I developed/run a private server for the game that many members of the Diablo team made after Diablo II (Hellgate: London) and the same technique is used for random dungeon generation, but using 3D assets and 2d flat floormap shapes to make the combinations. In fact, it's even called the same thing, DRLG!
Sometimes the game tries a few thousand combinations when a player enters a dungeon before being satisfied by the results (there are configurable constraint sets).
The random dungeon generation was, for me, absolutely the top killer feature of the game and was the main thing that kept me coming back through so many replays.
For some reason, not a single other similar game I've played has had the same effect on me. Maybe there's bias going on (obviously I was younger then), but I've played a huge number of games since, and not a single one quite scratched the same itch. I will be saving this article.
They really didn't. I can't fully pinpoint why, but it had something to do with the combination of the level and all its contents being randomized a certain way, which from what I experienced of the sequels wasn't replicated enough for me.
I enjoyed the descent of Diablo 1 a bit more. First play through you’re descending into this place that seems to get darker and more dire the deeper you go. Diablo 2 was probably the best adventure of all of them, but Diablo 1 made up for a meh story with the dungeon design.
Funny but as a kid I devoured and adored D2. I played it many many times through. Later I decided to try D1 and I couldn’t get past how slow it felt, and in some ways primitive.
Diablo 2 was a bit more on the rails to me: if I play a specific amount of time, I can acquire the specific pieces of gear that I want for a specific build. Diablo 1 it felt like you were chasing pieces, but grand scheme of things, you just wanted the best in slot more often than not. Story was more there too and the quests appreciable. They didn’t phone in the sequel and really delivered a great game.
Diablo 1, however, I enjoyed stacking Intelligence as much as possible so that I could have pretty substantial levels of casting skills from books (since you had to have x amount of points in it for the next level book to work). In that way, it felt like you could have a God Character in a way that D2’s 20 max assignable (more could be gained as mods on gear) didn’t. You had to also work at that, which felt rewarding.
I don't think I've heard Diablo called a roguelike before. It's got the dungeon crawling and procedural generation, but the permanent death seems to be one of the most quintessential features of the genre. Diablo does not have that.
Diablo is basically Angband in real time, with commercial polish.
If you choose to define Roguelike to include permadeath or that it must be turn based, then it is not, of course. But if Diablo is not a roguelike, its next door neighbor is.
These days games like Diablo are often put in the genre "roguelite" indicating a roguelike but with some toned down or missing features (e.g. permadeath, turn-based, ascii...).
Yeah, fair. But that's an optional niche game mode and this piece is about Diablo 1.
Another comment linked the creator's postmortem; he said he was influenced by roguelikes but specifically called out that Diablo doesn't have permadeath as opposed to roguelikes. His original design doc included permadeath, but it didn't make it far into production.
I thought that was interesting as well. It seems like a fair comparison, and permanent death for a mass-market game in 1996 would have been a big non-starter.
Coincidentally, I played Diablo II with a self-imposed permanent death rule several years ago. I made it to the... Desert, I think it was? and then got killed and never really got back into the game. It just wasn't designed for that.
That’s the algorithm, sure. The tiles aren’t marching squares though, and calling them after how they were generated is confusing! If they produced rectangular or other non-square tiles would you still insist on not calling them Wang tiles?
I'm not sure what you are expecting. I'm describing the algorithm used, so yes, I used the terminology of the algorithm.
Wang tiles doesn't say anything specific about the arrangment of tiles as there are many tilesets that have the Wang Tile property.
And Diablo tiles aren't wang tiles, it would be inaccurate to describe them so. It's not to do with the shape, it's to do with connectivity. You can have a floor tile north-west of a wall tile, you can have dirt north-west of a wall, you can have floor north-west of floor, but you cannot have dirt north west of floor. Wang tiles do not permit restrictions like that.
> The tiles aren’t marching squares though, and calling them after how they were generated is confusing!
I don’t understand how this could be misunderstood by someone reading in good faith. I think that, if a reader remarks “this is confusing because…”, you shouldn’t tell them they’re wrong and it’s not confusing.
As a kid I never thought about it's level design, but retroactively I have huge respect for the developers (and to the author of the article for bring it to me).
I wonder if such things are taught in today's game design courses.
No. Kids going through game programs can barely program, except for a few outliers. The courses tend to be quite lightweight, and the programs tend to be very non-selective. They tend to be marketed to and attract poor students who couldn't get into / don't see the value in a real university.
Instructors may be aware of these techniques but would be hard-pressed to turn them into assignments that wouldn't fail all but a couple of people in the class.
Sometimes the game tries a few thousand combinations when a player enters a dungeon before being satisfied by the results (there are configurable constraint sets).