Working to limitations- Are realistic graphics always better?

Over the past decades graphical processing has come on leaps and bounds, with advancements in how it is done, such as the invention of physical based rendering and faster graphical processing units (GPUs). in fact, GPU speed has come on so far that it currently follows Moore’s law (that the number of transistors on a microchip will double every two years), but Huang’s law, named after the CEO of Nvidia, due to the fact that GPUs have increased speed by 25x in just 5 years.?

This processing increase can clearly be seen in the quality of games, you only need to look at Final Fantasy 7 from 1997 and its remake, released in 2019.

This is an example of where graphical enhancement has clearly changed the quality of experience for a player, however, comparing 1997 and 2019 is a bit unrealistic in terms of what I’m actually addressing. Obviously in a time when we can produce such amazing graphics as on the right that players can interact with in real time, no game developer will opt for the left. But is it always needed to go as far as the right? Does pushing the limits always make for a better gameplay experience?

While it’s pretty conclusive that older games are visually worse than modern games, their players didn’t seem to enjoy them any less. Games designers were forced to work within bounds in order to make a playable and fun game, whether this be a limit on memory, processing speed, money or time. Limitations still exist both in making AAA games and indie games, they're expensive, risky and can very easily go off track if people aren’t careful, so the kind of graphics a dev uses and how big they let their project expand should be something on everyone’s minds from the beginning to the end. That being said, AAA studios clearly have far more resources, money and reputation to give them more wiggle room, indie studios are forced to work simultaneously within the box and outside of the box and work with scopes in a way most AAA developers won’t have to, and it often leads to creative and interesting results, especially because indie teams are often made multitasks job roles.

This is where the psychology of limitations comes in, limitations often mean people try to maximise how good an idea can be within the given constraints, a lack of constraints means people often take the path of least resistance, not necessarily the best one. Working to restraints means you’re constantly thinking about how to get what you want and the best ways to do it, removing unnecessary things in favour of those that are actually beneficial. This works amazingly in games, sort of similar to abstraction, ‘ignoring irrelevant details and focusing on key ideas and information’. This means focusing on gameplay, focusing on story, focusing on characters, instead of focusing on unimportant details when you simply don’t have the resources to.

In terms of graphics this means if you’re running low on time or money investing your time only into the features of the background that are important, and focusing on gameplay, in many cases may be more beneficial. Don’t go for realism if it's not important to game play, use a lower poly art style and save your GPU and your 3D artists a headache of putting together a fully textured dragon statue that the player will see for 2 seconds and says nothing about the world. Backgrounds are important parts of worldbuilding, but the player is far more likely to notice if they clip through the ground or if the enemy AI is painfully bad. A more stylised and less realistic art style can also be a selling point for a game, it makes it stand out from its counterparts and allows for more creativity in character and environmental design.?

Changing to simpler, abstract art styles is far more commonly seen in indie development at varying degrees. this can clearly be in Untitled Goose Game, a game about a chaotic Goose controlled by the player, trying to terrorise citizens of a town, that earned viral acclaim on its release. The game is simple, has few controls, allowing the player to move, honk, use their beak and spread their wings, and the art style is entirely low poly, flat coloured meshes. This is a clear example of abstraction which allowed the devs to focus more on gameplay, and didn’t take away from the content of the game at all. This art style also increases efficiency, when the game starts, all of the objects are loaded in and ones that aren’t currently in use are put out of the player accessible map. This decreases load time as instead of having to render the model it is already pre rendered, the models are simply moved in place when needed (e.g. a mug breaking is 2 objects, when the mug breaks the broken model will be swapped in), a large part of why this is possible is because the game takes up less memory due to its art style, so has the memory to spare to hold objects that aren’t in use. This is only possible because the devs were kept within a limit of what they could actually do, if they had tried to push that limit and expanded too far, adding more environments, more detailed graphics, etc. they may have been left with a worse game overall, or no game.

Looking at an AAA game that used realism to its disadvantage, we have CyberPunk 2077. The game was advertised? as having a large open world, with realistic textures and multiple paths through the game. But on release the game offered little in terms of customisation, low NPC interaction outside of scripted missions, and issues with the wanted system. This is all ignoring that it was also riddled with bugs, ranging from small to completely fatal. Unlike Untitled Goose Game however, CyberPunk 2077 is one that benefits from a realistic art style. The problem was that going for too realistic and too open-world actually ended up with a world that felt less lived in. Many players said that the world itself felt barren, which is an issue with such large projects. The environments didn’t show enough about the kind of people that lived there or the culture of the place. Having such large open spaces where you can access every room of a building sounds good, but having every room say nothing about the people that live there, in the end may be worse than only having one room that is decorated well. It is an example of where a game didn’t realise it was going over its restraints early enough in the project, and aimed for larger than was realistic to handle.?

Compare that to What Remains of Edith Finch (Giant Sparrow Games, 2017) an indie game that uses realism, it does the kind of environmental storytelling I described CyberPunk failing at perfectly by realising its limits and adjusting its scope. The entire game takes place in Edith’s old family house, exploring different areas of the winding house linearly, one by one. You are taken through each room that belongs to different family members that have been passed on but been completely preserved, and though we get to see very few of the characters, their rooms and the minigames within them are designed to tell you about who they were as a person and the life they lived. The house is incredibly detailed, but not meaninglessly, in places that are unimportant to the story there is very little detail, and there are no rooms in the main house that aren’t important to the story line. it's a game that clearly worked within limits but also makes itself appear not to, it’s a place where realism and detail can exist, because that detail is so core to the storytelling, not because the developers wanted to show off.

In conclusion, like in Edith Finch, there are places where realistic graphics can be good, and this is usually when they actually benefit gameplay. Realistic graphics have their place when the developers want their players to feel grounded in the world they’re creating, but it should not be the default. Developers should explore different art styles, and which would work best with the gameplay, the story they’re telling, the tone they want to set, and whether the style is feasible to do, making sure the cost of it isn't taking away from gameplay. Playing around with styles and combining them can lead to interesting results and better, more memorable games, and should not just be reserved for games aimed at children like it currently often is.?

Further reading and references:

GPU speed increase ref: https://www.motherjones.com/kevin-drum/2020/09/moores-law-is-dead-long-live-huangs-law/

Constraints and innovation; https://hbr.org/2019/11/why-constraints-are-good-for-innovation

Out of bounds untitled goose game footage: https://www.youtube.com/watch?v=wvYqf3NLKNM