Yes, I know that PA is alpha. I'm asking this question more out of curiosity than anything else: Why isn't PA's pathfinding as good as it can be from the start? I was under the impression that the developers behind TA and SC are working at Uber. TA and SC had pretty good pathfinding for the most part. Wouldn't it be a matter of just using the same pathfinding code more or less from those older games to PA? I've read comments on the PA forum that seem to be unanimous about the need for PA's pathfinding to improve. So is Uber recreating the wheel so to speak when it comes to PA's pathfinding? Is there something about PA that makes pathfinding inherently difficult to code for? I'm not a programmer. But as a banker and lawyer, I've been hired by different companies in my career to create policy and strategy. In each case, I've pretty much created policies and strategies as good as they can be from the start by using my inherent knowledge and experience, rather than recreating the proverbial wheel. Granted, I'll tweak the policies and strategies here and there as needed. But the impression I get from comments about PA's pathfinding is that more work is needed than just "tweaking".
Because its not done. Because they don't own the rights to these games and cannot use anything from them in PA. Basically yes. Well path finding on a sphere is a bit different than on a flat map but I am not sure if that really makes it much harder or not. That's pretty much what Uber is doing. It just takes a while to put everything together and get it working with everything else since so much of it is new.
Ladies and Gentlemen, the internet. Welcome to the wonderful world of Game DEVELOPMENT PA is a new engine with different code, so it can't just "use the same" stuff as TA did decades ago. PA is also based on sphere-based maps, so it's a whole different beast to tackle. Experience in finance and business doesn't mean you know a thing about game development, much as you'd like to think
It would be easier for me to understand if Uber's development team were new to RTS gaming and the like. But they're not. They're highly experienced, skilled developers who have worked on RTS games in the past like TA and SC, the spiritual successors to PA. I'd have thought that something as inherently basic in RTS games like pathfinding would be a cinch to develop, as if it were the "backs of their hands".
Then you clearly have no idea how gaming works. You can easily re-use code and pipelines if you're using the same engine as before, but TA is VERY old, and SupCom isn't exactly recent: neither used similar engines, and PA is having everything for it built from the ground up. (The only game so far that's used the in house PA engine is Outland Games, which is a basic iOS app). If you don't have a knowledge of how game development works, just stop now. And from what basis are you getting that this is "inherently basic"? Please lend me your credentials in AI.
If you're interested in learning more about he Pathfinding, I suggest you view the Jan 10th and Mar 22nd Live Streams, they both feature some discussion on it. Otherwise it's as kvalhiem said, They can't use the older systems for a variety of reasons, Luckily the guy they got working on it is pretty much the expert of flowfield pathfinding with the Game Industry. The Aim is to create a responsive system like what was seen in SupCom2, without the skating and clipping and focusing more on having units move realistically as they did in SupCom1/FA. It's something I've talked with them about on the forums a few times, and they'd set my fears to rest, now we just wait to see how it pans out long term and to tackle and problems that crop up. Mike
Mike post link to stream where Uber explaining well why SC pathfinding wasn't good and why it's doesn't fit PA needs. Main problem is old-style pathfinding create big bottleneck on CPU, in PA it's should work fine even with hundreds of thousands units because engine not calculated unique paths for each unit.
That's good to hear that PA's pathfinding will eschew elements like the skating and clipping of SC2. I had seen all the hype about SC2's flowfield pathfinding only to discover firsthand that the units were basically traveling through each other. Talk about a big let down. Thanks for the links on pathfinding. I'll check them out. If your fears are laid to rest on pathfinding, then I take that as an excellent sign that bodes well for PA. I'm familiar with you and your team's excellent work on SCFA's Black Ops mod. Your team doesn't settle for anything less but the best. I take it English is not your primary language, is it? I said "inherently basic in RTS" as in pathfinding is a major component of RTS games. Understand now?
1. My first language is english, and "inherently basic in RTS" is not the same as "major component of RTS". It sounds far more like your saying pathfinding is the most trivial of tasks. 2. SC2 = Starcraft 2, not Supreme Commander 2 3. They could not have used the pathfinding from SupCom 1 or TA even if they owned the games. They are not based on flowfield at all. 4. Just like you said that you have everything planned out completely from the get go when doing your job, i can guarantee that the pathfinding was completely planned out in the same sense, but just because its planned out does not mean that it is a finished product. Also, as far as i know, the world of Law and politics is fairly static whereas Pathfinding is still a major front of research in the world of Computer Science, meaning new things pop up all the time and could warrant a change to a developing system like this.
1. English may be your primary language but you still have work to do when it comes to interpretation of prose. "Basic" does not mean "trivial" in its use in my sentence. It means "primary", as in DNA is the basic building blocks of complex organisms like man. 2. I've seen SC2 used to refer to Supreme Commander 2 so that's what I'm going with. Feel free to use whatever acronym you wish. 3. So Uber is basically creating the pathfinding from scratch. Great. 4. Agreed. That might be true in your country, but here in the US, law and politics are dynamic. Just today, the US Supreme Court announced 2 rulings. And more are to come. They always do. But I'm sure that tech elitists will rationalize that their industry is more dynamic. Most stubborn people always do....
I don't think techies think their world moves faster than others, but it does move pretty fast! Especially when it comes to games taking advantage of the latest hardware. Which PA is aimed at, even though it will run on older machines too. PA is trying to be bleeding edge with a lot of its feqtures - the stuff these guys are writing now could be re-used a lot in other games over the next 5-6 years, especially the replay stuff as I think that will be quite portable. I'm a programmer, and the whole idea is to make stuff re-usable in the long run, to get value out of the initial work. So because they're doing so many new things, it just takes time. it just becomes bad if it takes too much time, and its still not right. The Flow field (path-finding) stuff is pretty new from what I've heard, and its also all done on the Server to make it easier for your own PC's processor, so again, the model is totally different to SupCom etc. where everything is run on every PC in the game, so the game runs at the speed of the slowest PC, even if some people's are great. Im only a beta backer, so have to wait until September, which seems like a long time, but they will get it right by release, and it will be interesting to be part of the beta given I'm on the receiving end this time. Mike
Main reason you can't just immediately apply all previously learned lessons from the games we've worked on is, quite simply, we're making a very different game. We're creating procedurally generated planets for one things, something not normally done in an RTS because it creates pathing problems. We're creating round planets for another, which has pretty much never been done in a large scale RTS. Which causes a huge host of problems - most pathing we've done in the past assumes a plane. We're also procedurally adding/subtracting a large amount of CSG for the planets, so you can't even just assume land/water. You have to look for crevices. Hills. At any rate, there are a number of reasons why you'll see "solved problems" not appear to be quite so solved. Unless you're just making version 5 of the same game, the problems you're solving are usually pretty new. In our case, very much so. Hope that helps to understand the state of things. While we may be making a game that borrows from a number of our favorite games, we're very much making a brand new game on most of the fundamental levels.
Garat and sower85, Thank you for your answers. They were very educational and enlightening (and I appreciate your giving them without any traces of condescension or sarcasm). A few more answers like that and I may just become an alpha backer of PA yet! Thank you again.
Yeah they are a great bunch of guys. Always helpful and professional. The job of being sarcastic falls to other members of the community.
Those other members of the community would serve PA and Uber better by following the example set by Garat and sower85.
Omega, As much as we try and post in here, there is still a lot of posting going on here, and no real patience for thin skins. As I've told others, if you feel like you're getting attacked, take a second to see if they're attacking you, or attacking your post. If they're attacking your post, just roll with it and respond to the points. If they're attacking YOU, however, that's when you use the flag post to get mine, or one of our great mods, attention so we can knock some sense into the overly pejorative poster.
