What is with indie games and procedural generation?

You can talk about almost anything that you want to on this board.

Moderator: Moderators

User avatar
Gilbert
Posts: 450
Joined: Sun Dec 12, 2010 10:27 pm
Location: Hong Kong
Contact:

Re: What is with indie games and procedural generation?

Post by Gilbert » Thu Dec 17, 2020 10:29 pm

I think apart from Spelunky, Dwarf Fortress also had a huge inference on the popularity of procedural/randomly generated contents in Indie games.

While in many cases it's probably due to laziness in designing real contents and to prolong gameplay, sometimes it's very well implemented to bring life to a game. I remembered it worked quite well in Spelunky and probably great on DF too considering how popular it was (I hadn't really played it as this kind of games was too complicated for my tiny brain). Some of the better Rogue like games have wonderfully generated floors too. But of course these are the exceptions on top of much much more poorly implemented ones.

I suspect writing a good procedural level generator(possibly has something to do with a capital A and a capital I) would be more work than designing good levels manually, so it's definitely not laziness in such cases (though it IS in many other cases).

Oziphantom
Posts: 1069
Joined: Tue Feb 07, 2017 2:03 am

Re: What is with indie games and procedural generation?

Post by Oziphantom » Thu Dec 17, 2020 11:01 pm

when the levels are random, its comes down to stats and if you can "make it". So you can make an Algorithm to then test the levels. Thus you start to make a GA to tweak the numbers and then let you computer run for a couple of days making and testing levels to see if "passes".
Some game styles let you break it up in to pieces.
For example Diablo, the level layout just needs you to be able to make it from A to B, a simple places connected blocks and then run a maze search algorithm and you are good.
Then for monsters you know the level of the player, so you can the make some base line stats for monsters, then randomize them a bit and then place the in clusters. Since you know how rooms join you can then put a "DPS" per "tile" and then keep leveling it out until it reaches a good enough average for the players.
Then for items, you know what the currently have so you can then do a bell curve around it to offer better and worse stuff.
The 3 systems don't really need to interact
What separates the good from the bad, is Diablo's makers spend 4 years tweaking it to be "just so" while indie game devs spend 2 months tweaking it so "it mostly passes" ;)

But if you look at the source code for spelunky there is one here https://github.com/yancharkin/SpelunkyC ... er/scripts
It makes room types form a list, then uses prebuilt room layouts see scrRoomGen1~5 and scrLevelGen.

Post Reply