So, the Brackeys Game Jam 7 is done. That was fun!

In this post, I wanna go over the experience of making Deadbot, the things that worked, and what didn't.

First days

When the theme first got announced, I had no idea what to do. This is the third game jam I attempted after doing the previous two GMTK jams so I thought I'd have this early process figured out by now. But with the theme being "it is not real", I didn't have any real ideas to go with.

Some of my rough early ideas were

• Enemy isn't real
• Detective game where you figure out which person is real, like clones?
• murder mystery
• A "human apocalypse" within a robot world, there are humans among us
• Wander around and ask questions, find clues but some of them are fake
• Robots will sometimes lie to you
• Have a time limit with more kills happening as the game progresses
• Include a joke about a captcha but for robots (very funny)
• Gameplay to mainly consist of walking around and asking questions
• Alternative idea: like Her Story but some videos are fake
• You are a human disguised as a robot detective and you must blend in with other bots

A lot of these are a broad description of what Deadbot ended up becoming. I still wasn't sure how well this would fit the theme though. Sure, having one robot be "not real" could force the game to fit in, but it's not as good as having the gameplay itself tied to the theme. There's typically two types of games in these jams -- those that are intrinsically designed around the theme, and those which could have the thematic part easily removed. Deadbot, I think, fits into the second group. I spent the first two days not coding anything because I wanted to find an idea that is built around the theme. Eventually, I had to start working on something so I decided to just keep going with the murder mystery idea and try to tie it to the theme somehow.

The Plan

So I went for the murder mystery idea. 

The game was gonna be set in either a mansion with multiple floors and a small outside area, or in a small town. It would be filled with robots. At least one of these robots would be a killer and it would kill other bots as the game goes on. 

As the player, you could access a list of all bots with their number and rough visual appearance, you would be able to freely make notes about any of them. As the game progresses, you would gain access to more features, like security cams and shortcuts around the map. 

The main way you would figure out who the killer was, would be by asking questions. You could interrogate anyone, but each bot would be assigned a personality type and a "trustworthiness" score. These would be hidden to the player but they would impact how the bots responded. For example, a kind personality type with a high trust score would be very unlikely to lie and would answer in a helpful way. Bots with lower trust scores would be more likely to lie when questioned. Either way, when asked, they would stick to their story when asked again (unless something changes). You could any bot the same questions, for example asking them about who they think the killer is, have they seen a kill happen, where they were recently etc. The bots wouldn't be all-knowing, and when asked about the murderer, even an honest bot that wasn't around any kills wouldn't tell you. Similarly, even if a bot was near a murder, they may not be entirely sure who done it. It would be up to the player to gather info from the bots, freely make their notes in essentially an in-game text editor, and deduct who the killers are.

The killers would be smart, they would avoid killing anyone when the detective (the player) was in their line of sight. They would also avoid crowds and try to lure victims to remote areas of the map. When questions, killers would always answer in a way that benefits themselves, bypassing any trust score they may have had. So they would almost always lie. But just because you catch a liar, doesn't mean the liar is the killer.

The Actual Game

As is usually the case, the scope for the actual game had to shrink by a large amount. The building that it was set in has a single floor with a couple rooms. The outside area was scrapped entirely.

The gameplay had to be simplified by a lot. You can't ask questions or make notes. To balance for that, the killer also isn't smart so it can kill right in front of you.

In the end, the final game looks much different from what was planned. Was it still fun though? I wasn't sure, but those who tried it seemed to have fun with it.

During development, I ran into some issues. A lot of these are in my gameplay / game dev video for Deadbot, which you can watch here:

One interesting issue I had was to do with light maps. I never tried baking lights before in Unity, so I was probably doing it wrong. I needed my scene to have two light maps: one with the lights on, and one with the lights off. But it seems that Unity doesn't really like you having more than one lightmap and changing it at runtime. It even tries its best to delete the previous lightmap whenever you generate a new one. I found some complicated-looking community scripts that allow for lightmap switching but I was late in the development process by that point and I didn't want to spend much time figuring out something that may not work. In the end, I found a workaround that saves the lightmap when the game starts, and then generates a new empty one at runtime to switch to when the lights need to be off. This was very... unstable. When I tried it, it seemed to work, but later it didn't and created the abomination that you can see below. Deleting and remaking the lightmap seemed to fix it...

Probably the biggest feature that I missed out on was audio. If you tried out the game during the jam you might have heard... nothing. I wanted to add some background music (I figured some jazz might fit a detective game) but as is the case with many things, I ran out of time. In fact, I was working on this game up to the final 10 minutes before the deadline. One reason for this was burnout. I spent the first few days brainstorming and just messing around with spawning AI-controlled bots that randomly wander around. But I really couldn't figure out how to tie it all together. So the days in the middle of the week were spent doing mostly nothing productive. Nearing the end of the week, I actually began implementing the core gameplay elements, like the cameras and the actual player -- things that I'd normally make first. I worked all night and day (with a couple naps) for the final two days of the jam and I really had no time left by the end of it. Things that seem simple enough to just squeeze in, like a basic background music loop, just weren't priorities compared to systems that allow you to start and finish the game.

Some other systems that I intended to implement in the revised game plan but didn't have time for:

• Main menu
• Pause system / confirm close (ESC instantly closing the game is not ideal)
• Camera disruption event -- the cams temporarily breaking, similarly to lockdown and lights out

The future of the game

Will I continue working on this? Maybe. Maybe not.

I like the original plan and it's something that I think would be fun to play but I'm not sure if I'll be motivated enough to make it. There is a sense of urgency wanting to finish a project for a game jam, which goes away the second that the jam ends. Even implementing simple things like audio to the original demo feels like a chore now that I don't have to do it.

Regardless of how good the plan is, there will always be problems and design decisions to run into along the way. Sometimes I spend weeks on an idea to realise that it's not really all that fun, or that I don't know how to make it into a full game. Just see my grapple hook game demo.

I don't want to say that Deadbot is dead but it's not really alive right now. Who knows, maybe I'll pick this project up again in the future. For now though, it remains as is and I'm probably taking a break from game dev...

Until the next jam

:^)