Welcome to Keen Software House Forums! Log in or Sign up to interact with the KSH community.
  1. You are currently browsing our forum as a guest. Create your own forum account to access all forum functionality.

The Major Problems that Ruin Survival.

Discussion in 'Survival' started by Project 06, May 24, 2016.

Thread Status:
This last post in this thread was made more than 31 days old.
  1. kruleworld Trainee Engineer

    Messages:
    23
    It's the same with drones and meteors. They spawn quickly and keep punishing you, not giving you time to fix the damage from the last wave.
     
    • Agree Agree x 2
  2. odizzido Junior Engineer

    Messages:
    683
    For me it's mostly the bugs that ruin survival. I made a scenario with the objective of death by lack of resources being your main obstacle. I made a custom planet with no ice or uranium to do this which was bug free and easy which is great.

    Power isn't terrible for bugs but even if you have zero uranium it's not hard to get because you can keep building more and more batteries which start off with quite a lot. Adding in an option at world creation for batteries to start with zero power(and grind back into power cells) would help a lot with making power more difficult.

    Oxygen just simply doesn't work at all right now. I mean, it works in that you need it, but I keep getting a bug where my oxygen tanks just fill up from nothing making having enough trivial.

    I also planned of having my scenario time limited so that if you managed to survive at all, you'd still need to move quickly to complete your objective. But then of course the scenario time limit doesn't work either so adding up free oxygen, no time limit, and easy power my hardcore survival scenario turns into easy mode.

    It sucks because I love this game and want to create a hardcore survival series but I just can't right now. I have been following the development of this game since I got it though and from what I have seen development is very active so I am hoping it won't be too long before this is possible.
     
  3. Sagi Apprentice Engineer

    Messages:
    265
    They altered a while ago that the batteries started with power because many player made stuff (mostly torpedo's) where unable to be used properly since the battery need to recharge first.
    It would be a nice idea to have both batteries one that should start empty and is used for storing power and one with a much smaller capacity but wich has power when build.
     
  4. Azirahael Apprentice Engineer

    Messages:
    154
    • Like Like x 2
  5. ViroMan Senior Engineer

    Messages:
    1,123
    see this is why I encourage people to post links to mods that they believe are relevant to peoples problems. This is one of thoes things I never knew I needed until I saw it and now I say... "why the F didn't someone tell me this existed!?"
     
  6. Azirahael Apprentice Engineer

    Messages:
    154
    Glad i could help :)

    Yeah, it's a little thing, but it's so useful.
    Great for small torpedoes, projectors, work lights etc.
     
  7. ViroMan Senior Engineer

    Messages:
    1,123
    This is relevant to me because I need this for transporting crap so easy and for connecting large to small grids easy.
     
    • Like Like x 1
  8. Comicsluvr Trainee Engineer

    Messages:
    34
    My main problem with Survival is the pass/fail way of things. Either you turn on all of the threats like meteors, spiders, pirates and get crushed or you leave them off and within 10 hours you've established a foothold and death (unless you do something dumb) is no longer a problem. I'd like to see other threats to our health and safety like Radiation Storms, Solar Flares, the occasional LARGE rock from space etc.

    I'd also like to see more control over the threats we have already. Like Pirates once every 24 hours means that for the FIRST 24 hours you get none and THEN the counter begins. This gives you a chance to get started.

    Right now the game is either too difficult (I can't find ANY Nickel and I can't build Metal Grids without it!) or not challenging enough.
     
    • Like Like x 2
  9. ViroMan Senior Engineer

    Messages:
    1,123
    Wolves are NOT a threat. They are an annoyance in that they make noise. They can hardly do any damage at all to heavy armor. All you have to do is build your base on stilts with the 3 lowest blocks on each stilt of heavy armor. It takes a long time for wolves to damage them. If they manage to take out a leg... you deserved it. Anti-personnel turrets are cheap and deal with wolves rather quick as well. You can't complain about putting up 3-4 of those turrets if you don't want to personally deal with the noisy dogs that for some reason carry pipes and computers and ore.

    Seriously... are these like trained carrier dogs that got sidetracked to attack people that are on there path? Why are they carrying crap now. It was OK'ish when they were cyber dogs but, now that they are wolves this makes no senses.
     
  10. McTraveller Apprentice Engineer

    Messages:
    118
    So what I'm hearing here is that the challenges in SE are essentially static - even meteors, once you get established to survive even apocalyptic meteors, you're basically done.

    Trouble is, challenges that adapt and continue to get more and more challenging are currently beyond personal-computer-level computation.

    Sad thing is, if you want that level of challenge, you basically need multiplayer, and that's got its own issues...
     
  11. ViroMan Senior Engineer

    Messages:
    1,123
    This is true. They could make an easy adjustment to pirate AI by having them spawn more aggressive units(bigger/heavy armor plated/more weapons) after X amount of losses or have them start spawning more units per spawn. As for the others... it would be silly to have them spawn more wolves or more meteors.
     
  12. halipatsui Senior Engineer

    Messages:
    1,253
    One could fill a planet with pirate bases in creative and then start survival. Maybe even enhance(or have a friend enhance them for surprises)
     
  13. Sirhan Blixt Apprentice Engineer

    Messages:
    467
    Like Hell.

    Postgraduate comp-sci students are doing experiments in at least simulating intelligent behavior in games all the time. You can walk into most Barnes and Noble locations (chain bookstore in Yoo Ess) and find books on "How To Program Game AI" that discuss very simple ways to at least simulate intelligent behavior. Yet little of this finds its way into actual, published games because the focus is typically on making things prettier. The "state of the art" of game development has kept pace with the abilities and feature sets of contemporary video cards, and little else.

    Seriously-- are there any recently-published games that one could say, with a straight face, couldn't run on older systems because they'd ask too much of anything except the video card? Except for the overhead required to shove data into that video card, not much is demanded of the rest of the hardware. Computer AI is at least as butt-stupid as it was in games published ten or even twenty years ago. There are still many games published today that don't take advantage of multi-core architecture, even though few contemporary PCs have a CPU inside with less than four of them.

    And let's not even mention existing games with AI players that at least pretend to be able to adapt to changing conditions brought on by human players and each other.

    If the developers say that they don't have the resources to spend on making in-game entities act more intelligently, then that's fair. They operate under constraints of time and budget in an environment where the corner-office folks are happy if the game looks shiny enough. Stupid, simplistic in-game "AI" is due to deliberate choice, not absolute limitations imposed by the state and capabilities of contemporary PC hardware.
     
  14. ViroMan Senior Engineer

    Messages:
    1,123
    FEAR was very high tech in the AI department when it came out. NPCs actually communicated with each other to converge on and flank you, used cover very effectively, and even made cover by knocking things over. FEAR's AI never cheated, they seemed to act like a real squad of soldiers would. I don't even see these qualities in newer titles. Today's AI always cheats and acts rather stupid. Once you alert the AI in an area it always knows where you are.
     
  15. odizzido Junior Engineer

    Messages:
    683
    I blame a lot of this on consoleitis. They have been holding back what games could have done ever since things started going multiplatform.

    One of the big things has always been a lack of ram. If you looked at the amount of ram fear and some other pc only games used around that time you would have seen that it is actually more than most other games used that came after it for many years. Almost all games capped at around 800mb so they could run on consoles.

    It wasn't until the xbone and ps4 came out that the 800mb limit vanished. Of course this has been replaced by extremely poor single thread cpu performance on the new ones. I don't know why they decided to go with the amd cat cores for what should be a performance system.

    Anyways my point is that the 360/ps3 were crap and didn't leave much room for smart ai in terms of system resources and then when they finally released the bone/ps4 they crippled them with netbook class processors so there still wasn't any processing power for ai. It's no wonder that things have been going backwards.
     
    • Agree Agree x 4
  16. Devon_v Senior Engineer

    Messages:
    1,602
    AI is hard. Cheating is easy. Most companies won't spend time making an opponent who is smart and fair when they can just use a couple of quick hacks and make something that cheats in a balanced way.

    For a current game that adapts, check out Infested Planet. It's an RTS puzzler, and each time you take a point the AI tries to reason out what you did and evole a counter for the next point. It's not very smart, but it makes an effort. (I haven't played the new random campaign mode, the AI's adaptations may work better in a global random world vs. the canned levels in the original. )

    SE doesn't even need strong AI, just somehing like Left 4 Dead's Director or Rimworld's Storyteller would be great. Just a system that gets a feel for what you're doing, and what your capabilities are and sends specific things at you to test you.
     
  17. Sagi Apprentice Engineer

    Messages:
    265
    Or if the game get's more stable more human opponents could appear.
     
  18. odizzido Junior Engineer

    Messages:
    683
    yup. I don't play SE much anymore because I am just waiting for the multiplayer to become, for me, playable.
     
    • Agree Agree x 1
  19. McTraveller Apprentice Engineer

    Messages:
    118
    Ok, so yes, I was probably a bit overreaching in my assertion, and maybe not accurate in saying "desktop level computational ability" but rather something along the lines of what you said at the end about not being enough of a priority for game developer studios.

    That's all still a maybe though: I feel like people are forgetting that I had the word 'adaptive' in my post - not that an AI could start out difficult and do "complex" things like flanking or whatever. But one that that learns as you play against it - and has some kind of goal it is trying to optimize. And this is tricky, because if the game AI learns too fast compared to the player, the game ceases to be fun - a game AI has to be "just hard enough" as a player develops. Maybe this is the thing that trips people up - because there's "challenging" and then there is "realistic." I'd rather face a realistic enemy than a challenging one - even if the realistic one is in fact challenging.

    Or maybe you just need some kind of configuration to say "This is what I consider 'fun'" and the AI can learn that. Note that this is all also especially directed at "AI in a sandbox environment" rather than in an FPS or RTS or even tower defense or something. In an open, persistent world, AI is difficult to balance well and provide a nice gaming experience.
     
  20. Sirhan Blixt Apprentice Engineer

    Messages:
    467
    The word "adapt" was included in your text I quoted, so I hadn't forgotten. The "FEAR" example is one of simulating intelligent behavior -- giving AI opponents some ability to share information, but otherwise using client-side scripting and map navigation resources to make them aware of what paths are available and what dynamic in-world entities they ought to hide behind. Even "Half-Life" did this, but "FEAR" was a bit more slick about it. I'm not saying that this simulation of intelligent behavior is unworthy, just that there's no learning or adaptation going on, just different execution paths in a Lua script. Developers of first-person shooters and similar should consider themselves to have an incentive to refine this behavior, as was done for "FEAR", because with their increasing usage of shiny graphics dohickeys, there is an upper limit on how many dynamic actors there can be in-scene. They need to make the most of them, because just flooding the room with dozens of enemies went out with sprite-based rendering.

    Regrettably, there is little room in those kinds of games for genuinely adaptive AI, because there are few tactical variations that the player can come up with that cannot be countered with more scripting. Especially when most single-player maps are just linear paths from one cutscene to the next.

    On the other hand, there are types of games that benefit greatly from adaptive AI behavior: RTS, 4X, and similar where there are multiple dimensions of progress within a game ruleset that allows wide variation in how progress occurs. Pretty much anything that's not a shooter, really. I just get wrapped up about shooters because some do a really good job of simulating intelligent behavior -- learning and adaptation -- without actually implementing any.

    And here you're paddling into Unsupportable Generalization Waters again. What is it about an open, persistent world that would make balancing AI behavior more difficult than any other game balancing? Unless you want to expand the conversation to particular implementation details, I don't think it's fair to say that balancing AI behavior vs. any other game behavior would present such an increase in difficulty that said difficulty in itself would cause adaptive, learning AI behavior to be eschewed in such a game.

    Balancing game behavior is always a serious challenge, and one that few development teams get right the first time. One of the most serious symptoms of a failure in this regard that always offends me is when one difficulty setting gives you pathetic, easily defeated opponents, but the next notch up gets you owned. In the face. Every time.
     
  21. odizzido Junior Engineer

    Messages:
    683
    The thing that offends me the most it's giving the ai different rules.

    For example giving the ai in civ4 no penalty for building too many cities so it can expand without killing itself. You end up seeing it build cities in the snow with nothing useful around. If you built a city there it would do nothing but harm your progress but because the ai has a different set of rules it actually helps it.

    Worse still is something like empire earth where the ai gets free resources. You try to strategically cut off iron because it makes sense from your ruleset but since the ai doesn't need to gather resources at all it's pointless. It really sucks.
     
  22. Sirhan Blixt Apprentice Engineer

    Messages:
    467
    Also not a fan of cheating "AI". I still remember the crushing disappointment I felt playing the original "Civilization" and seeing another civ's trireme float by even after I'd figured out that mine was the only civilization on that particular land mass.

    What that guy said.
     
  23. McTraveller Apprentice Engineer

    Messages:
    118
    I suppose you might be correct there again - balancing is balancing, eh?

    But more specifically: I think one of the issues with 'balance' (AI or otherwise - many player-only games have balance issues) is that there isn't really a universal way to define what 'balance' means. This is what I meant by "perhaps have a kind of configuration to say what the player considers 'fun'". Most discussions of balance I've seen seem to center around ideas like "everything has a counter" or "everything is equally viable" or "more powerful things should be more rare / more expensive". But as others in this thread have noted, it may also include things like "AI should have the same rules as players, and not 'cheat'".

    So when I said balance is very difficult in an open sandbox, I was referring to the fact that there is likely no feasible definition of balance in the first place - which is quite different than an RTS or shooter or even a sport like baseball where there are more clear boundaries in which to define balance.

    The best approach I could see in a sandbox is not 'balance' but what I would call 'consistency' - more like what the real physical universe has. It isn't really what you'd call balanced (e.g., quantity of raw materials isn't really tied to their chemical usefulness), but it is consistent (everything follows the same laws of physics). In space engineers, what this would mean is something like the discussions in the balancing thread about how block capabilities and component requirements should have fixed scaling rules rather than just arbitrary assignments; square-cubed laws, torque of a rotor should scale with number of required motors (it doesn't!), all that kind of stuff.
     
  24. Sirhan Blixt Apprentice Engineer

    Messages:
    467
    It's a tough needle to thread, to be sure, but having AI players being subject to the same rules as meat players is a good first step.

    Being fun, or even playable, is also a requirement not every game development team manages. It's regrettably easy to come up with a product that is challenging and follows natural laws, like physics, closely ... but which has crossed the line from "game" to "anti-game". This potential pitfall is particularly acute in "survival"-styled games -- which is why we have this thread.

    In all fairness to KSH, Space Engineers' survival mode is more game than many other offerings in the "Survival" category. This is the part where I throw tons and tons of shade on Day-Z, with horrible balance when I last played it. There, if you're on just a bad server, or otherwise don't roll sevens enough times in a row, your playthrough is fifteen minutes of gnawing frustration ending with either starvation or being mauled by a zombie. Anti-game.
     
  25. tankmayvin Senior Engineer

    Messages:
    2,864
    Once you are more seasoned at the game no combination of the in-game threads will crush you as a player.

    Meteors can be painlessly avoided by hiding all your grids away from sun behind asteroid. Pirates are pretty wimpy and don't aggressively attack you. The bugs and wolves are merely an annoyance once you get a couple of interior turrets setup.

    All of these things merely slow you down once you know to prepare for them and that's what makes them crappy - they just annoy you.

    Pirates are the most fun because of the salvage opportunity.
     
  26. ViroMan Senior Engineer

    Messages:
    1,123
    The mods out there to beef up the pirate designs, spice that up a bit too. You can get a massive battleship that spawns once in a while.
     
  27. tankmayvin Senior Engineer

    Messages:
    2,864
    Problem is that they are all dumb in terms of AI so bigger/better armed ship really just means more turrets to kite/snipe.
     
  28. ViroMan Senior Engineer

    Messages:
    1,123
    yes but, it is a bigger prize for a longer time spent to get it. There was one pirate mod that spawns a carrier that builds mini bots to send out. I never got to try that one because a keen update broke it and they said F it and didn't fix it.
     
  29. System Error Message Trainee Engineer

    Messages:
    96
    im hosting a survival server and playing. With all elements (wolves, spiders, meteors, etc) its a lot nicer. Every once in a while a meteor hits my ship but the damage is negligable (only a few light armour blocks). The physics issues and game bugs are what need to be fixed first, things like pistons and rotors on dedicated servers. I had to disable meteors because everytime a meteor hit a planet or asteroid surface my game crashes. When a gattling turret kills a wolf it flies away instantly like as if the bullet as an infinite force. with block destruction disabled sometimes ships just start jumping around like crazy and i have had serious desync issues even on LAN.

    I think the critters arent vicious enough. Spiders should be able to tear into your ship (though light armour and interior walls) and climb around. Infact the atmospheric lander comes with ammo for your machine gun but not for turrets and you can loot them so there is a reward for shooting them. At the start mining resources and scavaging critters is a way. The spiders should remind you of starship troopers, their capability to burrow through structures. Critters need an objective, such as to guard an area (wolves preventing players from getting near resources spots), spiders that go for players and also a particular component in the vehicle (could be for uranium or even some sort of component or part), something that can be lost but doesnt mean having to restart.

    The game becomes balanced when you mix in all elements together rather than only part of it. I have asteroids that spawn in planatary gravity fields but in space so that makes it easier to start however some people forget they're in a gravity field. It keeps the npcs at bay but its easier to mine an asteroid than on the planet itself. Mars doesnt seem to have critters but has no oxygen which really isnt a problem if you do include for ice in your search. Realistically some metal ores give oxygen when processed (like iron as it naturally exists as iron oxide). Pure iron is grey/black in colour. None of the spawnships however are capable of a proper start which is why i built my own so the player can start on any planet/moon/asteroid but stuck to the same loadout as the atmospheric lander.

    Meteors target players, when you're not around they dont strike. When you're on the server they follow the setting used and target the area around you. So if you move away from your ship during a meteor storm you can save it. Gattling turrets are very helpful but you should construct some turrets and dedicated them only to missiles and meteors (and characters?).
     
  30. Soup Toaster Apprentice Engineer

    Messages:
    221
    How good are turrets at stopping meteors on the Stable version? It's been a while since I played with them turned on as they used to kick you butt pretty relentlessly so the only option was to hide underground.
     
Thread Status:
This last post in this thread was made more than 31 days old.