Same board, different elements...
Essentially, I''m asking about boards which remain geometrically the same, but have elements that are positioned differently...
The game Phantasy Star Online for the Dreamcast seems to have a grid system for each of its lush, premade levels, whereby different elements like switches and obstacles are positioned differently on the grid each time you enter. You may enter the same room a dozen times and find force-field pillars, rocks, light-switches, and door controls all in different places. In the case of puzzles, you might find the same puzzle key (switch-lock combo) in different rooms as well.
I was weighing using the same idea and expanding the number of elements. The levels / boards would be premade pieces (like rooms and corridors) that are meant to be joined somewhat randomly. Internally, each board would have a 3-dimensional grid detailing where elemetns like tables, equipment stalks, ladders, beams, catwalks, and partitions could be placed. Even though players might travel to a wide variety of different environments (which the game would call by different names), there''d only be a set number of templates (warehouse, ruins, apartment room, generator room, etc.)
Potential downside: Eye strain and boredom due to repetition. I think the only way to evade this would be to make the levels (or rooms) visually appealing. This means a lot of effort put into animated textures and ambient animation. Levels would still be repetitive, but at least they''d be nice to look at.
Potential upside: We game designers without million dollar budgets might be able to create expansive game worlds and have plenty of gameplay for each board (provided we come up with the right algorithm for element generation...)
Thoughts?
--------------------
Just waiting for the mothership...
--------------------Just waiting for the mothership...
I think that''s a great idea. I think if used for the right roms/areas, it would be just as effective and much easier than building every area manually. However, I think that there should still be the ability to create fixed rooms, so that the key areas could be designed. To counter the obvious difference between that areas that were random and the areas that were designed, the majority of the room could be fixed but some of the less important objects could still be random, like plants, style of furniture, rock formations etc. It would give a random feeling but still give the designer to create key events.
One issue would be storing them for saved games. Having to store all the rooms/areas could be huge. Maybe a random seed could be stored and then when that room/area was needed, it could be recreated exactly as before.
tj963
One issue would be storing them for saved games. Having to store all the rooms/areas could be huge. Maybe a random seed could be stored and then when that room/area was needed, it could be recreated exactly as before.
tj963
tj963
Procedurally generated content is the way of the future. There just aren''t enough level designers to churn out the amount of content needed by today''s multi-player games. I''m surprised more places aren''t doing this, really. Diablo did it, and the Rogue-like games it was based on did it. Daggerfall did it, and I expect Morrowind does too. All these games take prefabricated rooms and corridors and join them together to form dungeons or whatever. Daggerfall generated whole towns this way, I think.
Essentially, it''s not a difficult task at all. The main problem, in my eyes, is that it takes a programmer and a designer (or a joint programmer-designer ) to work together on the system so that it creates satisfying worlds/levels/dungeons while being technically feasible. This is in contrast to the traditional system where the programmer can throw out a world-builder tool and leave the designer to get on with it.
I will be following this approach when I start on my RPG, if only to generate a large world upon which to add plot-specific detail later.
[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions | Organising code files | My stuff ]
Essentially, it''s not a difficult task at all. The main problem, in my eyes, is that it takes a programmer and a designer (or a joint programmer-designer ) to work together on the system so that it creates satisfying worlds/levels/dungeons while being technically feasible. This is in contrast to the traditional system where the programmer can throw out a world-builder tool and leave the designer to get on with it.
I will be following this approach when I start on my RPG, if only to generate a large world upon which to add plot-specific detail later.
[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions | Organising code files | My stuff ]
August 11, 2002 08:57 PM
(Off topic)
In Morrowind all the maps (both exterior and interior) are created by hand (which is quite impressive considering how large the world is). They use a similar principle though, in that prefabricated rooms/corridors/etc are combined to form larger structures, only it''s not automated.
quote: Original post by Kylotan
[...]
Daggerfall did it, and I expect Morrowind does too. All these games take prefabricated rooms and corridors and join them together to form dungeons or whatever.
[...]
In Morrowind all the maps (both exterior and interior) are created by hand (which is quite impressive considering how large the world is). They use a similar principle though, in that prefabricated rooms/corridors/etc are combined to form larger structures, only it''s not automated.
I have to mention board games !!!
If you dont know them, you might wanna give a look at Space Hulk or Advanced HeroQuest (the non Advanced version was published by MB but had a fixed board. With the MB version of Space Hulk which was called Space Crusade, they did 4 separate board that you could position in a few different ways, adding a bit of variety).
The idea is simple : create a few corridor, branching, corners, rooms and pick and mix. It''s damn easy to understand isnt it ?
Add the odd "unique" room, themes for the levels so that you dont get too odd results, possibly link monsters to some types of environment (to avoid the polar bear in the jungle), and possibly room (the bears in a cave, not in a little house...)
I think it''s simple, elegant, and actually allows for a lot of variety as can be seen in Roguelike games (which have much less variety, as a matter of fact).
My own interest is in the binding of events/monsters/items with the actual topography of the places created.
That is, dont generate a maze, then put random monsters, then items.
But bind each step with the others, pick an environment, pick a possible type of location/building that can happen in this environment, in those sublocations, put monsters/items/furnitures/etc that could/would be there, some elements increasing the probability for others to happen.
For instance, if a king was put in a place, even if it originally had a very limited probability, the fact that it''s there would tip the global balance to add elements that would normally be found with the king around. e.g. given that the king is there, the probability for the royal guard to be also present becomes 99.9% instead of the normal 2% if it''s in a castle, and 0.1% in wilderness, you get the idea, I hope.
So Wavy, tell me : what''s the problem with that ?
Sancte Isidore ora pro nobis !
If you dont know them, you might wanna give a look at Space Hulk or Advanced HeroQuest (the non Advanced version was published by MB but had a fixed board. With the MB version of Space Hulk which was called Space Crusade, they did 4 separate board that you could position in a few different ways, adding a bit of variety).
The idea is simple : create a few corridor, branching, corners, rooms and pick and mix. It''s damn easy to understand isnt it ?
Add the odd "unique" room, themes for the levels so that you dont get too odd results, possibly link monsters to some types of environment (to avoid the polar bear in the jungle), and possibly room (the bears in a cave, not in a little house...)
I think it''s simple, elegant, and actually allows for a lot of variety as can be seen in Roguelike games (which have much less variety, as a matter of fact).
My own interest is in the binding of events/monsters/items with the actual topography of the places created.
That is, dont generate a maze, then put random monsters, then items.
But bind each step with the others, pick an environment, pick a possible type of location/building that can happen in this environment, in those sublocations, put monsters/items/furnitures/etc that could/would be there, some elements increasing the probability for others to happen.
For instance, if a king was put in a place, even if it originally had a very limited probability, the fact that it''s there would tip the global balance to add elements that would normally be found with the king around. e.g. given that the king is there, the probability for the royal guard to be also present becomes 99.9% instead of the normal 2% if it''s in a castle, and 0.1% in wilderness, you get the idea, I hope.
So Wavy, tell me : what''s the problem with that ?
Sancte Isidore ora pro nobis !
-----------------------------Sancte Isidore ora pro nobis !
I think I would choose a few unique level designs over a ton of repetitive ones.
Whereas I would not. A procedurally-generated layout can be just as interesting as a hand-made one.
[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions | Organising code files | My stuff ]
[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions | Organising code files | My stuff ]
Don''t quote me on this, but I believe that Halo used the method your describing. Something like click-together level objects, there was alot of repatition in Halo.
Kylotan & Zegfried : well, at least we have people on both side of the fence.
To be honest, I *do* mind repetition, but whereas in a randomly generated environment I will expect it, hence not be as shocked, I''ll be sorely dissapointed in a game supposedly *handmade* where all towns are really the same big things with different textures.
Random elements are useful if you want to add variety without having to wreck your head. Think of a similar problem : you want to make a Starcraft (or any other RTS) map, with a lot of grass. But always the same tile would be repetitive. Thankfully the artists did a few different tiles for the grass, so you can have a bit of variety.
Now, wouldn''t it be nice if instead of having to put yourself the variety (by changing the tile you want) you could simply put grass everywhere, and the program would automagically add the random tiles in the grass)
Similarly, in HalfLife maps, you can have a random set of textures.
I dont find it too hard to imagine that you could extend this concept to something more than simply texture.
Have a few models of houses that occupy a certain standard surface, and in your editor, simply paint the map with a generic town house indicator. The editor would randomly insert a house among the variants you selected...
The machine is here to help you with boring tasks, so why not just let it do its job ?
Sancte Isidore ora pro nobis !
To be honest, I *do* mind repetition, but whereas in a randomly generated environment I will expect it, hence not be as shocked, I''ll be sorely dissapointed in a game supposedly *handmade* where all towns are really the same big things with different textures.
Random elements are useful if you want to add variety without having to wreck your head. Think of a similar problem : you want to make a Starcraft (or any other RTS) map, with a lot of grass. But always the same tile would be repetitive. Thankfully the artists did a few different tiles for the grass, so you can have a bit of variety.
Now, wouldn''t it be nice if instead of having to put yourself the variety (by changing the tile you want) you could simply put grass everywhere, and the program would automagically add the random tiles in the grass)
Similarly, in HalfLife maps, you can have a random set of textures.
I dont find it too hard to imagine that you could extend this concept to something more than simply texture.
Have a few models of houses that occupy a certain standard surface, and in your editor, simply paint the map with a generic town house indicator. The editor would randomly insert a house among the variants you selected...
The machine is here to help you with boring tasks, so why not just let it do its job ?
Sancte Isidore ora pro nobis !
-----------------------------Sancte Isidore ora pro nobis !
quote: Original post by ahw
Kylotan & Zegfried : well, at least we have people on both side of the fence.
To be honest, I *do* mind repetition, but whereas in a randomly generated environment I will expect it, hence not be as shocked, I''ll be sorely dissapointed in a game supposedly *handmade* where all towns are really the same big things with different textures.
Well, repetition is a fact of life. Almost everything is made out of just protons, neutrons, and electrons, yet there is more variety in these combinations than any of us are ever likely to see. The trick is in combining them in an interesting manner. As long as procedurally-generated areas for games is not a simple case of "for each tile, randomly pick a terrain feature", then there is hope.
[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions | Organising code files | My stuff ]
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement