I’m still focused on blocking out the main park area and modeling areas of interest this week. My goal is to have a mostly complete explorable layout before going back to work on vignettes and side areas. 

Thinking with Nodes

I use a node-based approach for planning layout to keep things as simple as possible. First, I list the areas or points of interest I need in a layout then sort them in order of importance. I judge ‘importance’ on two qualities: The first involves areas which the story can’t survive without, the second is scale; in terms of time and how much space they’ll likely take up. Once I have a bunch of free-floating nodes on a map I  start to draw connections between them. The 'hero' areas have a sort of gravity that catches the less-important areas in their orbit (or like a big tree where other smaller plants grow it it's shade, if you'd prefer a different clumsy metaphor). In this way, a hierarchy starts to assert itself and an internal logic begins to grow.

As I draw connections I'm thinking about the ideal player path and narrative experience. I start considering obstructions along the main routes and any kind of gates. Some paths are only traversable from one direction, some might be left more vague to provide some challenge in wayfinding, and so on…

Other forms of on-paper planning that are more detailed than this aren’t helpful for me. In the past, I’ve gotten distracted making pretty maps, this format keeps me focused. Also, much of this initial work gets scrapped or significantly reworked several times over once I move into the editor. While I think it's important to have a plan before you start tinkering in 3D I also think it's important to jump to 3D as quickly as possible. I keep distances and proportions pretty loose in these diagrams. I won't have a good feel for spaces until I test them with the character controller. 

Tools

I’m using Unreal Engine 4 at the moment, but the best tool is probably the one you're most comfortable using. I'm a 3D artist first and foremost, so I find Unreal's suit of rendering and shader features to be the most convenient and flexible to get me what I want without much fuss. I do wish it had better level design tools for modeling in-editor, like Valve's most recent Source engine (I think UE5 has added some of that functionality but I haven't checked it out). However, the Maya to Unreal pipeline is pretty streamlined so I don't have a lot of issues doing quick iterations on geometry.

Here I'd like to recommend all of Mary Nate’s plugins (available on Itch.io or Fab), I'm not affiliated in any way, I just think they're great. Particularly I find the Prefab Tool really invaluable for environment art. Having an easy way to make and edit modular chunks of level geo or clusters of objects is really essential. It works a bit like Unity's prefab system or Packed-Level Instances in UE5. While Blueprints can technically do this, I find moving and duplicating geo inside Blueprints kind of clumsy and slow (and why isn’t there vertex snapping inside Blueprints?!).    

Being Wrong Faster

On this project, I've pushed models a lot further earlier in the process. I wouldn't do this on a project where I wasn't the sole level-creator, but I find low-detail ‘greybox’ layout can be difficult to parse spatially and not very fun to work with. Having a bit of extra detail and texture also inspires me with new ideas more organically. Even if those early decisions turn out to be ‘wrong’ they still get me moving in specific direction. 

Ultimately my goal is to reach a point of specificity where the environment can dictate it’s own terms. Or to put it another way, a logic emerges from the environment which rules out other options. I think this is where the practical concerns of level design and the conceptual work of narrative merge to form a whole more significant than their constituent parts. Internal logic need not be the logic of reality but rather the logic of whatever world I'm creating.