Well just walking around and doing stupid dance Vs doing complex things(fast movements,shooting shield,etc..) that should be transferred with no lag between 50 or more players are 2 very different things you know?That what you have in Arccorp I been seeing in Crytek engine before in the game known as Entropia Universe that is build in CE 2,when are we going to see bigger battles in AC that can handle 24+ players with no stability issue or lag problems?
The Star Citizen Thread v 3.0
- Thread starter Jenner
- Start date
- Status
Well just walking around and doing stupid dance Vs doing complex things(fast movements,shooting shield,etc..) that should be transferred with no lag between 50 or more players are 2 very different things you know?That what you have in Arccorp I been seeing in Crytek engine before in the game known as Entropia Universe that is build in CE 2,when are we going to see bigger battles in AC that can handle 24+ players with no stability issue or lag problems?
"Grabs a video of one older update on a match that faced several network issues."
>>
"Acts like the whole game and players have the same experience on the last updates."
lol
(Side of that 40 players on Arcorp handle very well, much less netcoded content, if the netcode revamp process wasn't happening they would not double the player counts. They won't put more players on AC not being only on netcode, it's also a small area, to know is how many will SC 2.0 allow)
- - - Updated - - -
The AC likely won't face number boosts until SC 2.0. And i'm most interested to see how many will 2.0 handle, because there's one incredible increase of netcode, from stations, multi-crew ships, Quantum Drive, a huge playable map, etc....
2.0 is the fire proof for the netcode, and i'm already predicting things won't be pretty by the time it releases, no matter how much you test, once things go massive, things happen!
Last edited:
Haha true. The last 5 matches I had were smooth as butter. So was Arccorp.
Fair enough. Show us a 16 player dog-fighting video in 1.1.6 then.
Last edited:
I've got to say that Mr Smarts source for his information about a book deal is impeccable
.
Last edited:
Show us a video then with these 40 players plus hundreds of NPC's moving around. Let's look how very well its handling then.
- - - Updated - - -
I've got to say that Mr Smarts source for his information about a book deal is impeccable
Well... she said in an interview that was trying to ask permission for CIG to make a book based on their universe. Some guys quoted that in a forum with the link to the interview.
So, why do you think that its false or invalid as source?
/facepalm
Last edited:
You're a bit outdated, the current version is 1.2 on stable, 1.3 on PTU. There is no more versions for AC as all development streams are now merged on the same build.
Well Nowak might better do that for you as i don't play much of AC, however found some cool things people are doing on 1.3, as boarding a spinning avenger: (also the WIP interior physics is not finished that's why the lack of gravity inside currently)
[video=youtube;he0q3TdVLmM]https://www.youtube.com/watch?v=he0q3TdVLmM[/video]
I guess you don't follow Star Citizen because version 1.1.6 was 2 months ago. The current version is 1.2 with 1.3 pending public release. Networking is way better now. Last year I had several issues that I faced when I was playing multiplayer when I played it this week it was very smooth but I am pretty sure the system isn't bulletproof for everyone. Things will break with new content and will be fixed again. Netcode wise though they have their base tech in place.
Youtube has a couple of 1.2.0 videos. I am pretty sure everyone can do the research.
[video=youtube;hm1KJiZGx7Q]https://www.youtube.com/watch?v=hm1KJiZGx7Q[/video]
[video=youtube;ZKI9E6s27-o]https://www.youtube.com/watch?v=ZKI9E6s27-o[/video]
No. Seriously. You claimed that its an old build. I agreed.
1.1.5
So show us a video with the current released one? 1.1.6 or if they are calling 1.2, does not matter. Just bring it on.
I am not outdated. That is the current released version to all backers. What they have behind the doors, or in whatever controlled environment where just a few players can jump in... is just... "speech" for me. Not facts.
So, please, show us that 16-player video in 1.1.6 (or whatever number is the current released version - not PTU) to prove us how smooth it is, as you guys claimed.
Otherwise, your comment that "I got an old build", was pretty much, irrelevant.... silly.
Last edited:
You're a bit outdated, the current version is 1.2 on stable, 1.3 on PTU. There is no more versions for AC as all development streams are now merged on the same build.
Well Nowak might better do that for you as i don't play much of AC, however found some cool things people are doing on 1.3, as boarding a spinning avenger: (also the WIP interior physics is not finished that's why the lack of gravity inside currently)
[video=youtube;he0q3TdVLmM]https://www.youtube.com/watch?v=he0q3TdVLmM[/video]
Yeah that's why it's a WIP, AI adds to netcode, but that is nothing compared to what SC 2.0 will have to handle on netcode, that yes, i'm sure it will give the devs here nightmares on their sleep, ahah
Sorry but you're being silly now, 1.3 PTU is available to all the backers, there is no more passes and alpha access remember?
It's nothing else than the latest version being tested for usually 1 week before fully release.
*Also the latest stable build is 1.2 not 1.1.6.
Have rep to sooth your facepalm.
Ok. It's 1.2
Where's the 16-player demonstration in 1.2?
And where's the 40-players Social module with hundreds of NPC's walking around demonstration? (because talking that 40 players are walking in a ridiculously empty map, and claiming that its handling well, is silly).
Last edited:
Nothing on AI currently, if we saw something of AI was on the SQ42 demo. The SC 2.0 shown no AI implemented as well. All we know is that AI will be added on a update to the Social Module.
AI at the moment is only Vanduul. Human AI is being developed and will be integrated in Star Citizen 2.0 when you go out to do missions you will either encounter players or AI. Here are a few recent updates regarding the AI. I will post the August and September updates below.
August:
Some months we find ourselves focused on two or three big features, but August was one of those months where we found ourselves doing lots of small tasks: bug fixes, feature improvements, responding to feedback from designers. They were all important and useful, but they don’t always make for exciting reading for you, the backers. So we’ll spare you the talk of merge conflicts and build configuration bugs, and stick to the cool stuff!
Design
An interesting piece of design work we’ve been doing is a refactoring of the Kythera perception system for characters. Now that designers have spent a reasonable amount of time building behavior trees in Kythera and using our perception system, they’ve found that they want some aspects of the perception system and the way it interacts with behaviors to work a bit differently. In particular, they want behavior trees to have more control over how AIs respond to certain events in the world such as hearing weapon fire or some unexpected noise, and to also control when AIs look for better targets and when they stick to their current one.
Right now the Kythera perception system will take in stimuli from different sources (vision, sound, and tactile events for characters; radar signatures for ships), perform calculations on whether enough stimuli have been received for an AI to have noticed something, and then pass this information through to the target selection system, which will look for the best target at any point in time. There are various parameters that each AI can set to affect how both their perception and target selection work, but from the behavior’s point of view, it is just told what the current best target is.
This setup allows for simpler behavior trees, and has worked really nicely for ships, but for characters we’ve found that the behavior trees tend to be set up quite differently from ships. In human behaviors, where the acting element is so complex and vital, more control is needed, so it’s desirable to move some of the logic of the perception system into the behaviors, even though this can make them more complicated. So we’ve been working on a design for an improved perception system that will allow behavior trees to be authored with the control that designers want to have. We’re also looking at whether we can improve ship behaviors as well by making similar changes to their behaviors, and that’s something that will be ongoing in the next few months.
Engineering
We made quite a few good improvements to ships this month, particularly with regard to improving their behaviors for Pirate Swarm. Some of the highlights are improvements to approach and retreat behaviors that better take into account max weapon range and current shield levels; changes to make AI missile usage less predictable; and some improvements to avoidance so AI are less likely to crash into other ships.
We also fixed an interesting bug where AI would sometimes behave strangely when going to fly on a spline in scripted situations such as the tutorial. We use the reported thrust values from IFCS (Intelligent Flight Control System) in order to plan out ship movement, but sometimes IFCS wasn’t fully online when we were planning, and so we weren’t working with correct values. So we added a way for a ship behavior to make sure IFCS is fully online before doing something that depends on it.
One nice change we made to character visual perception is to allow them to see things at greater than 180 degrees if desired. Our visual perception system had both a primary and a secondary vision cone, where the primary cone mimics regular vision abilities, while the secondary cone is intended to mimic peripheral vision. That means targets in the primary cone will generally be detected fast by the AI, while the secondary cone takes longer.
The problem is, AI often seem stupid when their peripheral vision is less than 180 degrees, so the obvious solution is to increase the field of view of the peripheral vision. The issue there, though, is that the primary and secondary vision cones are precisely that: cones. The mathematics of the view cones completely breaks if you try to go above 180 degrees. So to fix this, we needed to change the geometric shape that defines the vision space of an AI from a cone to a more appropriate shape when the field of view is 180 degrees or more (which is basically a sphere with a cone cut out of the back of it).
We also continued to make some improvements to the behavior tree editor in DataForge. We added the ability to now specify inputs to BT nodes as a dropdown list of predefined values when this makes sense, such as when telling a character to change to a new stance. Behavior trees can now also embed other trees within them, which allows designers to put a common piece of behavior in its own tree and then add that to their trees in multiple places as needed. As the trees get bigger and more complex, this will be invaluable for keeping them readable and avoiding duplication.
Finally, we made some great performance improvements to the Kythera Recording Server, which is the system that allows designers and programmers to record AI behavior and then play it back with detailed debug information to help figure out bugs or look for ways to improve behaviors. There is potentially a lot of debugging information that needs to be saved for this system to be useful, so we improved the system to be better at detecting what data has changed since the last update and what hasn’t, which means it can better compress the recordings. This is both good for disk space and for making it easier to export recordings to give to other developers.
Source: https://robertsspaceindustries.com/comm-link/transmission/14937-Monthly-Studio-Report
--
Last edited by a moderator:
There is videos and footage of it, you can search them, you are supposed to inform yourself on the claims you make, not make the claims and say "it's this bad, now prove me wrong", that for is me is silly.
Currently i played AC on 1.3, and Social Module on 1.3. There is up to 40 people on the social module, it's a lot of people everywhere (even with its expansion), the social module has no AI, but added dozens of buggies around the map that players are riding with (you can also have 2 players per buggie), killing themselves, also re-spawning on the med-bay. You can now play with 1 of the several FPS armors on it, it's no longer everybody the same model. I don't see server issues. I also don't see hundreds of AI over-killing the netcode, one AI is one entity but by far processed as one player entity, it only has a script to it, AI is pretty much the biggest addiction the Social Areas will have before the shops actually start working.
EDIT: Oh yeah, weather / day & night cycles also under works. But that's not to do with netcode.
As i said, the biggest challenge to the netcode is SC 2.0, not hundreds of NPCs on a planet land-able area.
Remember we are talking about the game where 95% of its population Are NPCs, and its' not the first MMO that has more NPCs in an area than actual players.
Last edited:
I actually talked to Chris about the PU population. He said he wanted about 7-8 million AI and that the background Universe Manager was technically capable of managing all this. He explained the background systems for 15 minutes.
Yeah i remember about some discussions about that.
I don't know if it's like a WHOLE server supposed to manage the AI on the game (independent of instances), if that is done it shall be incredible.
I heard Germany Studio is working on new tech on Cryengine to allow like make multi-crew ships their own instance and then if you leave the ship you enter the exterior instance, different servers but streaming directly to each other as necessary, that allows 1 Ship count as 1 Player, even if the ship has 20 Players inside!
So if that tech is applied to AI management, a server can simply interact with all the instances to make the AI work, and avoid take load from the actual instanced servers.
They are legit giving refunds now and it can't last forever.
This could be the last chance to recover any money from SC, you should do it now because it's likely you won't be able to in the future.
This could be the last chance to recover any money from SC, you should do it now because it's likely you won't be able to in the future.
- Status