This game might not be as (un)playable as you think.
noisywalrus
Join Date: 2010-12-15 Member: 75694Members
Join Date: 2010-12-15 Member: 75694Members
<div class="IPBDescription">Opinions about balance are skewed right now. Here's why.</div>Until last week, I had this 2.4Ghz quad-core Intel processor (http://ark.intel.com/products/29765) and an Nvidia 8800GT. The game at lowest settings was playable, but dragged a bit. Trying to hit anything in CQB was impossible. Even dead center shots did not register.
For work, I did a hardware refresh last week. Better stuff, i7 + ATI 6850 + 120GB SSD, but not extreme by any measure. Absolutely within consumer range, but still probably higher spec than a good portion of the NS2 player base.
The difference is shocking. I can solo fades with a shotgun about 30% of the time. Everyone looks like a keyboard player. I'm not a very good player due to being out of practice for several years, but it was almost trivial to kill... anything.
This dichotomy affects balance so much that I think that balancing the game right now is a task approaching Busywork. Before hit detection, movement smoothing, etc. is less locked to client/server overhead, the only productive opinions (positive or negative) about balance can come from people who can play at a rock-solid 30+ fps on a server maintaining a good tick rate. That shouldn't be.
To Unknown Worlds: From one software engineer to another, I know that resources are limited, but stuff like your server programmer working on VOIP codec integrations (even if it's just a day) seems unusual. Server stability. Client stability. Even if you have to implement a mode that plays with no gear, no IO/CPU-hogging structures, low-poly everything, etc. I am fully aware that on the other side of that equation is marketing and promotion and screenshots, but remember that Minecraft, League of Legends, and countless others make a jillion dollars. If it's fun, people will play it. If it's not fun, no amount of pretty screenshots will make up the difference. When was the last time you had any friends who were itching to play Crysis online?
To Players: If you think that other posters are being needlessly whiny or nerf-batty, consider that they may have no idea what this game is supposed to look and play like on good hardware. Watching YouTube videos does not convey this. I know it's asking a lot, but try to play on a machine with a smooth framerate and see if you think "X is OP".
For work, I did a hardware refresh last week. Better stuff, i7 + ATI 6850 + 120GB SSD, but not extreme by any measure. Absolutely within consumer range, but still probably higher spec than a good portion of the NS2 player base.
The difference is shocking. I can solo fades with a shotgun about 30% of the time. Everyone looks like a keyboard player. I'm not a very good player due to being out of practice for several years, but it was almost trivial to kill... anything.
This dichotomy affects balance so much that I think that balancing the game right now is a task approaching Busywork. Before hit detection, movement smoothing, etc. is less locked to client/server overhead, the only productive opinions (positive or negative) about balance can come from people who can play at a rock-solid 30+ fps on a server maintaining a good tick rate. That shouldn't be.
To Unknown Worlds: From one software engineer to another, I know that resources are limited, but stuff like your server programmer working on VOIP codec integrations (even if it's just a day) seems unusual. Server stability. Client stability. Even if you have to implement a mode that plays with no gear, no IO/CPU-hogging structures, low-poly everything, etc. I am fully aware that on the other side of that equation is marketing and promotion and screenshots, but remember that Minecraft, League of Legends, and countless others make a jillion dollars. If it's fun, people will play it. If it's not fun, no amount of pretty screenshots will make up the difference. When was the last time you had any friends who were itching to play Crysis online?
To Players: If you think that other posters are being needlessly whiny or nerf-batty, consider that they may have no idea what this game is supposed to look and play like on good hardware. Watching YouTube videos does not convey this. I know it's asking a lot, but try to play on a machine with a smooth framerate and see if you think "X is OP".
Comments
I agree, with the slowing rounds, fades are pretty easy cake with shotguns, even solo.
Then again when I play fade I feel unstoppable as well.
<ul><li>a low, barebones graphics setting that allows for high FPS gameplay (ex my rig gets 150 FPS avg in TF2 and only 25 on the lowest setting in NS2 - the largest deterrent to playing NS2)</li><li>high server tickrate (L4D uses 33 max and that is not high enough)</li><li>no noticible input lag (a lot of my friends complain about this currently, though it might be a symptom of low FPS)</li></ul>
These changes make for a more enjoyable and smoother experience. Until these changes occur, this game is difficult to recommend to our friends, and we have A LOT of friends ready to jump on the next promising competitive FPS.
Well, if you could point out potential (specific) improvements I'm sure the devs would be very thankful?
Also, UWE does recognize this fact and makes progress towards this goal in nearly every patch (even if some look like 2 steps forwards and 1 step backwards). Part of the problem is that when UWE makes a significant performance increase, most servers negate the improvements by upping the playercount till it is just as laggy as before. 12 player servers used to be unplayable, now their common. Its progress, but I can certainly see how it might not look like it to some.
I suppose there's some truth to that, but now that servers are running 12-16 slots I don't think it'll be increased much further. People understandably just wanted to get out of those dreadful 6-8 slot situations, where no proper game could be played really.
True, but optimal playercount is mostly related to the map. 6-8 was pretty good for rockdown, 8-12 is decent on tram, 12-16 seems to work for summit. If UWE releases a larger map or a larger custom map becomes popular, then expect to see more servers striving for a playercount between 16-24.
Also, there is always the server owner who likes to push the boundaries (see current 20 playercount servers). Remember, UWE's max playercount goal for NS2 is 32. That's still a ways off, but I expect servers to mostly keep negating part of any performance increase until that max is reached.
However, we do need to be working on multiple things at a time, not simply performance. It's not simply about programming, but art, sound, balance, etc., so we must work on many things in parallel. Also, as we add new features, that changes our performance footprint radically.
Are you sure? I thought NS2 was using client-side hit registration. If so, it's not a wash if a crappy client machine is ticking below the server rate. That's a definite disadvantage, due to the difficulty of putting reticle on target at say, 12 fps, while someone else is ticking at 70.
Why do skulks feel hard to hit to some? Is it because players are bad/learning, are they lagged, are the animations weird, or something else?
Why do marines feel unresponsive to some? Is it because they're being slowed on hit, do the guns not ready fast enough after sprinting, or something else?
Why do people feel like aliens are losing all the time even if your numbers say they aren't? Are the results skewed by early alien wins, is turret spam a problem, is the cyst mechanic incomplete, or something else?
You guys are doing an awesome job, no doubt. I hope this suggestion isn't taken as an insult: You need to take lag out of the equation. Making "plays smooth" the number one priority for any team member who is qualified to choose one over the other will drive success more than any new weapon, map, or mechanic. I know that a texture artist isn't going to do double duty as an engine programmer. But is the driving principle right now truly Stability First? Without a good feel on the client, it is difficult for testers to make informed suggestions about gameplay.
I am convinced that as someone who had a much slower computer last week in the same build, the players aren't even playing the same game. The game that I can play now really makes sense.
We would love to, but that is SOOO much easier said then done. There are any number of different optimizations to do, and that we are doing. And those effect other things in the game, create more issues, and need extensive testing. Sometimes it turns out something that we thought would make a big difference ends up not being as noticeable of a lag reduction as we thought it would be, and vice versa. Generally a lot of playtesting and server performance tests need to be done to check out the changes. There's also a lot of time spent just thinking of what areas could be tackled to improve the server performance and how to go about it.
Even with all of the programmers fully focused on addressing the lag, which is not ideal anyway because that's not where everyone's strengths are, it would still be quite some time before we'd be able to get rid of lag from the equation. Whole systems have been reworked, which again is something that doesn't usually make sense to have more then one person messing with. To have the game essentially come to a stand still development wise during that lag killing phase is just not practical.
And, as Charlie mentioned, adding new features changes performance significantly. For example, cysts had an impact on server lag. We could spend months eliminating lag, and then with a few new features lag the game back up. The larger the amount of final game features are in, the better overall picture we'll have about performance. If more elements we know are causing performance issues, the easier it is to make optimizations that include all of those instances, rather then leaving them out because we didn't even know they would be a problem down the road.
Its pretty standard mode of development for the game industry, and if anything, we are actually addressing a lot of the performance issues sooner in the development process then many games do, because we actually have a public release version of the game already.
--Cory
True, but I'm mostly thinking that people with lesser hardware are more or less evenly distributed between the sides for most matches. Also, I suspect that a lot of NS2 beta players who are actually playing the game on a regular basis have good enough client specs that the limiting factor is the server. I don't know anyone who would voluntarily play a game where they only get 5-15 fps for any significant period of time.
Even with all of the programmers fully focused on addressing the lag, which is not ideal anyway because that's not where everyone's strengths are, it would still be quite some time before we'd be able to get rid of lag from the equation. Whole systems have been reworked, which again is something that doesn't usually make sense to have more then one person messing with. To have the game essentially come to a stand still development wise during that lag killing phase is just not practical.
And, as Charlie mentioned, adding new features changes performance significantly. For example, cysts had an impact on server lag. We could spend months eliminating lag, and then with a few new features lag the game back up. The larger the amount of final game features are in, the better overall picture we'll have about performance. If more elements we know are causing performance issues, the easier it is to make optimizations that include all of those instances, rather then leaving them out because we didn't even know they would be a problem down the road.
Its pretty standard mode of development for the game industry, and if anything, we are actually addressing a lot of the performance issues sooner in the development process then many games do, because we actually have a public release version of the game already.
--Cory<!--QuoteEnd--></div><!--QuoteEEnd-->
Cory wouldn't it be easier to ask someone to do it for you? You could pay them to do the job or at least pay them so they can give you guidance? There's many professional programmers which is specialize in server/client performance.
I think that it would be easier, but doing it yourself gives a better understanding. I also think the "pay" part would cause problems.
My motivation here isn't to say how things should be done, but more to raise awareness for both players and devs. Regardless of the internal development situation, the number one hindrance for me (and others, by the sound of it) to spreading your game is that playability is extremely sensitive to lag and hardware. Without a healthy server and great hardware, frustration abounds. Hits on stationary objects don't register. Animations don't play. Everything feels like it's covered in glue.
<b>Suggestion: Post recommended specs for testing.</b> Tell people, "You need x, y, and z to play the game like it's intended right now." Tell them exactly what to expect when they use anything less. (e.g. "Yes, we know that you can't hit anything right now with subpar hardware.") Revise it often. Rational people only want to calibrate their expectations.
Sigh. No.
It does seem kind of odd there are only 1 or 2 guys working on the C++-side of things, and for the sake of argument I'm assuming they're not Abrash'. NS2 is supposedly an indie-title, but has the production-value approaching that of a triple-A.
Works like this: Server saves the last few ticks --> Client tells the server that he just shot in direction xy and the game-time when he did it --> Server rolls back to the game-time from the client and checks if something was hit.