fanaticThis post has been edited.Join Date: 2003-07-23Member: 18377Members, Constellation, Squad Five Blue
edited April 2014
I think there are two ways of looking at it.
From a developer's point of view: Only UWE knows if it was worth it or not. That's an assessment of their bottom-line and player retention numbers.
From a player's point of view: I played both games and followed development of NS2 from the beginning. Put simply, Spark has been a huge disappointment. Now, I have no idea if any other engines would've fared better on performance, I'm no programmer. All I know is that for the entire lifetime of NS2, both client and server performance have been depressingly poor and many important features are still missing, more than two years after NS2 was released. Even something as simple as demo recording, a prerequisite for both reliably catching cheaters and making good fragmovies (compare NS2's 0 fragmovies to NS1's 20+) is still missing. Mass spectating (ala HLTV) will likely never been seen, and first person spectate is still essentially broken. If some other engine had been used, then at least some of these features would have been more or less ready for use, maybe even all of them. Regardless of whether or not other engines could've outperformed Spark, that utility cost remains. So, no, as a player, I don't think it was worth it.
Yeah the fact is a lot of dev resources were spent on the engine rather than being spent elsewhere on improve game play features and content, to make things worse the engine isn't that special, we never got proper dynamic infestation which was promised and supposedly one of the reasons they went and made their own.
Imagine if they had used source or unreal engine, the game could be something great today, though it's not all bad, i guess it served its purpose but it took a long time and wasted a lot or resources for little real benefit, at least from a players point of view.
A big reason why the devs decided to make their own engine, i at least assume, is due to the lack of dynamic lighting in other engines around the foetel stages of NS2 development. Pretty much no commercially available engine at the time of NS2s dev start could push out more than 3 dynamic lights per scene. For a game based around changing lighting conditions, shading, and shadows... that meant creating a new engine.
I think a lot of people may forget what the game engines of 2007 or so looked like and were capable of... basically where all lighting was created through time consuming baking and did not allow for complex shading or real time anything. This applies to UE3, Source, etc...
The only engine capable of something similar to NS2s goals would probably be idTech4, and as far as I know, was not really available.
A big reason why the devs decided to make their own engine, i at least assume, is due to the lack of dynamic lighting in other engines around the foetel stages of NS2 development. Pretty much no commercially available engine at the time of NS2s dev start could push out more than 3 dynamic lights per scene. For a game based around changing lighting conditions, shading, and shadows... that meant creating a new engine.
I think a lot of people may forget what the game engines of 2007 or so looked like and were capable of... basically where all lighting was created through time consuming baking and did not allow for complex shading or real time anything. This applies to UE3, Source, etc...
The only engine capable of something similar to NS2s goals would probably be idTech4, and as far as I know, was not really available.
While I doubt that it would have been available for UWE, the Cryengine 2 came out 2007 with Crysis 1, and it is capable of dynamic lightning in realtime, at least now but since you could experience day/night shifts in Crysis 1 I think it could do this from the start.
A big reason why the devs decided to make their own engine, i at least assume, is due to the lack of dynamic lighting in other engines around the foetel stages of NS2 development. Pretty much no commercially available engine at the time of NS2s dev start could push out more than 3 dynamic lights per scene. For a game based around changing lighting conditions, shading, and shadows... that meant creating a new engine.
I think a lot of people may forget what the game engines of 2007 or so looked like and were capable of... basically where all lighting was created through time consuming baking and did not allow for complex shading or real time anything. This applies to UE3, Source, etc...
The only engine capable of something similar to NS2s goals would probably be idTech4, and as far as I know, was not really available.
While I doubt that it would have been available for UWE, the Cryengine 2 came out 2007 with Crysis 1, and it is capable of dynamic lightning in realtime, at least now but since you could experience day/night shifts in Crysis 1 I think it could do this from the start.
Cryengine was still limited to about 3-4 dynamic lights per scene (the engine was forward rendered). Unlike the 100s which spark uses. But yes, Cryengine was ahead of its time by about 5 years
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
But there was no sdk at the time and the average cost to license game engines then was around $500k iirc? .. There's no way UWE had that on hand at that time.
And this is assuming the engine would have been fit for their needs (which by the sound of it, it wouldn't have been)
fanaticThis post has been edited.Join Date: 2003-07-23Member: 18377Members, Constellation, Squad Five Blue
edited April 2014
The Source SDK was released in 2004. As for the cost, that would depend on the terms of the contract. I very much doubt it would've been remotely close to the number you just suggested. The very, very early stages of NS2 development was actually done on Source -- see the first dynamic infestation concept video, for example. I think they decided not to use it because they thought they had to license the tools as well (which it turns out they didn't, so that's a shame), but my memory is a bit fuzzy on that point.
Soul_RiderMod BeanJoin Date: 2004-06-19Member: 29388Members, Constellation, Squad Five Blue
@fanatic - While I believe there was early work done in source, I am 90% sure that the dynamic infestation concept video was actually an early version of Spark which had BSP support.
I can't say I feel confident about my recollection of small details like this from seven years back, but the implications of these posts seems clear enough. There would be no need to add lua support to Source if they were already rendering things in their own engine.
This engine is something I've been working on for years and I use it whenever I want to prototype something in 3D. Since I've designed it to be really easy-to-use, it's generally much faster for me to do prototyping (which requires a lot of iteration) using it rather than a big complex engine like Source. I originally added Half-Life 2 BSP rendering to prototype the dynamic infestation ideas shown in this blog entry.
fanaticThis post has been edited.Join Date: 2003-07-23Member: 18377Members, Constellation, Squad Five Blue
edited April 2014
Ok, yeah, my bad. That actually clears things up a bit. At that point they were set on using Source to power the game (as evidenced by all the comments about Source in the blog posts I linked and the blog post you quoted from) but they wanted to use their own tools, which explains why the thought of having to license the Source SDK tools would be an issue for them.
I think it's a good engine. Looks amazing with all the shading, shadowing and stuff, it looks really amazing and nearly photorealistic in many places. And considering the complexity of the game, it's doing pretty well during gameplay with minimal quirks. I'm happy that they haven't used Source Engine. That thing has the worst netcode in the whole world with horrifying interp problems. I've spent more time fiddling with net settings in those games than actually playing. because even when i found one setting to work this hour, it was totally not working the next one. It's stupid. Where NS2's engine seems quite consistent.
I just wish they'd work a bit more on graphics optimizations, preferably adding AMD Mantle and utilizing more CPU cores.
I think it's a good engine. Looks amazing with all the shading, shadowing and stuff, it looks really amazing and nearly photorealistic in many places. And considering the complexity of the game, it's doing pretty well during gameplay with minimal quirks. I'm happy that they haven't used Source Engine. That thing has the worst netcode in the whole world with horrifying interp problems. I've spent more time fiddling with net settings in those games than actually playing. because even when i found one setting to work this hour, it was totally not working the next one. It's stupid. Where NS2's engine seems quite consistent.
I just wish they'd work a bit more on graphics optimizations, preferably adding AMD Mantle and utilizing more CPU cores.
I've never had problems with Source engine based game, they always worked great, even with tweaked interpolation settings you could really dial in your network performance if you knew how to tweak it. In comparison, NS2 always gives me problems. Random stutters, micro freezing, warping, rubber banding, etc. I find it laughable when you say it's "consistent" when literally almost every server suffers from bad tick rate drops when the mid-late game roll around including the official servers.
I couldn't ever get a single Source game to work right or vene more, consistently. Actually no, L4D2 worked fine, but mostly because you're not fighting other players but NPC zombies. Everything else, all CS, TF etc games work like shit. Hits registration is all over the place 99% of the time. So frustrating i stopped playing it because you can't even make a damn learning curve if aim is not the same even friggin twice. Ergo, no consistency. Sure, NS2 has problems here and there, but considering servers usually have 20+ players with all the building stuff going on massive maps, it's remarkable how well it works compared to 100% static world in lets say CS:GO with half sized maps. I only had noticeable problems on modded NS2 servers with 40+ players, but the blame goes on 3rd party mods and the fact the game wasn't really designed for so many players. But still, in like 75% of time, even these work perfectly fine.
NarfwakJoin Date: 2002-11-02Member: 5258Members, Super Administrators, Forum Admins, NS1 Playtester, Playtest Lead, Forum Moderators, Constellation, NS2 Playtester, Squad Five Blue, Reinforced - Supporter, Reinforced - Silver, Reinforced - Gold, Reinforced - Diamond, Reinforced - Shadow, Subnautica PT Lead, NS2 Community Developer
I've seen very few developers other than Valve use the Source engine well. Titanfall seems to be the big exception to the pattern. UWE is mad skilled, but given the cost of licensing the engine and tools and the hurdles of bending it to your will I think NS2 would have turned out a lot worse if they had stayed with that plan.
From a playtester perspective I cannot overstate the value of the game code being open source when it comes to finding bugs and for community members like SamusDroid to be able to fix all kinds of stuff. It also makes iterating on builds while testing and debugging much faster, and although the build process has slowed down considerably over time it's still fast enough that we've been able to test two or more builds in a single playtest session - an invaluable fact given that the PT team is composed of unpaid volunteers with limited time!
I couldn't ever get a single Source game to work right or vene more, consistently. Actually no, L4D2 worked fine, but mostly because you're not fighting other players but NPC zombies. Everything else, all CS, TF etc games work like shit. Hits registration is all over the place 99% of the time. So frustrating i stopped playing it because you can't even make a damn learning curve if aim is not the same even friggin twice. Ergo, no consistency.
You must be a very rare case. It's a bit odd that you think NS2 that runs at 30 tick rate is better over source games that run as high as 120. The higher tick rate will give you superior network performance, more updates = more precise. I also don't think the competitive scene for games like counter strike would be so huge if what you say is true.
Sure, NS2 has problems here and there, but considering servers usually have 20+ players with all the building stuff going on massive maps, it's remarkable how well it works compared to 100% static world in lets say CS:GO with half sized maps.
Not really. I would agree if the performance actually did well in the scenario you described, but it doesn't. You want to know what's really remarkable? BF3. It has gigantic maps, 64 players, tanks, jets, helicopters, explosions, etc - all while maintaining 60fps+ an entire match at 1080p on my rig that struggles with NS2 at times on 720p.
I only had noticeable problems on modded NS2 servers with 40+ players, but the blame goes on 3rd party mods and the fact the game wasn't really designed for so many players. But still, in like 75% of time, even these work perfectly fine.
No, don't try and put the blame solely on server operators. The blame goes on the engine and the server operators using under powered hardware, but what do you expect? NS2 rapes normal server hardware and you need these crazy overclocked hosting packages. If you want to blame the mods, why have I seen the official UWE servers that don't run any mods drop <10 tick on more than one occasion? The good servers are just less prone and you won't notice it much if you aren't very observant, but it still happens under heavy load.
NS2 is the only game I've ever played on PC where the performance is so hit or miss. You have no idea how a match is going to be performance wise since the performance label is absolutely worthless. Half of the servers can barely handle the mid game properly, let alone the miserable late game that follows.
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 April 2014
Nah, without Lua (or similar language) this game would not exist.
Granted, the performance implications of Lua was "slightly" more severe than anticipated, but since the Lua-JIT patch most of the performance problems are actually in the Spark itself; the time taken to ensure that all physical bodies are animated correctly dominates the performance problem areas - and those are pure C++ codepaths.
If UWE had tried to write NS2 in C/C++ the game would not have been released. They would have had to do everything themselves in the very unfriendly C/C++ environment, and considering how much of the NS2 code has been written or debugged by community members, they would have run out of money LONG before they could have released it.
I again state that the only problem with NS2 development was that it was unorganized and messy. Maybe also the fact it was 100% LUA (only some needs to be LUA). This was UWE's first standalone game, give them a break.
Unlike today, we had very few options for other engines. We would have had to pay hundreds of thousands of dollars or more in an up front cost, and/or given up a large % of our revenue on release. So, either we could just flat out not afford it, or we would have run out of money sooner, or we'd still be paying out large chunks of money to another company to this day.
The Source engine was a nightmare of a codebase to muddle through. Yes, it was a very tested engine compared to Spark and performed pretty well on low end systems, but to push it do the things we were wanting to do with NS2 would still have required a lot of effort with an unforseen outcome. It is far easier for a programmer to fix bugs and implement features in an engine they've built, especially when its been built specific to the game they are working on.
We did think we would be able to license the tools, even though we weren't licensing the engine, which did end up being a costly mistake. We ended up having to spend a lot of time on stuff like our own level editor instead of being able to license Hammer, and if we'd known that at the time it might...might...have swayed our decision, as that stuff sucked up a very large amount of development time.
What Matso said "without Lua (or similar language) this game would not exist." is largely true. I'm not sure I'd go as far as to say it wouldn't exist, but it would definitely have been even more challenging to finish, and to have as fleshed out a game as we ended up with. One of our main gameplay programmers, Andi, who has programmed a huge chunk of the game, would not have been able to work for UWE if not for LUA.
Yes, developing our own engine was a big learning experience. Yes, there were smarter and more efficient ways in which we could have developed, that we now understand. When we started, there was no usable version of Lua-JIT, for example, so if we'd been able to code specifically with that in mind from day 1, it would have made a big difference.
But that is where we are at today. We have our own engine, that we know the ins and outs of. We know the benefits and we know the weaknesses. Moving forward, we are developing Spark 2.0 for one our next projects, and we can now apply everything we've learned over the years, and rework and rewrite it to take advantage of our experience and the new tools that exist now. That puts us in a pretty good place, moving forward, so yes, I would say it was worth it.
I've not found the Spark engine to be particularly impressive at the time of NS2's release and given the quality of competing engines, I wouldn't use it to develop a game today. That said, I'm still blown away by what UWE have accomplished. A couple of guys working on a half-life mod to go on to hire staff and develop a full blown engine ahead of their first release during a period dominated by the big developers and prohibitive costs for indie developers to get into the market. All the pitfalls and mistakes along the way only make that accomplishment even greater.
Sure, the engine doesn't quite rival the capabilities of those created by the long established companies but it looks good and allowed UWE to begin creating the game they wanted without compromising on their vision, or waiting around for years for the technology to advance. I can't think of a single indie developer to do what UWE did, to the level of quality that they've achieved on their first game release.
The question remains as to what happens to Spark now. There are other engines out there that are a better overall choice for developers if they're willing to give some of their future revenue up. I'm sure Spark has it's strengths over others in niche areas, which may warrant using it in future development of specific titles, but, as a player, I'll be taking a long look at it's performance before buying another game built on it...Unless it's NS3, and it's in another 5-10 years. The IP still has a lot of potential and I'd love to see UWE revisit it with some more experience and a much improved (or alternative) engine under their belt at a later date.
Was Spark a mistake for UWE? I couldn't possibly answer that, but given the options available 7-10 years ago, I'd argue that creating their own engine was the right choice for NS2. I'd never have taken the risk personally and instead would have compromised my vision for the game by fitting into the constraints of a third party engine as well as offer up a chunk of all my earnings for the privilege, assuming I didn't give up entirely when I realised I couldn't make the game I wanted too.
....The Source engine was a nightmare of a codebase to muddle through. Yes, it was a very tested engine compared to Spark and performed pretty well on low end systems, but to push it do the things we were wanting to do with NS2 would still have required a lot of effort with unforeseen consequences...
....The Source engine was a nightmare of a codebase to muddle through. Yes, it was a very tested engine compared to Spark and performed pretty well on low end systems, but to push it do the things we were wanting to do with NS2 would still have required a lot of effort with unforeseen consequences...
Comments
From a developer's point of view: Only UWE knows if it was worth it or not. That's an assessment of their bottom-line and player retention numbers.
From a player's point of view: I played both games and followed development of NS2 from the beginning. Put simply, Spark has been a huge disappointment. Now, I have no idea if any other engines would've fared better on performance, I'm no programmer. All I know is that for the entire lifetime of NS2, both client and server performance have been depressingly poor and many important features are still missing, more than two years after NS2 was released. Even something as simple as demo recording, a prerequisite for both reliably catching cheaters and making good fragmovies (compare NS2's 0 fragmovies to NS1's 20+) is still missing. Mass spectating (ala HLTV) will likely never been seen, and first person spectate is still essentially broken. If some other engine had been used, then at least some of these features would have been more or less ready for use, maybe even all of them. Regardless of whether or not other engines could've outperformed Spark, that utility cost remains. So, no, as a player, I don't think it was worth it.
Imagine if they had used source or unreal engine, the game could be something great today, though it's not all bad, i guess it served its purpose but it took a long time and wasted a lot or resources for little real benefit, at least from a players point of view.
I think a lot of people may forget what the game engines of 2007 or so looked like and were capable of... basically where all lighting was created through time consuming baking and did not allow for complex shading or real time anything. This applies to UE3, Source, etc...
The only engine capable of something similar to NS2s goals would probably be idTech4, and as far as I know, was not really available.
While I doubt that it would have been available for UWE, the Cryengine 2 came out 2007 with Crysis 1, and it is capable of dynamic lightning in realtime, at least now but since you could experience day/night shifts in Crysis 1 I think it could do this from the start.
That's amazingly accurate. I bought it some weeks ago and had zero complaints about graphics
And this is assuming the engine would have been fit for their needs (which by the sound of it, it wouldn't have been)
This is a later post where they talk about adding LUA support to the Source engine: http://unknownworlds.com/ns2/prosumers-and-natural-selection-2-2/
I can't say I feel confident about my recollection of small details like this from seven years back, but the implications of these posts seems clear enough. There would be no need to add lua support to Source if they were already rendering things in their own engine.
I just wish they'd work a bit more on graphics optimizations, preferably adding AMD Mantle and utilizing more CPU cores.
I've never had problems with Source engine based game, they always worked great, even with tweaked interpolation settings you could really dial in your network performance if you knew how to tweak it. In comparison, NS2 always gives me problems. Random stutters, micro freezing, warping, rubber banding, etc. I find it laughable when you say it's "consistent" when literally almost every server suffers from bad tick rate drops when the mid-late game roll around including the official servers.
From a playtester perspective I cannot overstate the value of the game code being open source when it comes to finding bugs and for community members like SamusDroid to be able to fix all kinds of stuff. It also makes iterating on builds while testing and debugging much faster, and although the build process has slowed down considerably over time it's still fast enough that we've been able to test two or more builds in a single playtest session - an invaluable fact given that the PT team is composed of unpaid volunteers with limited time!
You must be a very rare case. It's a bit odd that you think NS2 that runs at 30 tick rate is better over source games that run as high as 120. The higher tick rate will give you superior network performance, more updates = more precise. I also don't think the competitive scene for games like counter strike would be so huge if what you say is true.
Not really. I would agree if the performance actually did well in the scenario you described, but it doesn't. You want to know what's really remarkable? BF3. It has gigantic maps, 64 players, tanks, jets, helicopters, explosions, etc - all while maintaining 60fps+ an entire match at 1080p on my rig that struggles with NS2 at times on 720p.
No, don't try and put the blame solely on server operators. The blame goes on the engine and the server operators using under powered hardware, but what do you expect? NS2 rapes normal server hardware and you need these crazy overclocked hosting packages. If you want to blame the mods, why have I seen the official UWE servers that don't run any mods drop <10 tick on more than one occasion? The good servers are just less prone and you won't notice it much if you aren't very observant, but it still happens under heavy load.
NS2 is the only game I've ever played on PC where the performance is so hit or miss. You have no idea how a match is going to be performance wise since the performance label is absolutely worthless. Half of the servers can barely handle the mid game properly, let alone the miserable late game that follows.
Granted, the performance implications of Lua was "slightly" more severe than anticipated, but since the Lua-JIT patch most of the performance problems are actually in the Spark itself; the time taken to ensure that all physical bodies are animated correctly dominates the performance problem areas - and those are pure C++ codepaths.
If UWE had tried to write NS2 in C/C++ the game would not have been released. They would have had to do everything themselves in the very unfriendly C/C++ environment, and considering how much of the NS2 code has been written or debugged by community members, they would have run out of money LONG before they could have released it.
It is far from 100% LUA.
Trolololo coolitic
Unlike today, we had very few options for other engines. We would have had to pay hundreds of thousands of dollars or more in an up front cost, and/or given up a large % of our revenue on release. So, either we could just flat out not afford it, or we would have run out of money sooner, or we'd still be paying out large chunks of money to another company to this day.
The Source engine was a nightmare of a codebase to muddle through. Yes, it was a very tested engine compared to Spark and performed pretty well on low end systems, but to push it do the things we were wanting to do with NS2 would still have required a lot of effort with an unforseen outcome. It is far easier for a programmer to fix bugs and implement features in an engine they've built, especially when its been built specific to the game they are working on.
We did think we would be able to license the tools, even though we weren't licensing the engine, which did end up being a costly mistake. We ended up having to spend a lot of time on stuff like our own level editor instead of being able to license Hammer, and if we'd known that at the time it might...might...have swayed our decision, as that stuff sucked up a very large amount of development time.
What Matso said "without Lua (or similar language) this game would not exist." is largely true. I'm not sure I'd go as far as to say it wouldn't exist, but it would definitely have been even more challenging to finish, and to have as fleshed out a game as we ended up with. One of our main gameplay programmers, Andi, who has programmed a huge chunk of the game, would not have been able to work for UWE if not for LUA.
Yes, developing our own engine was a big learning experience. Yes, there were smarter and more efficient ways in which we could have developed, that we now understand. When we started, there was no usable version of Lua-JIT, for example, so if we'd been able to code specifically with that in mind from day 1, it would have made a big difference.
But that is where we are at today. We have our own engine, that we know the ins and outs of. We know the benefits and we know the weaknesses. Moving forward, we are developing Spark 2.0 for one our next projects, and we can now apply everything we've learned over the years, and rework and rewrite it to take advantage of our experience and the new tools that exist now. That puts us in a pretty good place, moving forward, so yes, I would say it was worth it.
Second:
If that doesn't sound good I don't know what does.
Sure, the engine doesn't quite rival the capabilities of those created by the long established companies but it looks good and allowed UWE to begin creating the game they wanted without compromising on their vision, or waiting around for years for the technology to advance. I can't think of a single indie developer to do what UWE did, to the level of quality that they've achieved on their first game release.
The question remains as to what happens to Spark now. There are other engines out there that are a better overall choice for developers if they're willing to give some of their future revenue up. I'm sure Spark has it's strengths over others in niche areas, which may warrant using it in future development of specific titles, but, as a player, I'll be taking a long look at it's performance before buying another game built on it...Unless it's NS3, and it's in another 5-10 years. The IP still has a lot of potential and I'd love to see UWE revisit it with some more experience and a much improved (or alternative) engine under their belt at a later date.
Was Spark a mistake for UWE? I couldn't possibly answer that, but given the options available 7-10 years ago, I'd argue that creating their own engine was the right choice for NS2. I'd never have taken the risk personally and instead would have compromised my vision for the game by fitting into the constraints of a third party engine as well as offer up a chunk of all my earnings for the privilege, assuming I didn't give up entirely when I realised I couldn't make the game I wanted too.
We need an ^ @cory ^ as well :P
NS3 Confirmed! lol
I'm pretty sure I saw someone post something about it being 100% LUA. But I guess he was wrong then?