NS3 will be when the player in Subnautica finally reaches the depths where the Karaa live... unleashing a sudden explosion of Zenoforms that rapidly take over Planet 4546B
NS3 will be when the player in Subnautica finally reaches the depths where the Karaa live... unleashing a sudden explosion of Zenoforms that rapidly take over Planet 4546B
Maybe Subnautica is the prequal to NS? That would be sweet. The Subnautica player eventually discovers the Kharaa threat and its a race against time to fashion a communications tower to warn the human race before the kharaa absorb every life form.
Maybe Subnautica is the prequal to NS? That would be sweet. The Subnautica player eventually discovers the Kharaa threat and its a race against time to fashion a communications tower to warn the human race before the kharaa absorb every life form.
I have not played subnautica but this was very interesting! Ns3 would really benefit from a story stemming off of this. Hell It would be cool if NS3 was a continuation of Subnautica but a prequal to NS1. Like what happened after the virus / bacteria was brought off planet, prototype weapons, the chaos of all the Kharaa spreading. Like a more narrative story single player mode and obviously multiplayer. Co-op story mode would be boss in that setting. especially when the threat is still so mysterious, dangerous and still evolving.
For a start, I envision NS3 with larger scale conflicts involving multiple vehicles and similarly sized lifeforms.
I'd imagine 48 players or more. The current tech point concept would become a mobile unit for each team with a comm/operator. Each team could potentially field multiple mobile comm units with different builds and upgrades during the mid-late game. It would also revamp the res system in some very innovative manners (TBD). It would have a true dynamic terrain system for infestation and fortifications with power grid mechanics.
So due to the larger scale, the current NS gameplay model of tight melee vs ranged combat would have to evolve to incorporate new coordinated combat tactics like ballistic artillery strikes, flying transports, targeted missile swarms and other concepts similar to more modern shooters (see PlanetSide, HL2 mod Empires, Battlefield series, Evolve, Fortnite, Titanfall).
As far as the comm roles, I think the first comm would likely have the highest strategic perspective and responsibility, and the additional comm roles would act as sub-comms that would generally provide more direct battle interaction. This would make the comm roles more approachable over time to players. It's a difficult balance to make the comm important to team success but not make it discouraging to new players. By having sub-comm roles in games, this would provide a path to learning the roles.
I don't want any persistent unlocks or any of that type of shenanigans. Aim for a 25-35ish minute ideal round time.
NS2 is still so good though, I can only wish to see some improved iteration in the future.
I'm going to start from the tech end. Silicon CMOS is tapped out so you can't out-grow a performance problem. It has to perform well day one.
7 nm GPUs comming soon will have twice the transistor density (should be quadruple, but geometric scaling is failing), but consume only 30-40% less power per transistor (dennard scaling failed 2003); it's absolutely going to run hotter, probably clock a bit slower but you can cram more compute units in. Scaling with compute units is bad at best at this point for games. For the next 20 years graphics are going to get maybe 5-10 times faster. CPUs are going to get wider, maybe 16 or 32 cores will eventually be mainstream. Single threaded performance will probably drop a bit. Possibly heterogenous multicore CPUs (one massive core for best single threaded performance + 16 or whatever simpler cores for good multithreaded performance). 20 years is chosen carefully; it takes that long to go from a lab scale experiment to commercialisation without an apollo-programme-like effort; there's nothing on the horizon that will beat silicon CMOS right now, so it will take at least 20 years. After that all bets are off; it could be a golden age like the 90's again (~60% per year performance increase averaged over the entire decade + going from software rendering to 3D acceleration).
You can't really make a game like NS2 run faster by adding more cores. There's not much parallelism that you can extract from that. You need an independent problem that you can solve in parallel and since there isn't one, you'll have to invent one that can use those extra cores if present. Not quite sure what that is, but if you can invent something meaningful it's a free bonus since the cores otherwise would not be used. You really have to think long and hard what you load on these cores to make the game more fun; there's no point making an NS3 that uses 2-4 cores when 16x is standard. Some of this may solve itself when the rest of the industry shows what works and what doesn't.
I've never been a fan of real time lighting. NS looked so beautiful with such a shitty engine (quake was made for software rendering and beyoned ridiculously CPU bound in GL mode). It did so because it had pre-compiled lightmaps, a hack to overbrighten the game and CRTs had excellent black levels to go with the stark lighting choices. With real time lighting there's no global illumination, so it looks so sterile and hacked toghether with hand placed ambient lighting standing in for accurately precomputed reflected light.
Ambient occlusion is just a hack to try to claw some of that soft and beautiful ambient lighting back. Real-time global illumination is starting to become a thing but it would mean NS3 runs at poor FPS just to have as nice lighting as source did almost 2 decades before it (I'm assuming we're not getting NS3 any time soon). If it moves towards industry standard and UWE uses a standard engine NS3 will have it.
HMDs are relevant but VR is definitely not. What you can do in an HMD is simulate a display sitting in front of you, just like your desktop monitor does. That display can have depth like a "3D monitor" if you want it to, but a full VR experience would be projectile-vomit-inducing. Within the next generation or two of VR it's going to have extremely high resolution displays and eye tracking. The eye see's ~60 pixels/degree at the center and only 12/deg 5 degrees out. From there it drops to 1 pixel per degree in the periphery. Foveated rendering is going to give you a factor 100 or so performance savings; enabling 8000x8000 pixels per eye performance at 1440p hardware requirements, because you can cheat like nobody's business where you're not directly looking it can be made to look as good as doing 8000x8000 pixels per eye for real. I believe this is a huge killer app for HMDs; eye-tracking and foveated rendering on display sitting on your desktop is much, much harder; so I believe people will have HMDs for simulating screens of whatever arbitrary resolution or aspect ratio they need, with a really nice strobed OLED and they're only going to dip into full VR mode for cockpit-based games and other stuff where it's clearly the way to go.
Hopefully NS3 is late enough that it can take advantage without teething problems and not early enough to miss the train on foveated rendering. Foveated is the only place where I see a significant amount of graphical performance comming from; GPUs will keep gaining something like 20% year, gradually slowing towards 0.
Without foveated rendering NS3 looks only marginally better than NS2, like Crysis looks only marginally worse than a modern game, and really except for gameplay you shouldn't have bothered making it.
I hope it's not a large open world style game, because I just don't see how it could possibly work and still be close enough to NS to be a sequel and not a different game borrowing the NS title and settings for marketing reasons. Sequels have to innovate a little, but not so much that they are different games.
VR will also bring much needed focus on 3D audio; hopefully this will benefit everyone. It's beyond ridiculous that the best 3D audio cards released used the aureal A3D 2.0 chipset and API for windows 98SE and nothing after that has even come close. There were some hacked XP drivers for those cards but that's as far as it goes. Sound was a big thing in the original NS; in NS2 it's just so muddy there's just a slight hint of stereo separation and no hint of even a bad HRTF.
NS3 will need Valve or Blizzard to take over. Otherwise they will be absolutely no interest.
That would be good for UWE perhaps but it's not good for fans of either NS1 or NS2. It will be an entirely different game that I have no interest in playing. It's going to need to sell millions of copies if it's a huge AAA title and that necessarily means it will be designed for people who have no interest in NS1 or NS2 and everything that's unique and fun about the game will be sucked out and replaced with lootboxes, XP bars and similar horse hockey.
The Kharaa has evolved into living ships. Instead of a CC and Hive there are two massive command ships that must be defended. Commanders control these ships. There are fighter ships and also the main ships can be boarded by infantry/alien lifeforms. Players can choose whether to be either at any point. The ships can be repaired or have parts destroyed by players inside. Each command ship as a central reactor or something to blow up by infantry or can die from constant outside bombardment which is not repaired or stopped. Part of the game can be how many players are committed to inside and outside of the ships at a particular time. The command ships move about creating resource mines at locations (which can also be destroyed). Other side has to scout to find it, all hell breaks loose when the two ships come across each other. Something like that, I may have been playing a lot of Angels Fall First lately, boarding ships is great fun.
For a start, I envision NS3 with larger scale conflicts involving multiple vehicles and similarly sized lifeforms.
I'd imagine 48 players or more. The current tech point concept would become a mobile unit for each team with a comm/operator. Each team could potentially field multiple mobile comm units with different builds and upgrades during the mid-late game. It would also revamp the res system in some very innovative manners (TBD). It would have a true dynamic terrain system for infestation and fortifications with power grid mechanics.
So due to the larger scale, the current NS gameplay model of tight melee vs ranged combat would have to evolve to incorporate new coordinated combat tactics like ballistic artillery strikes, flying transports, targeted missile swarms and other concepts similar to more modern shooters (see PlanetSide, HL2 mod Empires, Battlefield series, Evolve, Fortnite, Titanfall).
As far as the comm roles, I think the first comm would likely have the highest strategic perspective and responsibility, and the additional comm roles would act as sub-comms that would generally provide more direct battle interaction. This would make the comm roles more approachable over time to players. It's a difficult balance to make the comm important to team success but not make it discouraging to new players. By having sub-comm roles in games, this would provide a path to learning the roles.
I don't want any persistent unlocks or any of that type of shenanigans. Aim for a 25-35ish minute ideal round time.
NS2 is still so good though, I can only wish to see some improved iteration in the future.
I wouldn't go as far as that with NS3. For me it would be closer to Halo 1/UT 2004. Just scale up a little bit and add some smaller vehicles.
I've never been a fan of real time lighting. NS looked so beautiful with such a shitty engine (quake was made for software rendering and beyoned ridiculously CPU bound in GL mode). It did so because it had pre-compiled lightmaps, a hack to overbrighten the game and CRTs had excellent black levels to go with the stark lighting choices. With real time lighting there's no global illumination, so it looks so sterile and hacked toghether with hand placed ambient lighting standing in for accurately precomputed reflected light.
These days you can do different types of mixed lighting where it bakes only the indirect lighting (global illumination) and the direct lighting is realtime. This would have same performance as NS2 + baked indirect lighting. Or, all close up lighting is realtime, far away is fully baked GI/shadows. That way you can get good performance by limiting the realtime light/shadow distance. The problem is obviously the map can't be dynamic, and realtime GI runs poorly even on todays machines.
Fully baked direct/indirect also means no shadows on objects, or some weird HL2 style projected shadow.
Ambient occlusion is just a hack to try to claw some of that soft and beautiful ambient lighting back. Real-time global illumination is starting to become a thing but it would mean NS3 runs at poor FPS just to have as nice lighting as source did almost 2 decades before it (I'm assuming we're not getting NS3 any time soon). If it moves towards industry standard and UWE uses a standard engine NS3 will have it.
You still need ambient occlusion when baking. Shadowed areas can still look flat without it. Modern realtime/baked AO techniques can limit themselves to shadowed areas only. NS2's just applies everywhere.
Comments
Maybe Subnautica is the prequal to NS? That would be sweet. The Subnautica player eventually discovers the Kharaa threat and its a race against time to fashion a communications tower to warn the human race before the kharaa absorb every life form.
I have not played subnautica but this was very interesting! Ns3 would really benefit from a story stemming off of this. Hell It would be cool if NS3 was a continuation of Subnautica but a prequal to NS1. Like what happened after the virus / bacteria was brought off planet, prototype weapons, the chaos of all the Kharaa spreading. Like a more narrative story single player mode and obviously multiplayer. Co-op story mode would be boss in that setting. especially when the threat is still so mysterious, dangerous and still evolving.
I'd imagine 48 players or more. The current tech point concept would become a mobile unit for each team with a comm/operator. Each team could potentially field multiple mobile comm units with different builds and upgrades during the mid-late game. It would also revamp the res system in some very innovative manners (TBD). It would have a true dynamic terrain system for infestation and fortifications with power grid mechanics.
So due to the larger scale, the current NS gameplay model of tight melee vs ranged combat would have to evolve to incorporate new coordinated combat tactics like ballistic artillery strikes, flying transports, targeted missile swarms and other concepts similar to more modern shooters (see PlanetSide, HL2 mod Empires, Battlefield series, Evolve, Fortnite, Titanfall).
As far as the comm roles, I think the first comm would likely have the highest strategic perspective and responsibility, and the additional comm roles would act as sub-comms that would generally provide more direct battle interaction. This would make the comm roles more approachable over time to players. It's a difficult balance to make the comm important to team success but not make it discouraging to new players. By having sub-comm roles in games, this would provide a path to learning the roles.
I don't want any persistent unlocks or any of that type of shenanigans. Aim for a 25-35ish minute ideal round time.
NS2 is still so good though, I can only wish to see some improved iteration in the future.
Lol that helicopter LMG
7 nm GPUs comming soon will have twice the transistor density (should be quadruple, but geometric scaling is failing), but consume only 30-40% less power per transistor (dennard scaling failed 2003); it's absolutely going to run hotter, probably clock a bit slower but you can cram more compute units in. Scaling with compute units is bad at best at this point for games. For the next 20 years graphics are going to get maybe 5-10 times faster. CPUs are going to get wider, maybe 16 or 32 cores will eventually be mainstream. Single threaded performance will probably drop a bit. Possibly heterogenous multicore CPUs (one massive core for best single threaded performance + 16 or whatever simpler cores for good multithreaded performance). 20 years is chosen carefully; it takes that long to go from a lab scale experiment to commercialisation without an apollo-programme-like effort; there's nothing on the horizon that will beat silicon CMOS right now, so it will take at least 20 years. After that all bets are off; it could be a golden age like the 90's again (~60% per year performance increase averaged over the entire decade + going from software rendering to 3D acceleration).
You can't really make a game like NS2 run faster by adding more cores. There's not much parallelism that you can extract from that. You need an independent problem that you can solve in parallel and since there isn't one, you'll have to invent one that can use those extra cores if present. Not quite sure what that is, but if you can invent something meaningful it's a free bonus since the cores otherwise would not be used. You really have to think long and hard what you load on these cores to make the game more fun; there's no point making an NS3 that uses 2-4 cores when 16x is standard. Some of this may solve itself when the rest of the industry shows what works and what doesn't.
I've never been a fan of real time lighting. NS looked so beautiful with such a shitty engine (quake was made for software rendering and beyoned ridiculously CPU bound in GL mode). It did so because it had pre-compiled lightmaps, a hack to overbrighten the game and CRTs had excellent black levels to go with the stark lighting choices. With real time lighting there's no global illumination, so it looks so sterile and hacked toghether with hand placed ambient lighting standing in for accurately precomputed reflected light.
Ambient occlusion is just a hack to try to claw some of that soft and beautiful ambient lighting back. Real-time global illumination is starting to become a thing but it would mean NS3 runs at poor FPS just to have as nice lighting as source did almost 2 decades before it (I'm assuming we're not getting NS3 any time soon). If it moves towards industry standard and UWE uses a standard engine NS3 will have it.
HMDs are relevant but VR is definitely not. What you can do in an HMD is simulate a display sitting in front of you, just like your desktop monitor does. That display can have depth like a "3D monitor" if you want it to, but a full VR experience would be projectile-vomit-inducing. Within the next generation or two of VR it's going to have extremely high resolution displays and eye tracking. The eye see's ~60 pixels/degree at the center and only 12/deg 5 degrees out. From there it drops to 1 pixel per degree in the periphery. Foveated rendering is going to give you a factor 100 or so performance savings; enabling 8000x8000 pixels per eye performance at 1440p hardware requirements, because you can cheat like nobody's business where you're not directly looking it can be made to look as good as doing 8000x8000 pixels per eye for real. I believe this is a huge killer app for HMDs; eye-tracking and foveated rendering on display sitting on your desktop is much, much harder; so I believe people will have HMDs for simulating screens of whatever arbitrary resolution or aspect ratio they need, with a really nice strobed OLED and they're only going to dip into full VR mode for cockpit-based games and other stuff where it's clearly the way to go.
Hopefully NS3 is late enough that it can take advantage without teething problems and not early enough to miss the train on foveated rendering. Foveated is the only place where I see a significant amount of graphical performance comming from; GPUs will keep gaining something like 20% year, gradually slowing towards 0.
Without foveated rendering NS3 looks only marginally better than NS2, like Crysis looks only marginally worse than a modern game, and really except for gameplay you shouldn't have bothered making it.
I hope it's not a large open world style game, because I just don't see how it could possibly work and still be close enough to NS to be a sequel and not a different game borrowing the NS title and settings for marketing reasons. Sequels have to innovate a little, but not so much that they are different games.
VR will also bring much needed focus on 3D audio; hopefully this will benefit everyone. It's beyond ridiculous that the best 3D audio cards released used the aureal A3D 2.0 chipset and API for windows 98SE and nothing after that has even come close. There were some hacked XP drivers for those cards but that's as far as it goes. Sound was a big thing in the original NS; in NS2 it's just so muddy there's just a slight hint of stereo separation and no hint of even a bad HRTF.
NS2 is basically NS1 but with the "we wanted to do that, now we can with our engine"
That would be good for UWE perhaps but it's not good for fans of either NS1 or NS2. It will be an entirely different game that I have no interest in playing. It's going to need to sell millions of copies if it's a huge AAA title and that necessarily means it will be designed for people who have no interest in NS1 or NS2 and everything that's unique and fun about the game will be sucked out and replaced with lootboxes, XP bars and similar horse hockey.
FUS RO... F4!!!
These days you can do different types of mixed lighting where it bakes only the indirect lighting (global illumination) and the direct lighting is realtime. This would have same performance as NS2 + baked indirect lighting. Or, all close up lighting is realtime, far away is fully baked GI/shadows. That way you can get good performance by limiting the realtime light/shadow distance. The problem is obviously the map can't be dynamic, and realtime GI runs poorly even on todays machines.
Fully baked direct/indirect also means no shadows on objects, or some weird HL2 style projected shadow.
You still need ambient occlusion when baking. Shadowed areas can still look flat without it. Modern realtime/baked AO techniques can limit themselves to shadowed areas only. NS2's just applies everywhere.