I don't understand why everyone loves source so much. I'm not a specialist regarding engines, but personally I think the source engine is horrible.
Performs well, has raw input, proper acceleration, easily moddable if your game is large enough to get attention from sourcemod developers, servers likely wouldn't need to be running 4.8ghz only to drop frames lategame, we would be able to see further than 40units. Another thing to note is if UWE didn't have to put people in charge of creating and maintaining the engine or dev tools, that would either be less employees (more money?), or more people working on the game itself. That's all I can think of off the top of my head.
edit: can't believe I forgot... while source's netcode isn't perfect, it is far better than what we're dealing with now.
keep in mind i'm just playing devils advocate, i'm positive they had good reasons to make their own engine (besides infestation)
While sources net code is better... Source games tend to have the worst hit boxes known to man. Just imagine trying to shoot a skulk when his hitbox is actually 3 feet ahead of the visual character?
i would like to see videos backing up your claim though. not only have I played comp on the source engine for years and never run into these issues myself, but I just asked a small handful of top players in different source engine games and they agree that it's just the go-to excuse for when people die. perhaps your settings are wrong? who knows, although I don't play or know anyone who has played l4d2 competitively, maybe the hitboxes get more shitty with faster moving objects (thinking of hunters/smoker tongue/w.e)
Netcode in source is quite similar to gldsource. Overall Source seems to handle higher speeds less gracefully than gldsource, but its still better than Spark (mainly because of adjustable rates).
Also, I have never seen bad hitboxes overall in any Source games - I have seen a couple questionable ones (I'm looking at you Dino-D-Day), but that was with the smallest dinosaur and guns with recoil so meh, and those were generally a specific model with the issue.
IronHorseDeveloper, QA Manager, Technical Support & contributorJoin Date: 2010-05-08Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
edited March 2014
@xDragon
Sigh, I knew I should have said more than two words, lol..
No net_stats showing makes that video useless, as I'm sure you know.
I can reproduce that video 20 times over (in almost any fps) if I recreate a poor network scenario - thus it needs to be ruled out.
Here's my typical offer:
Provide a clear 720p 30fps or better video with net_Stats on which shows hit detection or registration issues... and I will shower you with gifts and fame.
Here's my typical offer:
Provide a clear 720p 30fps or better video with net_Stats on which shows hit detection or registration issues... and I will shower you with gifts and fame.
That video was recorded on a good server, so I can almost completely be sure that it was 30 ticks and no lag at that point, I recorded enough to show that there was no correction of anything which would indicate any lag or tickrate issues. The point is that a video is completely useless in reproducing the issues, since there are 5000 things that could be happening in that game which you have no way of knowing about. Without a way to accurately test things on a live server, there's just no good way to test these things.
IronHorseDeveloper, QA Manager, Technical Support & contributorJoin Date: 2010-05-08Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
Yeah the only thing that could have occurred in that shot network wise that would be nice to rule out, would be a lag spike for the person recording - it wouldn't have effected your personal movement or your client predicting the enemy's position - if it was truly a spike and only momentary.
Something important to note about that video was how the rounds landed on the trailing portion of the model, which is consistent with past videos I've seen. Could be that predicted entity positions are off for the client, or what I've always suspected, (but told no it wasn't the case) that the hitboxes are sometimes too far ahead of the actual model. You can see this with a debug command, but I've been told not to rely on that because it is not accurately represented. *shrug *
@currently I'll look at it when I get to a computer again (on mobile)
matsoMaster of PatchesJoin Date: 2002-11-05Member: 7000Members, Forum Moderators, NS2 Developer, Constellation, NS2 Playtester, Squad Five Blue, Squad Five Silver, Squad Five Gold, Reinforced - Shadow, NS2 Community Developer
edited March 2014
When using multic-core rendering, the main thread hands off a copy of the data the render thread needs, but the collision world is not copied over. So when the rendering pipeline renders the collision world, it uses data that has been updated by the main thread while it is working on the next frame.
In other words, xDragon is correct.
Personally, I would have loved to implement a perfect server AND client side recording - that would have allowed for bug replication while also allowing for record/replay/broadcast/kill-cam shots etc... oh well, one can dream.
When using multic-core rendering, the main thread hands off a copy of the data the render thread needs, but the collision world is not copied over. So when the rendering pipeline renders the collision world, it uses data that has been updated by the main thread while it is working on the next thread.
In other words, xDragon is correct.
Personally, I would have loved to implement a perfect server AND client side recording - that would have allowed for bug replication while also allowing for record/replay/broadcast/kill-cam shots etc... oh well, one can dream.
When using multic-core rendering, the main thread hands off a copy of the data the render thread needs, but the collision world is not copied over. So when the rendering pipeline renders the collision world, it uses data that has been updated by the main thread while it is working on the next thread.
In other words, xDragon is correct.
Personally, I would have loved to implement a perfect server AND client side recording - that would have allowed for bug replication while also allowing for record/replay/broadcast/kill-cam shots etc... oh well, one can dream.
Yes. There were many reasons they dropped work on the Source engine version early, as indicated in the dev blogs/video updates back around 2008. Not sure why this argument keeps popping up over and over again.
Because most people don't lurk on the forums enough to understand that the ENGINE is fast and very good, its the LUA is slow and causes the game to be CPU heavy. (this is my understanding from reading these forums for the last year or so)
If the LUA were to be removed and the game rewritten in a compiled language, most performance issues would disappear (so I hear, anyway)
If only it were as easy as you say. The problem is that 100% of the games code was LUA. This was a mistake on their part (and they admit it). It made the game highly moddable but much of the code didn't need LUA even if the game was to be highly moddable.
However, the biggest problem was lack of experience. NS2 devs now know how to develop a game (mostly anyways) and if/when they make NS3 we should have a real treat.
Also @IronHorse, people have been abusing the flag button again.
Cannon_FodderAUSBrisbane, AUJoin Date: 2013-06-23Member: 185664Members, Squad Five Blue, Squad Five Silver, Reinforced - Shadow
I would rather the devs get subnautica out, get some cash flow, and start on NS3. NS2 is probably going to be as good as it gets in terms of optimization (I guess we can tinker at the edges). But I feel it is time better spent to start fresh and not make the same mistakes. It might seem cold and pragmatic, but I DO want UWE to continue to be successful, and feel putting too much resources in NS2 isn't really the way forward. As to Kick starter - I will donate for sure (I missed out on the Black Armor last time, and won't this time :-).
Having said all that, I do wish NS2 will continue to be worked on part time - the time for big patches and content over haul is over. And I think our great community mappers can add to the mix of maps (there are some really cool ones out in the workshop now). I just wish more servers will run them). It would be nice if UWE could run some of these on the official servers map rotations.
Cannon_FodderAUSBrisbane, AUJoin Date: 2013-06-23Member: 185664Members, Squad Five Blue, Squad Five Silver, Reinforced - Shadow
There is something about NS that has made me a fan boy. My Steam games (I have over 100) play time goes: NS2 558hrs, the next closest one on a recent install is Saints Row 3 (20 hrs). Nothing else comes even close to the amount of time I spend playing NS2. So what I mean by start NS3 is to start development for it (this could take 2-3 years), not to get it out the door next year.
L4D is the main culprit. Worst hitboxes I have ever seen. CS:S have odd hitboxes as well but not nearly as bad.
CSS has MASSIVE hitboxes, and they're almost never on the model - and when the game was forced to 66 tic it got even worse
CSGO is 'ok' on a high tic server, however if the server has low server side fps (which most do, it's probably costly to have several hundred 900fps servers) and you can disable clock correction - or lower it to give more fluid gameplay
The downside is people w/ higher ping will begin to feel 'worse' much like original quake
and tons of people disagree with me but I think games were waaay better with less lag compensation than the excess we have today.. seriously people are in my matches with 100+ ping playing normally/top fragging and we have to suffer through them 'appearing' on our screen.. its just silly
I dunno if they would make an ns3, but if they did and they got it to perform amazing it'd probably be a top online played fps.. i have sooooooooo many friends who would be playing but they can't.
I hope their next underwater title does well, unity engine is pretty awesome from what I've seen especially with large areas
Would love to see an NS3 in a few years from now. Although I really like this genre of game that is created, the rts/fps hybrid type, wouldn't mind a slight change like environment/teams instead of the the revitalization of NS into NS2. Which was AWESOME btw, I was worried in the early beta that the flower wouldn't bear fruit, but the fruit it beared was so sweet that I ate that peach for HOURS.
edit: realized my cake day is coming up.... 10 years. damn, thats a short list of games anyone can likely say they enjoyed for that period of time. Starcraft is the one that comes to mind.
Comments
Performs well, has raw input, proper acceleration, easily moddable if your game is large enough to get attention from sourcemod developers, servers likely wouldn't need to be running 4.8ghz only to drop frames lategame, we would be able to see further than 40units. Another thing to note is if UWE didn't have to put people in charge of creating and maintaining the engine or dev tools, that would either be less employees (more money?), or more people working on the game itself. That's all I can think of off the top of my head.
edit: can't believe I forgot... while source's netcode isn't perfect, it is far better than what we're dealing with now.
keep in mind i'm just playing devils advocate, i'm positive they had good reasons to make their own engine (besides infestation)
i would like to see videos backing up your claim though. not only have I played comp on the source engine for years and never run into these issues myself, but I just asked a small handful of top players in different source engine games and they agree that it's just the go-to excuse for when people die. perhaps your settings are wrong? who knows, although I don't play or know anyone who has played l4d2 competitively, maybe the hitboxes get more shitty with faster moving objects (thinking of hunters/smoker tongue/w.e)
Also, I have never seen bad hitboxes overall in any Source games - I have seen a couple questionable ones (I'm looking at you Dino-D-Day), but that was with the smallest dinosaur and guns with recoil so meh, and those were generally a specific model with the issue.
The bug definitely exists, but whatever it is that causes it is so random ive never been able to accurately track down any kind of relationship.
Sigh, I knew I should have said more than two words, lol..
No net_stats showing makes that video useless, as I'm sure you know.
I can reproduce that video 20 times over (in almost any fps) if I recreate a poor network scenario - thus it needs to be ruled out.
Here's my typical offer:
Provide a clear 720p 30fps or better video with net_Stats on which shows hit detection or registration issues... and I will shower you with gifts and fame.
9:02 What is my prize?
http://youtu.be/F1nrPTBVgVg?t=8m50s
Something important to note about that video was how the rounds landed on the trailing portion of the model, which is consistent with past videos I've seen. Could be that predicted entity positions are off for the client, or what I've always suspected, (but told no it wasn't the case) that the hitboxes are sometimes too far ahead of the actual model. You can see this with a debug command, but I've been told not to rely on that because it is not accurately represented. *shrug *
@currently I'll look at it when I get to a computer again (on mobile)
In other words, xDragon is correct.
Personally, I would have loved to implement a perfect server AND client side recording - that would have allowed for bug replication while also allowing for record/replay/broadcast/kill-cam shots etc... oh well, one can dream.
There is always kickstarter. Reencorded anyone?
I'd fund immediately.
If only it were as easy as you say. The problem is that 100% of the games code was LUA. This was a mistake on their part (and they admit it). It made the game highly moddable but much of the code didn't need LUA even if the game was to be highly moddable.
However, the biggest problem was lack of experience. NS2 devs now know how to develop a game (mostly anyways) and if/when they make NS3 we should have a real treat.
Also @IronHorse, people have been abusing the flag button again.
Never said it was easy, if it was, I'm sure they would have done it already :P
I do still have hopes for more game optimization (particularly server side)
Having said all that, I do wish NS2 will continue to be worked on part time - the time for big patches and content over haul is over. And I think our great community mappers can add to the mix of maps (there are some really cool ones out in the workshop now). I just wish more servers will run them). It would be nice if UWE could run some of these on the official servers map rotations.
CSS has MASSIVE hitboxes, and they're almost never on the model - and when the game was forced to 66 tic it got even worse
CSGO is 'ok' on a high tic server, however if the server has low server side fps (which most do, it's probably costly to have several hundred 900fps servers) and you can disable clock correction - or lower it to give more fluid gameplay
The downside is people w/ higher ping will begin to feel 'worse' much like original quake
and tons of people disagree with me but I think games were waaay better with less lag compensation than the excess we have today.. seriously people are in my matches with 100+ ping playing normally/top fragging and we have to suffer through them 'appearing' on our screen.. its just silly
I hope their next underwater title does well, unity engine is pretty awesome from what I've seen especially with large areas
edit: realized my cake day is coming up.... 10 years. damn, thats a short list of games anyone can likely say they enjoyed for that period of time. Starcraft is the one that comes to mind.