itch.io

Last sessions on BAD FUEL have been primarily for gas station and road segment layout. Even though the lighting means a lot of stuff is discrete or partially hidden, I think building a horror game really is like building a killer haunted house: if you cheap out on the subtle details, you lose something real even if the players can't fully articulate it.

So even though it's taking a while, I'm taking my time on these. I'm using a tool (Farclip) that helps with this, things like book keeping of layouts and analyzing how it flows. It's not really an AI in the sense of generative AI, as you primarily ask it questions or perform analysis but then you have to implement any changes -- but disclosing here just to keep it honest (since their website technically ends in *.ai lol). Whatever it is, this is the 3rd or 4th game I've made using it and it's awesome for stuff like this.

But regardless, TONS of work still to do!

Tested a bunch of approaches and landed on raycasting from the player. I only care if the player is lit right now, so light maps are overkill. Pixel sampling does this, but is too expensive for browser builds.

With some help from Claude (Fable), I built cheap raycasts to nearby lights. As my mum used to say, "if you can't see me, I can't see you." It gives softer falloff and silhouettes instead of hard cutoffs so corners, etc. feel right and also respects things like crouching.

The test level is working well, with WebGL holding 120fps.  But to really get that, I'm going to need some real levels, with real geometry. So next up is firing up Farclip to help me build some with decent sightlines for a stealth game.

Starting with basic functional block out, and as time permits adding refined visuals. But in horror, so much is driven by lighting and ambience rather than asset models so I want to see how far this will take me. Starting with a simple truck, interior cab and wired up fuel gauge and basic interactions.

I'll be using a treadmill approach, so the truck itself stays still and blocks of road cycle under you. That allows me to easily make the road infinitely long. My current thinking is to procedurally place things along the road based on various condition triggers.

I'll need multiple gas stations, so I'm thinking of ways to not need to make a bunch of unique layouts but still keep each stop interesting. Keeping pacing balanced in a smallish layout without feeling repetitive is always a challenge (at least for me).

Planning on keeping a devlog, in part because this is a particularly fun/unique jam (at least for me). So here goes.

Lighting is a big issue, especially with performance in the browser. Killer stealth experiences don't just need awesome level layouts (an issue I'll have to tackle at some point), but the lighting to match it so you can read what a good hiding spot is. This is particularly true where the game is focused on light bulbs, you make your own shadows and recharge in the light.

Sampling pixels off the GPU just isn't going to be an option, so I'm working on some ideas to keep cool visuals but not rely directly on shadows for visibility stats. Thief pulled this off on old hardware, so I'm starting by exploring how they did it. The challenge is it scales by light and by shadowcasting renderers, so clever light placement alone doesn't necessarily fix it if I want some engaging choices of what to extinguish. Raycasting? Swapping light maps? The latter is definitely going to hit some memory constraints unless I bake my own with a different representation than Unity's default.

Still thinking it through, but to make it quick to iterate I'm starting with a calibration hall and a stress field that will let me rapidly test both how well the visuals match the visibility meter and the performance in a WebGL build.

Your just a simple person, trying to drive your old truck home on Route 16 on a suspiciously dark afternoon. You know you'll need to stop for gas, so sure hope the stations are open and stocked!

First step for me is always a simple click-through so testing is controlled (and serves as a start screen when done) and a panel full of debug telemetry. Since I'm shipping in WebGL it also lets me monitor performance drops in editor and initial builds. Controls at the bottom are currently just a check list of what I need to build, may change.

Speaking of -- not making the mistake I made in the last jam and building today so I can kick any setting bugs.

Thanks! Sorry about the volume. The turning degrades with speed since I wanted path choosing a little more engaging than just the raw layout, but I agree it needs tuning to get that right balance. I'm giving your game a try now!

Cover image and page written. QA passed (and it is beatable!). Wrapped up and submitted, just in time.

I'm new to game jams, and looking at other submissions, I'm really impressed with how many people produced something that looks cool in such a short time. Looking forward to playing them, and would love to hear any feedback on Scaredy Cat!

https://pixeldr.itch.io/scaredy-cat

Biscuit looks horrid in pink...

This is the dreaded build issue where everything looks good in editor. My own fault, really wish I would have tested my build settings in the first few hours instead of waiting.... nice to have a panic at the end like this.

Mechanics are straightforward, which is great for such a short jam, just double jump, run, pick-up items to make you afraid and falling panels. So those didn’t take long to implement. Since the puzzle aspect is basically route-planning, you have to find the various ways to chain your fear pickups to get enough speed. So the hard part is balancing the levels so they are both solvable and scaling in difficulty.

Like this level in the screenshot took me over an hour to figure out a way that layout the vacuum and other stuff so it looked/felt good but for the longest time was just unsolvable.

I’ve been trying to use Farclip, so far it’s been a help. But balancing this type of thing is something I don’t feel like I have a good concept of how to approach. Any one have recommendations of methods they use or even books to read? So close but feeling so far off with only a few hours left.

Cross-posting with Discord since I'm a Discord n00b.

I'm building BAD FUEL, a first-person horror game about driving home at night and having to stop for gas. This theme was perfect for me and the game inspired by my years working at a gas station...

The worst part of any stop is the pump. You pay, you take the nozzle, and then you just stand there while it fills. Maybe run in for some junk food or to use a questionable bathroom. But at Perpetual Fuel stations, the pump is loud, and loud is what brings them.

See if you can stay alive, get home and maybe even figure out what's happening with the gas stations in this part of the county.

I plan to keep a devlog here as I build, and hope others will share some feedback and suggestions!

No time for fancy assets, so Biscuit is built from primitives with basic procedural animation for legs and tail. Also loves calming down by sitting on your laptop mid game.

• Loop Ideation/Story: Realmweaver
• Engine: Unity WebGL
• Coding: Claude Code(Fable for planning, Sonnet 5 for execution)
• Art/Assets: Blocking with stupid simple models built from primitives by Sonnet 5; considering using Meshy depending on rigging (but would love recomendations)
• Level creation & QA: Farclip
• Music: Hadn't decided, but based on other comments here I think I'll give Suno a try.

Thanks! Looking forward to the feedback. Good stealth feel, even not in the browser, is tough so we'll see if I bit off more than I can chew.

I'm building Scaredy Cat, a 3D puzzle platformer where you play as a cat trying to get out of a house that's coming apart. You have no attacks and for the prereq: the only thing that breaks anything is how fast you're going. And as cat owners know, the thing that makes you fastest is being scared. So up your fear by interacting with super scary objects to be able to hit the door hard enough and escape to the next level.

Thanks Neonfire for the discord tip, I'll be posting the rest there. 26 hours to go!

I'm doing a thief-inspired stealth game, set in the 1893 World's Fair. After midnight, two hundred thousand electric lamps and all of them dark. You designed the thing that lit it and an evil businessman has your patents, so you're breaking into your own exhibit to take three things back.

The bit I'm actually building the game around: you carry a coil that lets you turn off lights at some distance. It drains constantly, and it only refills off a live circuit. So darkness is safe but every stretch of dark you buy has to be paid for by standing in a lit room for some amount of time. A full coil also hums loud enough for a guard to hear, which means you're at your most audible exactly when you're best equipped. Since the theme is CHARGE! the last thing you steal is the master key, and turning it lights the entire fair. The game ends with ninety seconds of running across a fairground with no shadow left in it, charging to the exit!

Tools: Unity WebGL, Claude Code + custom harness, Realmweaver for ideation, Farclip for levels & QA
Haven't settled on asset or music/SFX generator just yet, but plan to use one.

Oh cool, thanks for the heads up!

I know it's short, but that's part of what makes it interesting to me. I've only done a few jams and the last one other kept devlogs here in the community section. Not sure how common that is, but enjoyed watching others build like at in-person hackathons.

Eexicted to kick off!