One More Night 2.0: Why We Tore It All Down
Six days ago, One More Night was an arcade zombie survivor. You dropped into an arena, mowed down waves of undead, leveled up, and tried not to die. It worked. People played it. The numbers were fine.
Today, it’s a completely different game — and I think it’s the one it was always supposed to be.
This isn’t a post about a patch or a content update. This is the story of tearing down something functional and rebuilding it from the foundation, in under a week, because “fine” wasn’t good enough.
The Problem with “Fine”
The Vampire Survivors formula is great, but it’s saturated. Every week another auto-attacking wave survival game shows up on Steam. We could polish forever and still just be “another one of those.”
More importantly, the arcade survivor loop didn’t match the fantasy of the game’s name. One More Night — that phrase implies dread, tension, scraping by until dawn. Not power fantasies and screen-filling explosions. The mechanics and the identity were pulling in opposite directions.
So on March 17th, I stopped patching and started dismantling.
Week of Demolition
First to go was the Phaser-based main menu. Replaced it with DOM — glass panel style, EchoForge accounts integration, proper polish. Sounds small, but it set the tone: this game was going to feel crafted, not hacked together.
Day two, I gutted the balance. Old system was for ten-minute arcade runs. New one needed tension over longer sessions. Complete difficulty rebalance, ultimates that feel desperate rather than overpowered. Also built a tile metadata system for walls, doors, furniture — the new game needs environments that matter, not empty arenas.
This is also when I committed to the dual release model: a premium Steam version and a free-to-play arcade version. Two different audiences, two different monetization approaches, same core game.
Building the Nervous System
Day three: make the undead dangerous again. I wrote a full DebuffManager — stun effects, control inversion, disorientation — so that getting hit by a zombie actually disrupts your plans instead of just ticking down a health bar. I replaced the programmer art with real pixel art sprites and built boss sprite atlases. The game started looking and feeling like a different thing entirely.
Day four was the big one. Old pathfinding was a hack — zombies got stuck on walls and a nudge system shoved them free. Replaced it with flow field pathfinding. Every zombie knows the best path to you at all times. But the real change: zombies now try doors before breaking walls. They don’t just path toward you. They look for a way in.
That one AI change transformed the whole game. Barricading matters. You’re listening for doors rattling and thinking about escape routes. The Vampire Survivors version never had a moment like that.
Under the hood, I extracted systems from what had become a monolith. WeaponRegistry, GameStateMachine, VariantRegistry, typed events, standalone DeathEffectsSystem. Doesn’t show up in screenshots but makes everything after it possible.
Bosses That Hunt
Day five was boss day. Redesigned the Brute with a chain charge — it picks a direction and commits, smashing through anything in its path. Added the Zombie Dog, fast and unpredictable and terrifying in tight spaces. Built a debug panel and burned through fifty-plus bug fixes in one session.
Fifty bugs sounds like a lot. When you’re rebuilding this aggressively, things break constantly. The debug panel wasn’t a luxury.
The 2.0 Pivot
Day six. Inventory system and shop UI, because survival horror needs resource management, not XP bars. Sealed crates as discoverable loot. Enemy drops so every fight has stakes beyond staying alive.
Then I built the flashlight.
The flashlight and stealth system is the heart of 2.0. Your light cone affects zombie detection range — shine it toward them and they’ll spot you from further away. Turn it off and you’re harder to detect, but you can’t see what’s coming. It’s a constant tension between information and exposure.
Then I added a scent trail system. You leave a chemical trail as you move and zombies smell it. Stay too long, they converge. Move too fast, fresh trail. There’s a deodorant item that suppresses your scent — silly name, real tactical value when you’re sneaking past a horde.
Switched to click-to-fire aiming because deliberate combat builds tension in a way passive damage can’t. Shrunk the maps and buildings — horror works better in tight spaces. And laid out a 10-location campaign, progression through distinct environments instead of one endless map.
From “Survive Waves” to “Survive the Night”
The old One More Night asked: how long can you last? The new version asks: can you make it to morning?
Different question. First one is endurance. Second one is decisions. Do you barricade and wait, or push through to the next building where there might be supplies? Do you use your flashlight to check that room, knowing the light might attract attention? Do you burn your deodorant now or save it for the gauntlet section you know is coming?
Flow fields, scent trails, the flashlight, door-seeking AI, the inventory, smaller maps — they all serve the same idea: every moment is a choice, and every choice matters.
What’s Next
The 10-location campaign needs content. The shop economy needs balancing. The stealth system needs edge-case testing — right now there are probably a dozen ways to cheese the scent mechanic that I haven’t found yet. And I need to build out the dual release properly so the arcade version feels complete on its own while the Steam version offers the deeper experience.
But the foundation is solid. In six days, One More Night went from a game I was maintaining to a game I’m excited about. Sometimes the best thing you can do for a project is admit that “fine” isn’t the same as “right,” tear it down, and build something worth staying up for.
— Bruno