Case Study — Level Design · Game Design
Starfall Crater
I designed and built Starfall Crater, a Minecraft dungeon centered on exploration, risk/reward, and player choice. The level descends through three increasingly difficult tiers, with multiple ways for players to approach or bypass encounters.
The project went from paper map → whitebox → playtest → final design review. The most useful result was something I didn't plan: a player found a way to bypass the combat entirely, and I chose to keep it.
The challenge. Design a level with a clear progression and defend the design decisions through documentation and playtesting.
Project
The course brief was to design a level, document it, build it, and defend it. I pitched Starfall Crater — a proposal for a randomly generated desert structure for players who want to explore: a meteor has crashed, and an alien biome has taken over (inspired, I have to admit, by the movie Evolution). From the ground you see the crater's striations running toward a hole in the ground. Below it, three tiers descend — each one darker, tighter, and harder than the last — and at the bottom, a treasure chest.
The full arc ran from pre-production to final write-up between December 3 and December 22, 2024: Milestone 1 delivered a milestone schedule, a Level Design Document with asset and feature lists, and a paper map. Milestone 2 whiteboxed the level and ran the first outside playtest. The final write-up reviewed the build against ten formal level design principles.
Design Goals
Six goals shaped the level, written before blocks were placed and defended against the finished build:
- High risk, high reward. A structure for explorers that punishes recklessness and pays out curiosity.
- A real challenge curve. At least three tiers that escalate visibly — darker, tighter, and tougher the deeper you go.
- Reliable encounters. Use planned encounters alongside randomized spawners so the intended challenge still happens.
- Nonlinear play. Multiple valid approaches — dig around it, cut through the middle, fall through a funnel, or go in as a guild.
- Deliberate disorientation. Moments that make a player lose their bearings, then re-orient them with risk/reward.
- Worldbuilding through space. A strange place the player reads through blocks, light, and sound — not text.
Level Design
The structure was planned on a paper map before a single block was placed: concentric terrain rings, hostile mobs and toxic spores noted per level, a critical path, and tunnels that drop you from the ground floor into the underground area. The strip-down of that map — the whitebox — became the playable level in Milestone 2.
From the surface, the crater reads as a question: striations of meteor-crash debris point at a hole, and diamonds glint from the dark below. The first tier is familiar (cave spiders, skeletons). The second drops you into evokers and husks as the ceilings tighten. The third is where it gets mean — pillagers, witches, and breezes, dense enough that a player thinks before they commit.
Two principles got special attention. Pathing intentionally disorients: fall-through tunnels can drop a player 40 blocks — a room is usually two to four blocks high — dropping someone straight from "mining along on the surface" into the thick of it. And verticality was the honest weakness: the tiers are clearly separated, but each level stays fairly even, with tunnels around rather than a real third dimension. I flagged it in the review rather than hiding it.
The alien theme did the immersion work: blocks shift from sand, dirt, and gravel into strange red mushroom-like growth, veins of diamond ore appear, and bioluminescent flora and pulsing vines mark the deeper tiers.
Key Design Decisions
1 · Difficulty by descent, not by gates
Each tier escalates — cave spiders and skeletons, then evokers and husks, then pillagers, then witches and breezes — with tighter, darker rooms as you descend.
Why. Let exploration carry the difficulty ramp. The tension should come from going deeper, not from a locked door.
Result. The difficulty increased as players moved deeper into the crater.
2 · Hand-placed mobs as the safety net under spawners
Minecraft spawners boot on time-in-area, so a player who drops straight down a funnel could land in a room that hasn't spawned anything yet. I placed key monsters by hand so encounters always fire.
Why. Difficulty must be reliable. Randomized spawners add variety; planned placement guarantees the beat.
Result. Hand-placed enemies made the intended encounters more reliable while spawners added variation.
3 · Nonlinearity on purpose — and emergent paths accepted
Dig around the whole crash, cut through the middle, fall through a funnel, or go in as a guild. The meteor site doesn't wall itself off; the sandbox stays an open world.
Why. Player agency is the point of Minecraft. A level that ignores the pickaxe is fighting the game.
Result. A playtester bypassed a tier by digging straight to the bottom — and I kept the route.
4 · The 40-block drop
Fall-through tunnels can send a player from the surface 40 blocks down, creating a sudden transition from exploration to danger.
Why. A spike of genuine tension creates a moment the player retells; disorienting them briefly is the setup for the re-orientation behind it.
Result. The drop created a sharp transition from open exploration to immediate danger.
5 · Risk made visible before the commit
Diamonds glint from the ground and through the descent; deeper tiers hold veins of diamond ore, and the bottom hides a treasure chest with real loot.
Why. A risk/reward trail needs an anchor the player can see before they take the risk — the reward says "yes" before the mobs say "no."
Result. The visible diamonds gave players a reason to take the risk of descending.
Playtesting
The first outside playtest ran on the Milestone 2 whitebox. What happened: the striations → hole → descent loop worked, but the tester died — there were simply too many monsters in the lower tiers. What I changed: I cut the mob density and set the next milestone to reward tuning.
The most useful lesson came during the final write-up. What happened: a playtester skipped the fights, dug a controlled shaft around the holes, and reached the bottom at a pace that felt safe. What I changed: nothing. Rather than reading this as a broken level, I kept the route — the design allowed it, and the level was better for it.
Technical & Implementation Notes
- Paper map and critical path first. Mobs and toxic spores marked per level before building began; the whitebox followed the map.
- Documentation discipline. A Level Design Document with asset and feature lists, and a milestone schedule updated across all three milestones.
- Minecraft systems as design material. Spawner timers (time-in-area), lighting, and survival-mode constraints shaped room design; hand placement compensated for spawner boot time.
- Custom alien mob. I modeled an alien creature in Blockbench, but the resource-pack setup needed to load it into the game remained unfinished.
What I Learned
- Emergent play beats choreography. The best moment in the level — the dug shaft — wasn't in my design. Leaving room for something I hadn't planned made the work stronger.
- Balance from playtest outcomes, not intentions. The difficulty was right on paper and fatal in play; the fixes came from the test, not from the plan.
- Name blockers honestly. The unfinished resource-pack pipeline went on the record as a blocker, alongside everything that worked.
- Named principles keep a review honest. Checking the build against ten level design principles surfaced verticality as my real gap — one I carry into the next level.
Video
The full arc in three videos — proposal, whitebox + playtest, and the final level design write-up.
Note. The original Minecraft world file is no longer available, so this case study documents the project through the surviving proposal, playtest, and final-review videos above.