Game Discussions Star Citizen Discussion Thread v12

Anyway...I melted the BMM and got my Corsair out of buyback with the store credits. The Corsair is set for flyable sometime during 3.18 and it looks functional enough...plus it has a decent loaner in the meantime. I had enough left in store credit to buy a Raft, a space forklift and a few new paint jobs for my ships...left $100 bucks in the store in case I need it :)
Took me a bit to find out what a BMM is. Still no idea what a banu is. I figure it's what @Viajero does to me when I'm not a good boy.
 
Some familiar tropes nestled amongst the May Glassdoor reviews.

Imbibe with the usual shovel of salt. (Although I do find longer, wayward posts like these more believable than average ;))



9 May 2022 — Software Engineer, Manchester, England

Pros:

Fairly good career development. You'll get promoted if you're working effectively, though perhaps promotions are handed out a little too easily.

Cons:

We have some directors saying that the work doesn't look good enough, even though it had previously been signed-off on. Seems counter-productive to sign something off, and then a month later say you've changed your mind.



10 May 2022 — Junior Programmer, Manchester, England

Cons:

Look I get it, the team I was on was under an insane amount of pressure, but I fail to understand why I was hired knowing I had just come out of University with no prior experience in commercial C++ programming.

Advice to management:

Don't hire graduates and not provide a support structure.



Not sure if this otherwise positive Dec ‘21 one was spotted at the time, but it also covers some familiar themes in some detail:

12 Dec 2021 — Anonymous Employee, Wilmslow, England

Cons:

-Production practices and quarterly release cycle make it very difficult to actually develop the game
-Workflows and toolchain are horrible, you get bogged down in stream management for hours each day; tools are slow and are not scaling well with the project
-Build stability is terrible, new crashes and bugs every day which makes it very hard to be productive
-Director-level leadership and communcation is not strong, and the project is so large, leading to inconsistent design direction and poor transparency with other teams. Some features are on their second or third iteration because the design direction changed
-The engine / core-tech is constantly being updated, which breaks existing features all the time
 
Some familiar tropes nestled amongst the May Glassdoor reviews.

Imbibe with the usual shovel of salt. (Although I do find longer, wayward posts like these more believable than average ;))



9 May 2022 — Software Engineer, Manchester, England

Pros:



Cons:






10 May 2022 — Junior Programmer, Manchester, England

Cons:



Advice to management:






Not sure if this otherwise positive Dec ‘21 one was spotted at the time, but it also covers some familiar themes in some detail:

12 Dec 2021 — Anonymous Employee, Wilmslow, England

Cons:

You'll get promoted if you're working effectively, though perhaps promotions are handed out a little too easily.

Congratulations! You can now shave! Senior developer material!
 
Ouch…

So I did myself a favor and forgot this game existed for about 2 years, haven't played since 3.8 and was wondering how playable the game is and what kind of new stuff has happened since then.

Is salvage working now? Ai blades/NPC crew? Can ships & their components be disabled during combat or they still just blow up at 0 health? Do servers run better or still lots of disconnects out of the blue? Was ship armor finally added? Are there any big events going on or just the regular individual missions?

1.5 out of 6 I guess…
 
Went on for a bit of SC earlier this morning...I might as well not have bothered. The server back end is so borked I couldn't recall a ship without getting an immediate parking violation with the ship being stored in the time it took me to get in the elevator to the hangar. On the one occasion I did manage to get to a ship, it kept randomly powering on and off and wouldn't let me escape the hangar.

Besides all that, the friends list isn't populating...even if it does, you can't jump in a server with friends or join up at the main menu and do a party launch.

I tried server after server, region after region and they're mostly the same with a variety of back end stuff completely broken or only randomly functional. Some folk theorising in global chat on the various servers I briefly visited were putting all those issues down to the Invictus free-fly...but I strongly suspect that 3.17's issues are more deeply rooted than a few newbies clogging the servers up. We'll soon find out when the free-fly ends later today 😐
 
Last edited:
Some folk theorising in global chat on the various servers I briefly visited were putting all those issues down to the Invictus free-fly...but I strongly suspect that 3.17's issues are more deeply rooted than a few newbies clogging the servers up. We'll soon find out when the free-fly ends later today 😐
Yeah tend to hear that quite a bit. It's the new guys ruining it for the rest of us!!

But more seriously, I look at new games that have hundreds of thousands of players trying to get on and the games work, albeit some queuing and general lag and then look at SC's PU/PTU which struggles with a smallish influx of additional players (by comparison). It just shows the design of their backend stuff does not scale at all.
It really seems like they are using a fail method of development. Build this, see where it fails, start again, build that, see where it fails, start again etc
 
Yeah tend to hear that quite a bit. It's the new guys ruining it for the rest of us!!

But more seriously, I look at new games that have hundreds of thousands of players trying to get on and the games work, albeit some queuing and general lag and then look at SC's PU/PTU which struggles with a smallish influx of additional players (by comparison). It just shows the design of their backend stuff does not scale at all.
It really seems like they are using a fail method of development. Build this, see where it fails, start again, build that, see where it fails, start again etc

Failure as a service?
 
In my case, Fail to Pay. Had again a bit of flying around, enjoyed some great sights, fell through a great city crossing a great quantum door, took a couple quantum lifts, took countless other normal lifts, took a mind numbingly large number of lifts. Got blasted off the sky a couple times, medical services stole my gun, bought some fancy bits of armor and a new gun only to have my inventory disappear, together with my ammo, and found myself on a barren moon with an empty gun and leg bags that couldn't hold stuff (and still no main inventory), brought back a box from said barren moon to Area 18 only to have the mission not updating, and staying stuck back there with my glowing blinking box. Nice assets, still not worth paying them though.

A nice farewell sunset, until next free fly.
ScreenShot-2022-05-29_18-10-21-048.jpg
 
I truffled through the AI Q&A. Here are some highlights:

Source: https://www.youtube.com/watch?v=JKhBrT6fh7w


On the mess hall still being a mess: "You don't get the crockery that you want" [2m42s]

Andrew Davies [Product Owner of the AI Feature Team]:

I'm also a senior developer and I've been working on, uh, getting our, uh, chow line, um, AI up to scratch, and this is something I worked on before but it's kind of lapsed for a little bit, but we're, uh, we're kind of gonna get running again so I can pass it along to the content team and… The chow line is a, a system that, um, allows our, is a social bit of a social AI, that allows our guys to go and grab a tray from the mess hall, grab some crockery, and then go and use lots of different machines that each produce bits of food and drink. And then go and sit down at tables and eat them and drink them and do that to their hearts content.

Um, and it's a system that has required lots of little different bits working in a way that not, if I if I'd worked on a previous game, you probably will have that all done in one canned animation. But because this is, uh, Star Citizen and we like to make things and we like to go that extra level with lots of detail. Each of these machines is a separate thing, and that means there's lots of places where things can go wrong, like you don't get the the crockery that you want, or something else, or let's say there's a, uh, whilst you're you're uh getting your food suddenly there's, uh, an alert over the tannoy and all of those characters have to drop what they're doing and leave their plates and trays and go off and and man their battle stations. So it's dealing with all of those things, like how do you come back to that, and, uh, deal with that. So that's, um, kind of taking lots of stuff that we've already worked on, and then putting it all together to create like a really complex system, and with all the complexities, all the bugs that come with that [laughs], um so, that's basically what I do.
...like I was thinking about in the chow line, and the, the mess hall. If there's loads of people having to go off, um, or the player leaving loads of trays behind, then there's going to have to be a janitor character who comes in and picks up those trays and takes them, takes them back, so we don't get a mess hall that's just full with, with trays. [30m30s]


The mocap studio being is for pick-ups and SQ42 sequels... [12m14s]

Lando:

...where we'll be able to do pickups, like, we just did, like Stewart just showed you, for, for, uh just any time we need to add more NPCs into, into places and make the place feel more alive, uh, to do pickups for Squadron 42. Things that we, you know, realize, oh, we need a bridge for this or that or whatever. Or even, when it comes time to it, production for, you know, additional, you know, sequels and stuff for Squadron 42 and stuff like that...


They don't seem to know how the engineering modules will work yet [16m30s]

Dan Baker [Senior Designer]:
And inside there you're going to have things like fuses, or whatever, other kinds of sub-components that might be relative to that that component. And they're going to affect the statistics of the the component in some way, most likely just like, you know, if certain fuses, um, were broken you would probably expect the thing to not work as well as if it had three fully functioning fuses.

[19m+]

  • The NPCs are supposed to be able to do this too. But for now it's SQ42 focused.
  • 'Assignments' - Currently NPC just stays there doing all their inspection actions because it looks good.


On getting NPCs to work with the existing world and assets [31m+]

Dan Baker [Senior Designer]:

Say you have something on top of the Gladius. The NPC needs to be able to get up there, you know, which means that they need to understand that, in order to get up there, I need to grab the ladders move them over to the Gladius. [Francesco Roccucci: And then the ladders are trolleys, you need to be able to use the trollies]. And then when they get up there you need to make sure that they've got the animations to enter, and it's all the correct metrics, which isn't always the case, you know. So then you've got to go back to the artist and say this isn't to metric, and usually you have to design a metric first, and then get them to change their, their baby into something that you've you've written down, months or sometimes years later.


Lando managing NPC crew expectations... [34m23s]

To properly manage, uh, expectations we are shooting for the moon with this thing. And, and, and, when, when it is complete, and when it is, uh, fully developed, it'll be a thing unlike any other.
NPC multicrew is not a 'soon' thing.


- [2m42s] Andrew Davies? [He says 'Davis'] - Product Owner of the AI Feature Team [so Producer?], and 'chow line' developer. On the mess hall still being a mess:

"And I'm also a senior developer and I've been working on, uh, getting our, uh, chow line, um, AI up to scratch, and this is something I worked on before but it's kind of lapsed for a little bit, but we're, uh, we're kind of gonna get running again so I can pass it along to the content team and, the chow line is a, a system that, um, allows our, is a social bit of a social AI, that allows our guys to go and grab a tray from the mess hall, grab some crockery, and then go and use lots of different machines that each produce bits of food and drink. And then go and sit down at tables and eat them and drink them and do that to their hearts content. Um, and it's a system that was required lots of little different bits working in a way that not, if I if i'd worked on a previous game, you probably will have that all done in one canned animation. But because this is, uh, Star Citizen as soon as we like to make things as we like to go that extra level with lots of detail. Each of these machines is a separate thing, and that means there's lots of places where things can go wrong ,like you don't get the the crockery that you want, or something else, or let's say there's a, uh, whilst you're you're uh getting your food suddenly there's, uh, an alert over the tannoy and all of those characters have to drop what they're doing and leave their plates and trays and go off and and man their battle stations. So it's dealing with all of those things, like how do you come back to that, and, uh, deal with that. So that's, um, kind of taking lots of stuff that we've already worked on, and then putting it all together to create like a really complex system, and with all the complexities, all the bugs that come with that [laughs], um so, that's basically what I do.

Lando: And a lot of your focus these days is on Squadron 42 but most uh much of that work will make its way into the PU once it's proven out.


- [9m20s] Stuart Charlton [3D Animator]: On the flow graph of using a usable. "By doing that we maintain a sense of realism, of performance, and just adding more to that expanding universe". References bartender ;). He's in charge "So i'm in charge of making sure that performance is authentic and that's why animation is such an important aspect of of the AI work.". [9m20s:
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=560s
]


- Talks about pick up mocap shoots for things like AI on trains looking bored etc

- [12m14s] - Lando on the mocap studio being for pick-ups and SQ42 sequels... "...where we'll be able to do pickups, like, we just did, like stewart just showed you, for, for, uh just any time we need to add more NPCs into into places and make the place feel more alive, uh, to do pickups for Squadron 42. Things that we, you know, realize oh, we need a bridge for this or that or whatever. Or even when it comes time to it production for you know additional you know sequels and stuff for Squadron 42 and stuff like that..."
[
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=12m12s
]

[16m30s] - They don't know how the engineering modules will work yet, but they have an animation: Dan Baker [Senior Designer]: "Things like, uh, shield generators, quantum drives, relays, whatever it ends up being. And inside there you're going to have things like fuses, or whatever, other kinds of sub-components that might be relative to that that component. And they're going to affect the statistics of the the component in some way, most likely just like, you know, if certain fuses, um, were broken you would probably expect the thing to not work as well as if it had three fully functioning fuses."
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=16m30s


[19m+] - The NPCs are supposed to be able to do this too. But for now it's SQ42 focused.
'Assignments' - Currently NPC just stays there doing all their inspection actions because it looks good. (Warbles on about 'usables' and heirarchies of the same. Will use torch if it has one etc).

[24m15s+] - Hireable NPCs, what's needed?...

[26m30s+] - Maybe engineers will get low morale if you assign them as a janitor. Claims they have the backend in place for this type of stuff. Just needs the usables to be usable etc...
- Persistence is in place he thinks [?]

[30m30s] Andrew Davies on mess halls again: "like I was thinking about in the chow line, and the, the mess hall. If there's loads of people having to go off, um, or the player leaving loads of trays behind, then there's going to have to be a janitor character who comes in and picks up those trays and takes them, takes them back, so we don't get a mess hall that's just full with, with trays."


[31m+] On getting NPCs to work with the existing world: Lando on ship variety being a difficulty for NPC transit / animations. Designer worried about dependencies. (IE where to get tool from). Designer is working on deck crew - NPCs repairing Gladius etc (cosmetically). So realises 'this ship isn't to metric' / get ship designers to change their baby. Sometimes has to invent the metric. Get animations and/or tech for them to use trolley, to move ladder, to climb up to bit that needs repair etc....

"Say you have something on top of the Gladius. The NPC needs to be able to get up there, you know, which means that they need to understand that, in order to get up there, I need to grab the ladders move them over to the Gladius. [Francesco Roccucci: And then the ladders are trolleys {IE you need AI to be able to use trollies}]. And then when they get up there you need to make sure that they've got the animations to enter, and it's all the correct metrics, which isn't always the case, you know. So then you've got to go back to the artist and say this isn't to metric, and usually you have to design a metric first, and then get them to change their, their baby into something that you've you've written down, months or sometimes years later."
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=32m22s


- [34m23s] - Lando managing expectations...
"To properly manage, uh, expectations we are shooting for the moon with this thing. And, and, and, when, when it is complete, and when it is, uh, fully developed, it'll be a thing unlike any other."
- "NPC multicrew is not a 'soon' thing"

[35m54s] - Static Nav Mesh - "When that's in, later this year"
  • Only for certain POIs to start
  • Hopes you can kite NPC wherever, in theory
  • Working in internal build now

[39m30s] - AI reinforcements / drop ships:
  • 'In theory' can unload vehicles etc as well as people
  • They can't support ships that use elevators or ladders yet. Only the backdoor / ramps by the sound.

[43m50s] - Traits system
  • Gung ho character vs cowardly one etc. Will affect how they will target others he reckons. (Sounds like this might be aspirational though). [It's chow line AI guy]
  • Factions who are amateurs vs pros. Animations to that effect. [SQ42 related]

[46m] - AI using healing on themselves etc
  • Planned for SQ42
  • Non-lethal takedowns planned. (And revival of same by NPCs)

[47m] - Allowing AI to loot dead players?
- More unknown. Just talks about plans for distributing loot. (Although suggests AI might know where to loot?)

[49m30s] - The illusion of AI not knowing
- Want them to check stuff they already know, re ammo drop, or vending machine, or whatever. (Or having a 'dynamic' conv while waiting in queue for vending machine...)

[51m] - Performance issues
- They have a lot of tasks to try and improve performance for the 'dedicated server'. Client isn't such a big deal as it's only streaming in what's local. If you have 50 players in heavy locations (Lorville etc), the server is going to be down at 5-7FPS.
-Try to talk with directors/designers and get them to limit the number of entities in the game.
  • 'Usable Idle' will play instead if NPC waiting on server / deciding what to do next.
  • 1,500 characters acting at the same time is a lot for one server.
  • "AI itself is definitely not that expensive, but it's everything that comes with the character, right. That is animation, updated animation, the rendering, uh, you know all the calculation for, I don't know, if they're piloting a ship with all the items."
  • They fixed an issue where combat AI wasn't turning to shoot not because the server command hadn't arrived, but because of a bug preventing them turning.
  • Trying to optimise the ray casts / queries they do every frame.
  • SM will help. Live now you can feel the difference when the server is 'fresh' etc. The experience is then 'closer to what regular games do right now'.
  • Not unusual to see stuff work on single player and test servers but fail in PU etc.

- [56m30s] - AI priority 'LODs' [?]

Fran: "At the end of the last year some things that we changed is the way we give priority to some of the calculation. So we give calculation to the combat more than the social, and to some things at the social a bit above the combat, and, uh. You know I was finishing to experiment with some other ways of key priority with a lot of different LODs you know, and this is the stuff that we have. We have already two version LODs on behaviors and movement and stuff like this, but we don't have yet and like, for example, what they want to do is a nice way of balancing importance of the request based on LOD:

...if you switch LOD right, if the player moves around and fights with, uh, you know, let's say, uh, out with Andrew right? And, and Dan is super far, maybe he was close to me, but I left him right? So if now it takes a little bit of more time for him to find a new cover spot, I prefer that Andrew's cover spot is quicker to give to him. [IE - AI close to you get higher update priority?]

Lightning Round: [59m]
  • AI grenades need a 'refresh' currently
  • Animal NPCs? Every animator wants to work on them. Some SQ42 WIP / upcoming work for fauna.
  • AI turrets work for NPCs etc, but just needs ways to equip etc for players.
  • AI feelings? AI will express moods. Frustration is one emotion. And the mood should be consistent etc.

  • Biggest challenge right now? [1h1m30s]
  • 'Set up'. Making sure everyone's doing the same thing the same kind of way in a big studio. Frustrating when different approaches & trying to debug.
  • Getting Mocap time, but that's getting fixed...
  • 'Intertwined systems'. For example looking at the transit systems, need to have a full understanding of how it works, need dedicated time with the team etc. Then new systems have an impact and can impact the old systems etc. Loadouts and inventories, the lockers = example of trying to prepare something with least impact possible [?]
  • Performance. But a good challenge. Especially with limited resources.
 
Last edited:
I truffled through the AI Q&A. Here are some highlights:

Source: https://www.youtube.com/watch?v=JKhBrT6fh7w


On the mess hall still being a mess: "You don't get the crockery that you want" [2m42s]





The mocap studio being is for pick-ups and SQ42 sequels... [12m14s]




They don't seem to know how the engineering modules will work yet [16m30s]




[19m+]

  • The NPCs are supposed to be able to do this too. But for now it's SQ42 focused.
  • 'Assignments' - Currently NPC just stays there doing all their inspection actions because it looks good.


On getting NPCs to work with the existing world and assets [31m+]




Lando managing NPC crew expectations... [34m23s]






- [2m42s] Andrew Davies? [He says 'Davis'] - Product Owner of the AI Feature Team [so Producer?], and 'chow line' developer. On the mess hall still being a mess:

"And I'm also a senior developer and I've been working on, uh, getting our, uh, chow line, um, AI up to scratch, and this is something I worked on before but it's kind of lapsed for a little bit, but we're, uh, we're kind of gonna get running again so I can pass it along to the content team and, the chow line is a, a system that, um, allows our, is a social bit of a social AI, that allows our guys to go and grab a tray from the mess hall, grab some crockery, and then go and use lots of different machines that each produce bits of food and drink. And then go and sit down at tables and eat them and drink them and do that to their hearts content. Um, and it's a system that was required lots of little different bits working in a way that not, if I if i'd worked on a previous game, you probably will have that all done in one canned animation. But because this is, uh, Star Citizen as soon as we like to make things as we like to go that extra level with lots of detail. Each of these machines is a separate thing, and that means there's lots of places where things can go wrong ,like you don't get the the crockery that you want, or something else, or let's say there's a, uh, whilst you're you're uh getting your food suddenly there's, uh, an alert over the tannoy and all of those characters have to drop what they're doing and leave their plates and trays and go off and and man their battle stations. So it's dealing with all of those things, like how do you come back to that, and, uh, deal with that. So that's, um, kind of taking lots of stuff that we've already worked on, and then putting it all together to create like a really complex system, and with all the complexities, all the bugs that come with that [laughs], um so, that's basically what I do.

Lando: And a lot of your focus these days is on Squadron 42 but most uh much of that work will make its way into the PU once it's proven out.


- [9m20s] Stuart Charlton [3D Animator]: On the flow graph of using a usable. "By doing that we maintain a sense of realism, of performance, and just adding more to that expanding universe". References bartender ;). He's in charge "So i'm in charge of making sure that performance is authentic and that's why animation is such an important aspect of of the AI work.". [9m20s:
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=560s
]


- Talks about pick up mocap shoots for things like AI on trains looking bored etc

- [12m14s] - Lando on the mocap studio being for pick-ups and SQ42 sequels... "...where we'll be able to do pickups, like, we just did, like stewart just showed you, for, for, uh just any time we need to add more NPCs into into places and make the place feel more alive, uh, to do pickups for Squadron 42. Things that we, you know, realize oh, we need a bridge for this or that or whatever. Or even when it comes time to it production for you know additional you know sequels and stuff for Squadron 42 and stuff like that..."
[
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=12m12s
]

[16m30s] - They don't know how the engineering modules will work yet, but they have an animation: Dan Baker [Senior Designer]: "Things like, uh, shield generators, quantum drives, relays, whatever it ends up being. And inside there you're going to have things like fuses, or whatever, other kinds of sub-components that might be relative to that that component. And they're going to affect the statistics of the the component in some way, most likely just like, you know, if certain fuses, um, were broken you would probably expect the thing to not work as well as if it had three fully functioning fuses."
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=16m30s


[19m+] - The NPCs are supposed to be able to do this too. But for now it's SQ42 focused.
'Assignments' - Currently NPC just stays there doing all their inspection actions because it looks good. (Warbles on about 'usables' and heirarchies of the same. Will use torch if it has one etc).

[24m15s+] - Hireable NPCs, what's needed?...

[26m30s+] - Maybe engineers will get low morale if you assign them as a janitor. Claims they have the backend in place for this type of stuff. Just needs the usables to be usable etc...
- Persistence is in place he thinks [?]

[30m30s] Andrew Davies on mess halls again: "like I was thinking about in the chow line, and the, the mess hall. If there's loads of people having to go off, um, or the player leaving loads of trays behind, then there's going to have to be a janitor character who comes in and picks up those trays and takes them, takes them back, so we don't get a mess hall that's just full with, with trays."


[31m+] On getting NPCs to work with the existing world: Lando on ship variety being a difficulty for NPC transit / animations. Designer worried about dependencies. (IE where to get tool from). Designer is working on deck crew - NPCs repairing Gladius etc (cosmetically). So realises 'this ship isn't to metric' / get ship designers to change their baby. Sometimes has to invent the metric. Get animations and/or tech for them to use trolley, to move ladder, to climb up to bit that needs repair etc....

"Say you have something on top of the Gladius. The NPC needs to be able to get up there, you know, which means that they need to understand that, in order to get up there, I need to grab the ladders move them over to the Gladius. [Francesco Roccucci: And then the ladders are trolleys {IE you need AI to be able to use trollies}]. And then when they get up there you need to make sure that they've got the animations to enter, and it's all the correct metrics, which isn't always the case, you know. So then you've got to go back to the artist and say this isn't to metric, and usually you have to design a metric first, and then get them to change their, their baby into something that you've you've written down, months or sometimes years later."
Source: https://www.youtube.com/watch?v=JKhBrT6fh7w&t=32m22s


- [34m23s] - Lando managing expectations...
"To properly manage, uh, expectations we are shooting for the moon with this thing. And, and, and, when, when it is complete, and when it is, uh, fully developed, it'll be a thing unlike any other."
- "NPC multicrew is not a 'soon' thing"

[35m54s] - Static Nav Mesh - "When that's in, later this year"
  • Only for certain POIs to start
  • Hopes you can kite NPC wherever, in theory
  • Working in internal build now

[39m30s] - AI reinforcements / drop ships:
  • 'In theory' can unload vehicles etc as well as people
  • They can't support ships that use elevators or ladders yet. Only the backdoor / ramps by the sound.

[43m50s] - Traits system
  • Gung ho character vs cowardly one etc. Will affect how they will target others he reckons. (Sounds like this might be aspirational though). [It's chow line AI guy]
  • Factions who are amateurs vs pros. Animations to that effect. [SQ42 related]

[46m] - AI using healing on themselves etc
  • Planned for SQ42
  • Non-lethal takedowns planned. (And revival of same by NPCs)

[47m] - Allowing AI to loot dead players?
- More unknown. Just talks about plans for distributing loot. (Although suggests AI might know where to loot?)

[49m30s] - The illusion of AI not knowing
- Want them to check stuff they already know, re ammo drop, or vending machine, or whatever. (Or having a 'dynamic' conv while waiting in queue for vending machine...)

[51m] - Performance issues
- They have a lot of tasks to try and improve performance for the 'dedicated server'. Client isn't such a big deal as it's only streaming in what's local. If you have 50 players in heavy locations (Lorville etc), the server is going to be down at 5-7FPS.
-Try to talk with directors/designers and get them to limit the number of entities in the game.
  • 'Usable Idle' will play instead if NPC waiting on server / deciding what to do next.
  • 1,500 characters acting at the same time is a lot for one server.
  • "AI itself is definitely not that expensive, but it's everything that comes with the character, right. That is animation, updated animation, the rendering, uh, you know all the calculation for, I don't know, if they're piloting a ship with all the items."
  • They fixed an issue where combat AI wasn't turning to shoot not because the server command hadn't arrived, but because of a bug preventing them turning.
  • Trying to optimise the ray casts / queries they do every frame.
  • SM will help. Live now you can feel the difference when the server is 'fresh' etc. The experience is then 'closer to what regular games do right now'.
  • Not unusual to see stuff work on single player and test servers but fail in PU etc.

- [56m30s] - AI priority 'LODs' [?]

Fran: "At the end of the last year some things that we changed is the way we give priority to some of the calculation. So we give calculation to the combat more than the social, and to some things at the social a bit above the combat, and, uh. You know I was finishing to experiment with some other ways of key priority with a lot of different LODs you know, and this is the stuff that we have. We have already two version LODs on behaviors and movement and stuff like this, but we don't have yet and like, for example, what they want to do is a nice way of balancing importance of the request based on LOD:

...if you switch LOD right, if the player moves around and fights with, uh, you know, let's say, uh, out with Andrew right? And, and Dan is super far, maybe he was close to me, but I left him right? So if now it takes a little bit of more time for him to find a new cover spot, I prefer that Andrew's cover spot is quicker to give to him. [IE - AI close to you get higher update priority?]

Lightning Round: [59m]
  • AI grenades need a 'refresh' currently
  • Animal NPCs? Every animator wants to work on them. Some SQ42 WIP / upcoming work for fauna.
  • AI turrets work for NPCs etc, but just needs ways to equip etc for players.
  • AI feelings? AI will express moods. Frustration is one emotion. And the mood should be consistent etc.

  • Biggest challenge right now? [1h1m30s]
  • 'Set up'. Making sure everyone's doing the same thing the same kind of way in a big studio. Frustrating when different approaches & trying to debug.
  • Getting Mocap time, but that's getting fixed...
  • 'Intertwined systems'. For example looking at the transit systems, need to have a full understanding of how it works, need dedicated time with the team etc. Then new systems have an impact and can impact the old systems etc. Loadouts and inventories, the lockers = example of trying to prepare something with least impact possible [?]
  • Performance. But a good challenge. Especially with limited resources.

Its kind of funny, because the reaction to this sort of stuff gives you more or less the same line from both faithful and skeptics.

"this is why its taking so long and costing so much money"

But whereas the faithful present it as a positive thing the skeptics present it as a negative thing.

Well, i think we skeptics usually put strong adjectives in there as well :p
 
Back
Top Bottom