look attractive) has not. The fact that the demo is rendered on a PlayStation 5 is also being used as a proxy for the capabilities of the next generation of consoles, and so a broader interest is understandable on that front. The skepticism is still perplexing. Epic’s demo will never be a game you can play and isn’t being sold as one. UE5, just like the PlayStation itself, is a tool that creators will use to bring audiences certain types of experiences. The question as to whether or not those experiences will be worthwhile will be answered over the next few years as games that take advantage of these new tools are released.
The worst take is the one that says an overwrought tech demo will inflame the worst behaviours of gamers. Specialists should be able to have a conversation about the capabilities of their tech and not have to reframe the conversation to accommodate the worst behaviours of a potential audience. The UE5 demo is under ten minutes and does not concern itself with features like multiplayer or an enemy AI and so I would not hold my breath for a 50 hour experience identical to the demo. That understood, I have no doubt that both the PlayStation and the Unreal Engine really do everything that was on display and the question will be how well developers draw my attention away from the things they didn’t have time to polish to absolute perfection. Consider what is depicted in this demo for Unreal Engine 4 on a PlayStation 4. How much of this would seem out of place in a modern title? How much of this would seem below the standard for a modern title? The expectations being set by the UE5 demo are for creators and what they can achieve with their tools, not audiences.
What did the demo show?
Very little of the commentary on the demo mentions the introduction in the video where team members talk about the technology that is about to be demonstrated. This is a shame because it really is the key to getting the most out of the demo. The headline features are two technologies called Nanite and Lumen. Nanite is responsible for all the detail in the rocks and statues, while Lumen is responsible for the lighting, though the introduction does explain how one complements the other.
If the reaction is “games already look like this” then it is a testament to the abilities of developers to bamboozle us into enjoying ourselves today. Acknowledging my lack of expertise on this subject, the most impressive graphics in today’s games are the product of taking very high quality art and applying all sorts of tricks to make the underlying object less demanding on the machine while still looking close to the original. The claim from the demo is that Nanite will remove the need for these tricks and allow the original high quality art to be used.
Lumen is a technology that enables dynamic global illumination. As with Nanite, the skill of modern developers presents the illusion that this is less of an improvement than it really is. To illustrate the benefit, I’ll use an anecdote from a small film I worked on:
One of my mentors in cinematography was off to the side in one of the sets that was not lit but was catching a little bit of the bounce off of the work lights. He had his personal camera out and was trying to produce a still life but kept frowning into the display and changing the settings. It’s hard to explain why unless you’ve lit a scene, but I immediately knew why he was frustrated. The bounced light was perfect and there was a nice little arrangement that he wanted to capture with his personal camera. Unfortunately the whole reason cinematographers and gaffers exist is because cameras cannot capture the world as we see it and require lighting departments to augment the world so that it can be captured by the equipment. No matter how many settings he changed, my mentor’s beautiful shot would not be realized unless he commandeered the film’s lighting for his hobby shot.
Similarly, light in games does not behave the way it does in the real world. The least demanding solution is to get someone with a good eye to prepare a gorgeous environment, fix the lighting in place, and call it a day. Since games are interactive, this works right up until we start exploring the environment and then our expectations about shadows and how the light should play over the various surfaces we interact with break the immersion. The more interactivity, the more work the computer needs to do to render the world. Games as we currently experience them are a series of tricks so that the parts we expect to exhibit more dynamic behaviour get extra attention, while the rest of the environment appears to be illuminated by the same source but uses the less demanding methods. Lumen is supposed to make it so that if you turn a light on in the game, it will be behave a lot more like a light in the real world, up to and including light bouncing off of diffuse surfaces and illuminating other parts of the map. It’s the tool my mentor wanted: shoot it the way you see it.
The demo does a good job of showing how impressive these technologies are in terms of making things that look good, and yet I am not sure this is the most exciting part about the announcement. The most exciting benefit of UE5 seems to be the time that is saved by not having to create the illusions that the current technology requires. This is not the kind of thing you can directly show in a demo, and yet it will likely create the greatest benefit to people who will play the resulting games.
Think about all the time that is spent turning a high quality asset into something that your computer can handle, and all the time that goes into lighting a level in a way that you are immersed in it. If Nanite and Lumen live up to their promise, the time spent on these tasks should be reduced or eliminated, meaning that people are freed up for other creative work and the time between idea and implementation is reduced. The headline is not ‘UE5 can produce bumpier rocks and lots of games with collapsing ceilings’ but rather ‘UE5 frees up talent to do the things that entertain you the most.’
While these gains may seem theoretical, recall this anecdote from Visceral’s cancelled Star Wars game, Project Ragtag. Early in-game footage had a moment where the main character touched a door on his way out. That single door touch apparently took months to accomplish, while other important work fell behind. One of the features described in the UE5 demo are improvements to contextual animation, and is demonstrated with a similar doorway touch (the 5:16 mark). It overstates the case to say that better contextual animation tools would have saved Project Ragtag, and yet time saved in one area can help a team focus their attention on problems that can become existential threats to a project.
We’re allowed to like things
Time savings are hard to display in a tech demo. A demo can display how certain technologies can be used and does its job if it upsets its audience’s assumptions about what can and can’t be done. In so far as it is used for branding for a non-specialist audience, the demo should get people excited about the potential for what creators can do with the new technology.
There is always room for managing expectations, but the dour response to Epic’s unveiling is the game engine equivalent of telling a wrestling fan “it’s all fake you know.” The cautions against inflated expectations are best when they have a counterweight of discussing what actually is exciting about the technology. Because of the lack of technical commentary, the reactions come off as a bunch of moping and cheap shots at what is a genuinely impressive demo. Very little of the discussion rises above some variation of the old saw that ‘gamers care much more about gameplay than graphics.’ If this were strictly true then companies would not be grinding their developers into powder to achieve a certain level of graphical fidelity, but this is beside the point. Unless the game is called Fortnite, gameplay is not Epic’s job here. Epic’s job is to tell developers “look at this amazing new tool we want to sell you” and to tell the broader audience “look at what’s going to be possible with this cool new tech.”
I get that there is a takes industry and that the biggest rewards to go the fastest response, no matter how asinine, but we should be expecting more from these discussions. The demo itself has people who worked on the technology discussing what the advantages are and so we have been given every opportunity to understand what is exciting about these tools. The reward for doing something interesting should not be a collective eye roll because everyone was too busy to think about what they just saw or ask an expert. I’m excited about UE5 because the demo did a good job of displaying improvements that I think will let the people who have already shown they can make things I like spend more time making things I like and less on wrangling the tools to realize their ideas. If the benefits fail to materialize there will be plenty of time to talk about why the games failed to live up to the demo’s promise.