Game level design, to be precise; those who have been doing electronics and struggled with level-shifting between all the various standards are forgiven for thinking this was about something else.
That would probably be called amplifier design then, attenuating signals is a comparatively simple problem (although in RF anything that seems simple can become black magic wizardry).
Level design can still be an engaging hobby. In fact there's never been a better time to be a hobbyist level designer. Never before have amateurs had access to so much tools, education, and resources.
Except if you want to do level design for existing AAA FPS games other than Counter-Strike, the days where user maps and modding were allowed have come and gone long ago.
There's nothing out there like there was for Counter-Strike 1.6.
No worries, one promising use of LLMs is that we could describe a level someday with all its key features and just have it created fully formed, instantly!
Imagine a game mode where you as a player can just describe what you want while the LLM builds something fun and challenging! I think in simpler games such as Doom this should be possible already.
I keep seeing comments like this and can't help but feel like that would be like suggesting someone who makes pottery as a hobby could make it way faster by just ordering the vase they want from someone else; I imagine the fun part in this, as in many other hobbies, is making it, because if just want to play something someone else made, you can already do that, can't you? Maybe it won't be exactly what you wanted, but the LLM generated one won't also.
"we could", but how many people can actually describe a level with all its key features? It's not this easy. In your other sentence you mention that the LLM builds "something fun and challenging", but how would an LLM know what fun and challenging is? You, the "prompt engineer", would have to do that. An LLM can only replicate what it already knows, meaning that level design would be copies of existing level design instead of something creative and novel.
Sure, it can be done. But you can already generate levels and environments generatively. How is AI going to be better at it than a person that clicks around in the Unreal editor?
That’s just procedural level generation with extra steps. Why ask humans for input when you can just as easily ask the LLM to generate a list of key-features that would be received well by user_profile 400231862?
use of grammar and overly excited? (or are you trying to make a point? like look how bad LLM's are... hard to tell on the inet. or x2 are you trying to emulate an llm to see how many people you can catch?... its all so confusing now)
Who wrote this book? I don't see that information.
The Internet is an odd place. Imagine a printed book without an author listed (even 'anonymous', which is a rare edge case). Imagine a non-fiction book without a clear thesis, and one stated up front.
In addition, we are overwhelmed with information and options on the Internet in ways humans with only printed material couldn't imagine. We need these things - especially theses clearly stated up front - even more. I'm not willing to read things without it - I just have too much to read and too little time.
Robert Yang is the founder of this project. As an indie game developer, he is most well-known for his Radiator games about sexuality and intimacy. In the past, he also contributed levels to projects like Black Mesa Source, and conducted interviews with level designers for Rock Paper Shotgun. (https://debacle.us)
Contributors ...
Andrew Yoder has worked on Paladins and Warframe, and specializes in multiplayer social spaces and combat design.
I clicked into the Overview from the Book menu in the main nav. That only shows the sections in "Book 1, Process", not books 2-4 and the appendix.
Overview took me into the content, which has a sidebar; I scrolled to the bottom of that and found the link to the page. It took me a moment to realize it did show that info about the author/contributor.
There was a fantastic resource about Multiplayer Level design from David "DavidM" Munnich. Although done in Unreal Tournament context it was really informative.
I don't know if it's still available (could not find it), taking into consideration it was hosted at the defunct planetunreal I doubt it survived.
That's the one I was thinking of too, but also, how relevant is that still today? "Levels" in various categories of video games have evolved far beyond the fairly basic arenas of the UT / Quake era.
In terms of geometry density (detail) they have definitely improved by many orders of magnitude but in terms of layout I think most games have actually gone backwards with vastly simple levels to navigate. That (exagerated) meme about level design in 2010 is still fairly relevant [1].
I've been reading "Level Design: Concept, Theory, & Practice" by Kremers. It's a dead-tree book, and more of a theoretical treatise on all kinds of level design concepts, tropes and patterns rather than a hands-on guide, but might still be of interest to you.
Not a book, but Cave Story is interesting because the first areas in the game are too large and shapeless, but later areas are very well defined. I feel like the author went through some sort of level design progression while making the game.
Seems a lot of it applies to 2D levels too, I did some quick skimming and didn't see anything that jumps out as not being applicable for 2D levels. Anything in particular that seems very 3D focused? I know the images happen to be 3D, but most of it is basically the same for 2D.
AFAIK, de_dust (and de_dust2) were hastily thrown together by a young amateur (not in a bad way) map maker, not a professional and carefully designed + user tested level.
So maybe what we need more of, are designers and developers throwing shit together with less analytics behind their choices? It does seem like the more data-driven a decision is, the safer but also more "boring" it becomes.
Yes: there are no "average gamers". Everybody gets something different out of games, and the beauty of early HL1 online play is that there was something for everyone. Severs that were AWP-only playing sniper-focused maps; servers where AWPs were banned. Servers that only ever played de_dust; servers that only played crazy user-made maps. Team Fortress Classic and Counterstrike and HL1 Deathmatch and Science & Industry and so many more variants that I've forgotten playing. And what one player wants to play changes with their mood and the phases of the moon. As long as there were at least 9 other people interested in what you felt like messing around with in that instant, you could do it.
If you make a game based on average metrics, you get a single, average experience. If you make a game for the average player, you make a game for nobody. (Although you do seem to make a lot of money ...)
Look around for game jams, short competitions where you don't have to worry about perfection and can just dig into the process of making a game. After a few of those, you'll have all the base skills you need to tackle something bigger.
One thing I've never understood, as someone who only has a passing curiosity in game design, is how people learn from them? It always seemed to me that the whole idea was 'Here's a theme, make a game.' But if you don't know how to make a game, where does the knowledge come from? I get you can do independent study, but you could do that without a game jam.
Don't get me wrong, I understand the appeal of game jams, I just don't understand why I see it offered as a learning experience so frequently, as opposed to a practice experience.
In my life and career, I've found deadlines make the best mentors.
In addition to the other commenter's great point about making connections and getting exposure to different paradigms, with a tight deadline you learn how to quickly research and implement new things, and how to make quick creative decisions.
You learn when something is "good enough", and after a few jams you start to understand the holistic process of conceptualizing, actualizing, deploying and collecting feedback on games.
After a couple years, you'll find yourself more knowledgeable than the average graduate in game design, and will either discover what you'd like to specialize in, or realize you enjoy and want to be involved with the entire process.
I learned an enormous amount from in-person game jams by chatting with other game developers at the event, most of whom were more skilled than I was. I've also found game jams made it easier find collaborators, easier to get feedback from other designers, and easier to see how other people approach the same theme.
> But if you don't know how to make a game, where does the knowledge come from?
If you're at the stage where you feel like you don't know how to make a game, you might find value in doing the exercises from the book "Challenges for Game Designers" by Brathwaite & Schreiber.
I'm an ex game-dev, but I never did much level design. I've been thinking about it a lot recently and I'd love to see more theory on what actually makes a good level design.
Godot software is free and very easy to do the "blocking" method in this article, there are nodes called CSGBox3D which are resizable boxes with collision that you can drag, rotate, etc. and box out your world
reply