A Course Correction

I hate the level design. It sucks.

The level design in the game, as it currently exists, was a product of me trying to lay things out to a grid with no thought to anything but being able to have sensibly contiguous streets. I didn't even succeed at that.

So I've decided I'm going to completely redo the small-scale level design, and I thought this would be a good occasion to talk about my thought process with such things, and how I'm altering that thought process to make the next iteration of the level design massively better.

Part 1: The Macro Layout

For games with open world progression, I'm very fond of a macro-level layout that looks like the diagram of the four colour theorem:

I won't explain the theorem here because it isn't relevant to the design, but the layout consists of 4 areas, 3 surrounding a middle one. For Trafique, there are more than 3 outer areas, and the result is that not all are touching, but this is the seed of the design.

For Trafique, the Boulevard links all the areas except for Bastille, the central zone. This is both to further the diegesis of Bastille being an older development, as well as reinforcing the idea of it being separate from the rest of the city. As a further reinforcement of this concept, Maçon-Tannerie, the poorest zone, is also the only one with no direct connection to Bastille.

By design, the Boulevard is the only normal connection between the zones that it runs through; other than Bastille for its connected regions, but Bastille is unlocked only in the endgame and is thus not a viable shortcut for the majority of the game. 

Part 2: The Zone Layout

My goal for the redesign is for each zone to be a cohesive loop made up of smaller interconnected loops. I want the map to flow effortlessly, and for there to be as few dead ends as possible.

Each region's circuit naturally has its Boulevard ramp as one end, and should have something significant at the other "end".

  • For Maçon-Tannerie, the end point is M-T Taxi, where the game begins. Because leaving the zone is the culmination of the first act, it makes sense for the Boulevard ramp to be at the opposite end of the zone's flow from where the act begins.
  • For L'Aciérie, the end point is Chinatown.
  • For Les Collines, the end point is Bourbon Hill.
  • Gaufrier and the Grotto do not have strongly defined end points due to their layout goals, and Bastille does not have a single major entry point from which one could designate an opposite end, other than possibly Harris Tower.

Part 3: The Track Layout

This is where things get hairy. To get the Burnout Paradise meets Crazy Taxi feeling I'm going for, there's going to need to be a lot more stunt opportunities, and it's going to need to be easier to pull off stunts. Which means once again overhauling the car colliders and control systems, but it's ok, it won't be as major as the manual transmission overhaul was. It will be almost completely under the hood, no pun intended.

Overhauling the map, however, does allow me to correct my greatest mistake: not including a train.

Part 4: The Mistakes I Made in the First Two Designs

Being beholden as hard as I was to a square grid was a massive error. I'm even familiar with Robert Yang's Whiteboard Test, so I've got no excuse beyond feeling like it fit the low-poly aesthetic. It's mostly just lazy and boring. Bastille will still be the only zone with roads that truly curve, but that doesn't mean that I need to use nothing but 90 degree angles for everywhere else.

Neither design considered actual moment-to-moment driving whatsoever. Both considered the navigation aspect of the game, which is very crucial, but it still needs to be fun to drive to where you're navigating to. There need to be more shortcuts and they need to have some element of risk, of skill. I mean, I could go on at length about specific elements I want to consider, but the fact that the maps were jotted down without thinking of play and then converted into 3D is an unforgivable level design sin.

Beyond just being beholden to a grid, a huge problem was that the grid wasn't nearly granular enough. I was mostly counting on adding some of that with the final models, at least as far as shortcuts are concerned, but the uniform building and road sizes are so boring to drive in.

My plan is to bring down the size somewhat, and hugely increasing the granularity. The current map design exists on a 54x54 grid, where each grid square is 42m x 42m. My plan is to bring the "chunk" size down to 36, but to further subdivide the grid into three (or 4, which is why I picked 36). The new design will exist on a 192x192 grid where each square is 12m x 12m, and angles are permitted to exist. We're going from Wolfenstein 3D to Doom.

(Yes I am aware that 192 / 3 is 64, not 54. I am making the number 64 to offset the reduction in chunk size, and also because as a programmer I'm a sucker for powers of 2.)

Leave a comment

Log in with itch.io to leave a comment.