Low-poly strategy

How we can use low-poly resources as high-res greyboxes for the first 6-12 months of development.

I think low-poly is a very natural starting point for a game that wants to be as modding-friendly as possible. If the average asset creator needs to create this kind of HD scene:

..then there’s gonna be a lot of people who tap out right away, because they can tell that it’ll take them too darn long to be making something of that detail.

Having to make something like this is much less daunting:

The simplicity of low-poly also makes it a more powerful target for novel rendering techniques. There are loads of smart ways to give low-poly a distinct, fresh look at low cost:

A short hike (video) Untitled Goose Game Bad North

Early-stage marketing

By far my strongest argument for investing early in low-poly assets is due to the marketing potential it brings with it. This is directly related to the "openness strategies for a moddable game" I lay out in #2.

Low-poly assets are a great way to go from boring, uninspiring grey-boxing to something that tickles your imagination. See for example the progress updates of Antorum, a Rust + Unity game.

As a modding enabler

See the success of Veloren for an exceptionally good example of this. They've significantly lowered the bar for art contributions by constraining themselves to the voxel-based "block-style". Grokking a tool like Goxel is 10x easier easier than something like Blender. It also makes it easier for anyone to make something that's decent-looking, because you only have so many pixel-blocks to work with and there's a limited amount of ways to paint a 64x64 pixel bear 🐻

Dust3D specializes in low-poly modeling and greatly simplifies character rigging & animation. That said, Blender is also great for low-poly, and it is easier to learn when confined to low-poly.

Low-poly as a guiding constraint

Low-poly affords us less smoke and mirrors. The action (combat) can’t just be made to feel dynamic and vibrant, like you can get away with on higher-def with fancy animations. In low-poly the action needs to truly be dynamic and vibrant in order to spark the observers' imagination. As Bad North shows, it doesn’t have to be anything outrageously complicated. It's just a combination of tiny things that can be adequately simulated with simple randomness rolls.

• Pushback
• Actual blocks;
• impacts that really hurt
• Basic swarm behaviors (boids)

Bad North is the most combat-fun I’ve had in a long time, while still being decidedly chill.

Procedural generation

We want to make a lot of game with a limited amount of resources. Procgen isn't something we should in any way be reliant on for our game to be filled with good-looking assets, but some cheap experiments with procgen can result in massive cost savings if we discover some legitimate uses for it.

Procgen is a lot easier to do for low-poly resources. For some assets, like rocks and trees, there's hardly any quality difference between generated and hand-crafted assets. And the generation-step can just as well act as a precursor to the final touches done by hand. As such, procgen is also an amplifier for asset modders.

Procgen is also closely related to modding; it's DYI art assets for programmers.

This is the same thing Embark is doing btw, except they’re trying to do it for AAA games, which takes 10x-100x more resources than what we’ll be attempting.