Monumental Failure utilises a low poly aesthetic, a look that has risen in popularity over the last few years. I wanted to take some time and collect my thoughts on employing this art style. It would be nice if I could organise this as a discreet list of pros and cons, but in the process of collecting my thoughts, I've realised that the work of employing the aesthetic is not broken down that simply. There's a lot of grey area, there are a lot of small decisions along the way that informed the final look. So, instead, I'll be writing about these decisions, and you'll infer the pros and cons from it. Sound good? Well, anyways, I'm probably a bit ahead of myself! First I should define what I mean by low-poly aesthetic.
Low poly can be literally understood as low polygon count. When I say low-poly, I mean models where the geometry is obvious. This is consistent through the world. We have hard edges. Light hits the models' faces at one angle only, creating big blocks of colour. No textures. When you see a rock, a brick, or a log, you know an artist spent the time modelling it and placing it. That's it, that's our starting point. With that in mind, let's get going!
The first thing I want to point out is definitely a pro, and was a huge boon in developing Monumental Failure in the time we did. Simply said, there are tonnes of freely available assets for low poly environments. A quick search through the asset store will reveal plenty of low poly models ready for you to drop in your game. If you find models that aren't low poly, It's pretty easy to edit objects to look low poly, otherwise the plug-in Polyworld has a utility to do it for you. Saving time on producing assets, especially generic stuff like rocks and trees is such a huge plus. The only stumbling point here is ensuring that these free models are cohesive with the rest of your aesthetic.
Speaking of sticking to aesthetic, one of the questions you'll have ask is what do low-poly effects look like? Things like fire, dust, smoke, or water, what are their low poly versions? A strict approach, i.e. producing something that has the obvious geometry like we described above, would have required a wealth of additional time and knowledge. Custom mesh based particle systems for fire, custom shaders for water, and custom cloud models for smoke, are a few examples of things I would have had to consider. Instead, for Monumental Failure, I decided that people generally wouldn't find a regular looking flame or smoke poof out of place, and trended towards "realistic" looking effects. For water, in the end, I think there are 3 unique takes on water in the game, and each one of those probably took a good 3 attempts before I settled on something I liked. It's not trivial!
On the theme of natural things, lets talk terrain. A feature of low poly artwork with respect to landscapes is there is typically more geometry in the foreground, less geometry in the background. As your eyes move from rocks to hills to mountains, the amount of polygons that describe each object decreases. The problem we run into, is of course that we're creating a video game, not a static image. We have a moving camera, meaning the area we want rendered in high detail, near the camera, will be changing.
The obvious solution is to have the parts of the world the player will be near be detailed, and keep backgrounds less detailed. This is probably feasible for something like a fist person game, where the spaces the player occupies will be relatively constrained. This is harder top do with a game where the environments are larger and less constrained, or when the camera's zoom frequently changes. Both these issues occurred within Monumental Failure.
I'm beating around the bush; there's a bigger problem. Fully modelling every environment vertex by vertex would be too much work! There's no way I have all the time to custom place an entire environments worth of ground. I need a tool like Unity's terrain engine to easily sculpt and paint terrain. Polyworld is that tool. Polyworld also has the advantage of providing flat surfaces with a low poly "texture", again saving us time modelling terrain.
For me, the Polyworld terrain was definitely an imperfect solution. It looks algorithmic, detracting from the "hand crafted" nature of the low-poly aesthetic. I was able to reverse some of this affect by manually placing rocks and buildings and other details where possible. Additionally, foliage helps distract from the repetitive nature of the Polyworld terrain.
These before and after screenshots highlight another aspect of achieving the low poly look. Your colour game has to be on point. The colours of the scenes your game renders out have to be cohesive, they should tie everything together. This starts with the colours and textures of all the models in the game. We then had the skyboxes and fog settings to play around with. The lighting was key to pulling this all together. Finally, we throw on some post process color correction to get all the previous work looking just a little bit better. With Monumental Failure, it would be nice to say we had this all figured out from the start, but truly it was an iterative process, adjusting and readjusting until we were happy, or at least complacent. 😛
The last thing I need to mention are the models our artist Jessica made. I can't specifically speak to all the challenges of modelling in a low poly style, though I'm pretty sure I can say that Jess never wants to model another brick again. I will however, speak to the aesthetic quality of the models themselves.
I think low poly really helps achieve a look of mystery and majesty. It's surprisingly good at communicating material property, those towers are made of stone, you know it! On the organic models it works great too. Trees and grass are trees and grass, no doubt about it. Our human characters too, even with their super low amount of detail convey huge amounts of personality. Jess knocked it out of the park!
That's pretty well the summary of how we got Monumental Failure to look the way it does. Perhaps you're wondering, would I employ this aesthetic in a future project? Short answer: maybe! I think in a game with less camera movement and more constrained environments, you would reap more reward from the aesthetic. The low poly look can save a lot of time and effort realising environments, producing a great looking game to boot. I also think it has potential to be combined with other aesthetics, maybe your environments are low poly but the characters are toon shaded. Ultimately, this experience will help me going forward to inform any other aesthetic choices, and maybe this devlog will help you too.
Thanks for reading,