Saturday, April 19, 2014

Truly Next Gen Ideas

As video game consoles shift into yet another gear (and PC hardware revs higher and higher), consumers are right to ask what "next gen" really means this time. It's hard to remember now, but the the previous console era was truly revolutionary: serious online gaming, HD graphics, motion control interaction, and downloadable content/games were all new in the PS3/Xbox360/Wii era. What do we have in store this time? The truth is that nobody knows quite yet, but I'm happy to offer my guesses.

Material Worlds

In the vast majority of games made so far, a wall is just a solid impassable object. It is a fixed barrier with no properties other than its placement, size, shape, and texture. This is already beginning to change. In early PS4 and Xbox One games like Battlefield 4 and Infamous Second Son, some special structures are designed to be destroyed. While this is a step in the right direction, the majority of the structures and objects remain indestructible, from the strongest reinforced concrete to the lowliest shrub.

There is only way to resolve this long-accepted trope: to have true destructible environments. That means that every object in a game needs to be "built" out of simulated materials. By the end of this generation, I expect that video game walls will have studs and drywall and pipes beneath their skin, just like walls IRL. Plant leaves will shred to pieces when shot at and no car will completely shield you from incoming rockets.

Interestingly, there is actually one game built this way already on PS4: Resogun. Every enemy, object, and player character in this 2D revolving shooter is made out of voxels (volumetric/3D pixels). That means when you shoot an enemy ship, it blasts into the hundreds of pieces of which it is constructed. There can be dozens of enemies on screen at once, and the game rarely dips in frame rate; an impressive bit of tech for a seemingly simple game. Of course, each object in this game is made out voxels with similar basic properties, so there's still a long way to go.

A 3D materials-based game engine suitable for an FPS or an open world game would be tremendously difficult and time-consuming to develop, but could pay huge dividends in immersion and graphical fidelity. Most crucially, it would allow for the creation of a physical world which could react properly to being battered by crowbars, shot at with small weapons fire, and blasted away with heavy weapons. It would also eliminate some of gaming's oldest design conventions in interesting ways. I can't wait to see how it plays out.

Procedure

Games have played around with procedural level design for years, particularly in the "dungeon-crawler" genre where the dungeons are often randomly assembled (e.g. the Diablo series). The sacrifice here is that the quality of the level design is highly variable, and usually the levels end up feeling bland or "same-y". The computational power now available to developers, either on-system or in the cloud,  opens up the possibility for much greater variability and success in spontaneous level design. Now a game could be designed to procedurally generate 500 levels using dozens of variables, "playtest" them virtually and select the ones with desired characteristics (most difficult, least amount of available cover, greatest vertical variability, etc.) in a matter of moments.

This will greatly enhance the variety and enjoyability of games using random generation, and most importantly open up this type of design to the types of games which have historically relied upon tedious and costly development. Imagine if an Elder Scrolls game could see that you need one more Daedra heart to smith that badass longsword and spawned a new dungeon, tailored to challenge your avatar's skillset, ending in a boss fight with a demon who drops the heart. Or imagine if every planet in the Mass Effect galaxy map had multiple bases and colonies who could really use the help of a Specter. High level procedural generation can make these things possible, and best of all give each player (and playthrough) unique experiences and stories of their own.

New Voices 

One of the grand promises of gaming is to craft your own unique story within a game world. Almost every game now has "RPG elements" like character customization, classes, leveling up with experience, and item acquisition/unlocking. Generally, games have gotten very good at providing different ways to play, but in therms of storytelling they all constantly fight a losing battle between personalization and quality.

Games like Mass Effect and the Elder Scrolls series are absolute favorites, but are also prime examples of this problem. In Elder Scrolls you have a huge (sometimes petrifying) amount of options in designing your avatar. There are multiple races to choose from, different sizes and shapes and skin colors and hairstyles and scars and ok I should probably just click DONE and start the game already. In my first playthrough of Skyrim, I was a lizardman (Argonian, for the initiated) named Seldon who looked legit OG dragon-born. That level of personalization is awesome, and the game does a pretty good job of reacting to your race and gender. The trade-off? You never speak. Bethesda isn't going to hire the twenty voice actors needed to read all that dialogue for all 10 playable races and both genders. And they shouldn't, because that money is better spent on creating the world environment and writing the script.

On the opposite end of the spectrum, Mass Effect games tell an amazing story and give the lead character, Commander Shepherd a strong voice. The trade-off here is that you have exactly one meaningful character customization option: gender. I appreciate that Bioware took the challenge to hire two voice actors to read all that dialogue when most players will only hear the male Commander Shepherd (not my sexism, just a fact; and most Mass Effect aficionados will agree that "FemShep" gives a stronger, more nuanced performance). Also, all the NPC's are forced to call every player "Commander" or "Shepherd" for three straight games, which is much better than "The Champion" of Dragon Age 2, but still not ideal. The game offers story-altering choices, but they don't branch too far from the tree. It's definitely a great "role-playing" game, but the role you play is fairly similar to those played by everyone else.

In order to provide high quality and highly customized story experiences, multiple technological advances are needed. I think the hardware may now in place to be able to create those solutions, so it will be in the hands of game developers to invent software solutions capable of delivering this promise. I think the answer is two-fold: artificial  voice synthesizers and simulated artificial intelligence.

Voice synthesizers have been around for decades, starting with text-to-speech devices designed by Ray Kurzweil and others to aide the visually impaired. Now everyone now has Siri (or something similar) in their pocket. I'm not going to pretend that these programs are going to replace Morgan Freeman anytime soon, but it is reasonable to assume that they will continue to advance, and within the proposed decade-long span of the current console generation they may be just good enough for video game use. If and when high quality synthetic voices become available, a game like Elder Scrolls could finally give a voice to the playable character and allow NPC's to call you by name.

That's the easy(ish) part. If you really want to have a dynamic game with unexpected turns, endlessly branching stories, and personal character interactions, the non-player characters must come alive via a crude form of artificial intelligence. Open-world games of the current era have blasted well beyond any reasonable expectations for the volume of dialogue writing. I can't imagine how long the script for a game like Grand Theft Auto 5 is, if you include every side quest, pedestrian quip, radio station, etc. It's really getting ridiculous, and ridiculously expensive, to make an open-world game and fill it with so many voices. There are always sacrifices, and in the case of the GTA series, it's pretty clear that the money gets spent on presentation and world design and not on gameplay improvements. As such, I find GTAV to be an impressive achievement but ultimately not very enjoyable to actually play.

Combining artificially generated NPC behavior and dialogue with a believable voice synthesizer could free game developers from writing volumes of banal background chatter for the sake of making an open world feel more real.

Summary

There is an over-arching theme to the ideas I've brought up in this article: my dream for truly next-gen gaming is that game creators will set up elaborate worlds with interesting characters that allow the player's actions and decisions drive a unique narrative each time. That and really, really, really high polygon count assault rifles of course.