People talk a lot about things like flow, pathing, balance, rotation, etc. These people seem to have clear definitions for each of these concepts, and they seem to design their maps with a clear understanding of what they think is appropriate in each of these categories. I think this means that these concepts can be clearly defined within an algorithm. That algorithm might then be used to, at least partially, procedurally generate a fun and well-balanced map. Other design elements may similarly be defined, or act as a set of additional rules for an algorithm, such as determining the amount of floorspace, vertical differences, weapon placement, etc. It seems to me that a lot of map design is people trying to follow their own set of rules, and then iteratively innovate with each new map. I think this is similar to following a procedure or algorithm by hand. And just like the designer's map philosophies, an alogirthm can similarly be changed over time. What percentage of map design, do you think, can be done using an algorithm? I am of the opinion that 100% of map design can be done with algorithms, even the art. That doesn't mean that the creation of these algorithms is trivial, however. They could be extremely complicated, or maybe stupid simple. I wrote a little Python script to essentially generate blueprints for forge maps. I've run it a few times and slopped each of them out, and each one of them seemed pretty decent to me.
Algorithms can be used to make all kinds of things, but, food is the original art. I'm not saying that it is impossible to make art with an algorithm, I'm just saying.. On another note, people often use the word "trivial" to mean "superfluous." But, did you know that the word "trivial" comes from the the greek, "Trivia." Trivia, is the god of "where three roads meet." (Tri meaning three, and via meaning road) Which was a big deal back then. I'm those days roads were long and dangerous, so any place where three roads meet was a place where people could exchange goods without traveling the whole way. This meant that people could return to their respective homes with, not only merchandise, but information, and stories from other lands. So the conclusion may be drawn, in retrospect, that the god Trivia is partially responsible for creating what we now think of as myths about some of the classic Greek jeeps (edit: heros) like Achilles, Hercules, and Odysseus. If we chose to think of these myths as a collective of similar tales that were adapted and woven together over time. (Edit: Sorry those are bad examples. Just the first three names that came to mind. Try instead, Hercules, Orion, and Theseus.) So why is this relevant? What do the ancient Greeks know about playing Halo? Not a thing, that's for sure. But then, what do we know about constructing a myth? A real myth. What have we done to contribute to the kinds of artistic work that has lasted for thousands of years and inspired innumerable individuals? To the point of influencing entire civilizations. Even, debatably, our DNA. Compared to that, what is designing a map? But then, again, who can name a person that DOES have first hand experience constructing a myth? How is it that we have come to recognize even the three names I mention here? In this way, don't we come to know mythical figures in the same manner by which we have always come to know them? With that in mind, what does it really mean to play Halo? But here I go, rambling about trivial matters.. Just food for thought, that's all.
How does that work? Do you have examples on forgehub? Does it actually make the art in the blueprint? Or do you add stuff in during the Forging process? What is the meaning of life? And does it make objectives like ctf?
You can only algorithmize something if you completely understand it. We're better off just sticking to our own personal calculations, because everyone has a different level of understanding.
But what if the algorithm he used makes blueprints that are better than what 343i can make on their own? Or at the very least, what if the algorithm makes something that is a reliable starting point for people just starting to Forge? Could that not make it easier to introduce people who are new to Halo to a side of the game they may not otherwise be familiar with?
That's not true. I can build a Python script to do completely bonkers stuff, even if I don't understand it. Also, I believe that people can have an extremely deep understanding of their own design philosophies, and therefore can write them down as a set of rules that an algorithm can follow. That doesn't necessarily mean that they know everything about map design.
Okay, then the algorithm won't make good maps if you don't understand what makes a good design and use that knowledge to build the script.
I agree, unless the algorithm designer gets lucky. But, people do know how to make good maps, right? So then do you think they ought to be able to build algorithms for this purpose?
I would say that the process of designing a map IS an algorithm of sorts in your brain, except a conscious person can implement their creativity and intentionally break the algorithm to mix things up.
Algorithms are what we create in order to HELP us understand things. The word comes from the influence of the greek "arithmos" (meaning "number") which combined with an older Arabic system of counting, called algorism, to eventually evolve into what we now call algebra. Yes, it is true that all thinking beings must rely on our brain's own internal algorithms, (and there could be thouands) to interpret the physical world. That's just one of the things that brains do. But, another thing our brains can do is create new algorithms in order to solve new problems, the kinds of problems that our bodies are not naturally equipped to handle. And I would, again, argue that this adaptation marks another major turning point in our evolution. I believe it continues to do so, to this day, despite any circumstantial evidence to the contrary. Just for the hell of it, I'll go further and point out how C.S. Lewis used the character of Aslan as an allegory for Jesus, in the ferociously popular book series, "The Chronicles of Narnia."
I sloppily sketched this the other day based on the premise that negative space is proportionately maintained throughout on the XYZ axes. I'm sure you could generate a script that could stack geometry in such a way and assign value to the way things are distributed. Hell, you could probably even create an algorithm that draws telelines between different areas and decides where to put sender and receivers.
Windows used to have a screen saver that did that It was called "pipes".Used an algorithm to make a three dimensional maze of pipes.Every one was different, although I suppose they would repeat at some point.
I don’t think it’s a question of whether or not something procedural can be made even more procedural. The focus for me is when to cut off the procedural and begin the personalized randomness and how much time to budget on it. Identifying that point is maybe just another process, but it’s the fun part. I don’t want to sell it for efficiency, just yet. An algorithm would make a good manager, though, ready to tell me when I have enough of some particular element. The subdivided limits for objects in previous Halos used to have that effect.
This is a very interesting topic from an analytical standpoint and computer programming. I think everyone should comment on a variable that would be consistent for map design. He could upload the majority of them in the scripting and share 20 layouts to see if they are too similar or have unique enough layouts. I know @Goat had mad a thread about map designs previously so if nothing else, @Spranklz could read that and implement the rules, but a group input might prove more beneficial to the findings. To keep the layouts design simple to read things like pillars or line of site blocker would be better to be left out perhaps and just let it create the hardwalls, deathpits and open space. Additional doorways or line of sight blockers would be easier to leave out until a map was "forged" possibly. I don't know if I would include the regular 3-lane, donut algorithms just to avoid generic designs, but adding those rules could end up in it being basic pathway flow structures. A separate algorithm could possibly be created where the user would input starting travel (aka spawn points) and see where a computer would determine the most common travel paths (heat map) would be for it. That would allow us to compare how it would actually play and even eventually allow it to develop suggested locations for teleporters based on the heat maps. IE: Maps that are for 4 vs 4's would not exceed 520x520 units (I believe this is a relative size of most 4 vs 4 maps) All walkways can't not be less than 16-20 units wide Each room can't be less than 20x20 units Each map must include height variation of XYZ height (rooms tall enough to jump in, but if they connect to a lower floor, said inclines can not exceed 30 degree's) All stairs / inclines can not exceed 45 degree's Doorways can't be smaller than 8-10 units in width Optional toggle on or off option to see how a map would be designed when run both ways - A jump over a gap should not exceed X width (spartan jump distance with or without clamber)
I know **** all about core level design and while I think you could do it, it would also take all the character, charm and soul out of the map. Knowing someone put hours of work into design a product for people to enjoy is what actually makes the map in my opinion. I get this wasn't part of the initial discussion and don't get me wrong, machine learning via input can result in some great stuff but I think it just wouldn't be the same. But good luck anyway, mate.
I have always enjoyed the machine learning vs man debate. No amount of algorithm's can ever give it character or even maybe the added foresight of the "tug of war balance" on a map, but it would be interesting to see what it churned out.
Depends on the type of game. Is it possible to make entire levels completable via algorithm? Of course. Can it ever make a level as meticulously crafted and unique as a skilled level designer can? I'd say no. It tends to work the best in open world type games, rogue-likes, and crafting/survival games, where parts of the world can fit together like puzzle pieces. I would never expect an algorithm to be able to build the Last of Us or Uncharted 4, but i do expect it to be able to make Dead Cells, Minecraft, and Spelunky.