Over the last few years, I’ve been trying to get better at art – in the specific sense of visual art, that is – drawing and painting and, most frequently, digital art. I think I’ve succeeded at the goal of getting better, though I still fall short of what I want to be able to achieve. Over the past few weeks, I’ve been trying to understand 3d art – in this case trying to construct an environment that’s visually interesting and feels reasonably natural to exist in. Most of my abortive forays into 3d work haven’t really concerned themselves at all with scale or natural placement or trying to set a scene – at best I’ve learned the basics, and now that I’m learning the basics again I suspect I didn’t actually learn them very well the first time.
It’s a strange sensation, sometimes, trying to extrapolate the things I’ve learned from 2d art into 3d art. When I’m doing a painting, I compose the view and then decide where the light is and, as I flesh the piece out, I try to remain true to those decisions. When I’m working in 3d, I have to position the entrance to the room and position the lighting to create the ‘composition’ the player actually encounters in the world, and the specifics of how that light gets rendered are handled by the 3d program. I’m still far short of where I want to be in this field, as well, but because it’s a newer skill to me I at least feel like I’m quickly improving.
The more I work at it, though, the more I notice there’s something missing. If I were to sit down in a drawing workshop and draw a model, then move my chair and draw them again, and do that several times, those drawings together would comprise something like a 3d interpretation of the model. Even if I did a very good job of those drawings, though, there would be discontinuities – part of the process of drawing would be to make decisions, exaggerations, corrections… there’s no such thing as perfect representation, because the lines we use to draw are largely conceptual, whereas the model is a person with a physical presence. Each drawing goes through its own artistic process, interpreting what I am able to see through my flawed eyes and converting it into a linear and shaded approximation. In 3d art, though, I just create the environment and leave that visual interpretation, what would be my drawing in this analogy, entirely up to the renderer. The gap, the thing that’s missing, is the 2d artist whose place is being taken by the computer – or, to be more precise, in this case the artist’s place is taken by Unity’s default rendering pipeline.
This isn’t to decry 3d art as in some way being less ‘real’ art, but to bring up the idea of the many kinds of art that it could be but currently is not. In our quest for consistency, for realism, we’ve left behind much of the power of 2d art – the ability to exaggerate, to portray the impossible and cartoonish. A good example of this would be the video game adaptations of cartoons like The Simpsons: In these shows, characters rarely face towards the camera because the style of the show usually only holds when they’re in 3/4 profile view. The particular cartoon squiggles that comprise the mouths and eyes only really make sense in that perspective, so whenever the script calls for a character to be seen from a different angle it looks off, weird and confusing and sometimes downright unrecognizable. The 3d game adaptations, however, require the characters to be viewable from every angle – so the style is collapsed into 3d models that makes approximate sense from every angle but also never really look like the cartoon original. Even the most elegant and well-executed cel-shaded outline shader can’t fix the issue that 3d rendering will, at the end of the day, be a faithful and uncreative depiction of the model data.
I know that you could affect the geometry of the model with a shader as well, but I haven’t seen it applied towards this problem – unsurprising since I expect I’m the only person who considers it a problem. Whether because of the constraints of technology or because of our worship of ‘realism’, the idea of making models that don’t appear the same from every angle doesn’t seem to have ever really taken hold. Every game establishes its own language, so in the end objects in a game can look like basically anything and the game continues to work. Just like playing make-believe as a child, a stick can be a sword if we agree it is a sword, a bush can be a dragon, the floor can be lava. If we can push things this far, why don’t we? I mean, we sometimes do, but usually only in the context of relatively low-tech ‘retro’ experiences, either using simple pixel art or low-poly 3d styles. We have yet to unleash most of the power on our disposal on the challenges of surrealism, impressionism, cubism – and, on the rare occasions when we begin to push in these directions, it’s usually only to try to emulate the most well-known 2d visual aspects of that style, rather than making any attempt to interpret how these might translate into a 3d space.
In general, the aesthetics of games fall into two R’s: Realistic and Retro. Recently there’s been a bit of leeway around ‘realism’, but it’s still the broad category AAA games fall into. While games like Dishonored and Breath of the Wild may not be attempting to appear real, they still try to emulate a version of reality, a world that is consistent in its rules and its appearance, a world where even if the particular appearance of a thing is stylized it still has the essential properties expected of a real object. Retro games, conversely, are willing to be weird, to be inconsistent, to be arbitrary and unreal – if they emulate the exact forms of inconsistency and arbitrary unreality that were the hallmarks of the nostalgic history of video games.
Any one of those quirks that are typical of retro games, though, could be harnessed now, and recontextualized into a modern space. The tendency for sprites to receive erroneous memory addresses and replace parts of an important game character with text or another character was caused by the specific implementation of pixel graphics used in old games, and has been used narratively in interesting ways by games like Undertale, but there’s no reason that must be married to a retro style – we could just as easily have a scene where parts of a 3d model flicker and are replaced with parts of another 3d model. It wouldn’t have the same contextual meaning as it does with sprites, where it comes to stand in for the concepts of corruption and elemental chaos – but that wouldn’t keep it from being visually interesting! Or, another artifact of retro games is pixelation – pixelation is still called back to in 3d contexts, sometimes, but only by creating low-resolution textures or occasionally creating 3d equivalents, voxels. However, that’s an interpretation rooted in a visual rather than a systemic understanding of what pixels were – what about an art style where all vertices of a 3d object snapped to a grid? What about an art style where models are rendered to a texture and then crunched down into a sprite and projected back into the world?
I bring up retro art, though, just because it’s the most understandable entry point into non-representation (or at least less-representation). Retro is the only area in game design where looking anything besides beautiful, anything besides representational, gets much allowance from potential players. We are prepared to accept retro because we know what it’s trying to be. We must be prepared to accept more, to accept the weird and ugly and inexplicable.
I understand why it has been necessary to attempt ‘realism’ for so long. Picasso drew a lot of normal very pretty paintings before he started painting blue shit and weird cube people. Eventually, we need to accept that we’ve got this reality thing down pretty well, and be willing to push outwards. We have the power to make worlds – why do we keep trying to just make this one, over and over and over again, with just slight thematic variations?