Senior Member
That would be great, that's what I meant by a more organic planning stage. Tell the AI where to go in a macro level, not micro manage it's very step through the evironment. Critical points would need go-codes, etc. You could setup the AI in blocking positions for suspected reaction routes by the AI and have snipers cover key areas. Would be wonderful.
I just don't know if today's AI could pull it off without a lot of scripting, then you fall back into linear play.
Senior Member
As much as some people are going to hate me for saying this...Originally posted by Defuser:
At what point did FPSs, particularly tactical shooters, become the baby's choice of genre? At what point was the decision made to make games blithely simple in this genre as though everyone who plays them can't handle anything more complicated than run and shoot? How come strategy games, which sell HUGE amounts, don't have to make such concessions to their players? Is the assumption that FPS gamers are stupid somehow? Where did this come from?
One major cause is the rise of so-called E-Sports and Person vs. Person (Multiplayer) competitive video gaming.
The growth in this segment of gaming has all but simplified what games, especially, FPS, are all about in terms of their complexity because the emphasis is now on indiviual skill (DM), or team skill in games like Counter-Strike and BF2.
And actually, it was only a matter of time before this really took hold with mainstream audiences and game developers as the FPS is the most accessible genre there is because the core game mechanics are very easy to understand: Run around and shoot things from a first person perspective. Compare this to more complex games like RTS and MMORPGs which involve a lot of concepts which can't be visually seen when spectating a match, for example.
I don't know if it is a good analogy, but it would be like comparing competitive chess to Baseball from purely the spectator point of view.
Chess is a cerebral game of strategy which means most of the competition and "exciting" game breaking "action" is going on inside the player's heads...
As opposed to Baseball, if the batter hits a ball far into left field, it causes an immediate chain reaction which can easily be seen and understood and that is the left fielder scrambles for the ball as the batter then tries to run as many bases as he can while the ball is then passed to another player and so on and so forth in a frantic effort to stop the progress of the batter as he runs.
Again, this may not be the best analogy, but I think it addresses the issue you raised (even if it was a rhetorical question) because even if there were no such thing as E-Sports, again, the FPS is the most universal game genre there is because the core mechanics are very straight forward and simple and therefore, anyone can relate to and understad it compared to more complex genres like mentioned above.
It's almost like a Catch-22, blessing and curse all in one.
Member
You've got a point. FPSs are the most accessible because outside of shooting and moving, you've got nothing else to learn. Where the tac shooters differ is that it is about more than just moving and shooting. This is the 'unseen' element that you spoke about, much as the 'unseen' element of strategy games are the decisions being made by the player to expose inadequacies in his opponent's defence, or to consolidate his own. Tactical shooters are just that - tactical. The tactics come from having to apply specific proceedures to get an advantage over your opponent provided the game is at least SOMEHOW realistic. By making these decisions and applying the tactics, you are immersed in the game. Immersion equals success!
Now, about E-sports like CS:S. CS:S CAN be very tactical. Not in any conventional sense. Certainly not in it's application of realistic tactics, most of which come unstuck. CS:S makes no concessions to realism despite being set in the 'real world'. Recoil does not equal realism. Realism, at an appropriate level, should affect all aspects of a game so that there is no incongruity between a gratuitous lack of realism in one instance and strict realism in another - it should feel complete. You should have a world with rules to be immersed in. Gross disparities break immersion by reminding you that this is a gameworld that is inadequately balanced - the best gameplay is one in which you don't detect specific design decisions because they fit in with the game's philosophy as a whole. Again, immersion, plus realism, equals tactics. Tactics, plus realism, equals immersion. Immersion plus tactics equals realism!
E-sports unfortunately seem to be based on corridor shooters with very little avenue for tactics outside of running and shooting. Painkiller, for example. I know somebody will tell you that there's a lot of poker-like psychological warfare involved in e-sports, along with split-second timing to collect power-ups and the such like as well as the need for pixel-accurate shooting. But that's Painkiller. It is one of a long line of corridor shooters designed to be played in DM, as a DM game, with it's own specific world and sets of rules and boundaries. It is not a tactical shooter because the tactics it concerns are not valid in a real sense. When we say tactical shooter we are making a concession that the tactics involved in such a game correspond to the ones that are in use in real life - not that the tactics involved in Painkiller are any less valid, but they are not the tactics people think of in terms of weapon handling and room clearing, like in real life.
What the designers need to be at pains to realise is that just because a game is situated in first person that does NOT mean that the game is subject to exactly the same considerations as your average 'shooter'. This is a game all of it's own. Tactical shooters should not even be in the same sentance as other shooters because there is so much under consideration IN ADDITION to accurate use of virtual firearms. Each stripping away of the opportunity to use realistic tactics results in the game becoming less and less tactical in a realistic sense (a tautology, but you get the point), and towards the unfortunate zone of bland grey mediocrity that cannot decide what it is - a shooter with realistic elements but the lack of ability to institute realistic tactics - Lockdown.
Any rhetorical questions in there? I do like to make use of the odd bit of rhetoric. I'm compulsive like that.
Senior Member
If you look at the team AI from games like The Regiment and SWAT 4 and yes, even Lockdown, there have been vast improvements since the Rogue Spear and Raven-shield days on how AI handles clearing rooms.Originally posted by DayGlow:
I just don't know if today's AI could pull it off without a lot of scripting, then you fall back into linear play.
The Regiment AI move fast, push corners and shoot to kill. The SWAT 4 AI do the same except they have the added restriction to call for compliance which I find often gets them killed. Lockdown's AI did fairly well in small rooms when they didn't flashbang themselves, but couldn't handle larger areas. Their largest problem in the larger areas was their refusal to move away from a few feet within the door. Instead of pushing forward and clearing the area piece by piece, they stopped and crouched behind cover just inside the large room and waited.
The biggest problem with SWAT and TR AI wasn't in successfully breaching a room they were ordered to, it was in what they did AFTER that. They often failed to watch the entrances to the room after they'd cleared it and that often got them shot in the back. That is not all that difficult to correct, really. All that's happened is that they had a priority, they breached the room, killed the enemies then called clear and.... ran out of priorities. After that, the default priority assigned to the AI should be to cover the doors/entrances from dominant positions until the go-code is given or their team leader calls for them to do something else according to the plan (fuse locks, secure hostages, stack/breach the next door, etc). It's not really a matter of scripting, just making their priority to secure and cover the room they're in when they have no immediate orders instead of standing around complaining like the Lockdown AI.
90% of the time with TR and SWAT 4, that was my biggest complaint about the AI. We would clear a room and then while I was moving towards one door, someone would walk in from another and just gun us all down. I found myself shouting all too often "Why weren't you watching the damn doors?!"
Additionally, speaking of SWAT 4, I would have liked to see them take some bloody initiative and secure weapons, suspects and civillians if they're left in a room with surrendered suspects and hostages and no orders for a period of time. I once told my team to breach a room, which they did nicely and had their suspect surrender, then I went around and completed the rest of the mission myself. When I came back to them, the suspect was still kneeling on the floor with his AK47 laying on the ground in front of him.
While having direct control when you want it is a good thing, you shouldn't have to babysit your team members and tell them every individual thing they're supposed to do. Granted we don't want them running off on their own, but part of properly clearing a room is ensuring that anyone IN that room is either dead or secured.
Senior Member
Good read Defuser(and others). I'd like my medal in gold and navy blue please.
The two things I find so ironic about the R6 series is how the core gameplay has gotten worse in the modern graphics era, and that same gameplay is what made the series popular in the first place.
I'm starting to think that they focus too much on graphics and the limitations of whatever engine of the game and less on the actual gameplay. Granted, LD uses the GR engine, but even that is designed for a somewhat different game style, Ghost Recon style.
As cool and flexible as the unreal engine is for RvS and Swat4, it still makes them work within its parameters as opposed to doing what they want. Then they have to build on to it. If they were using their own engine, they could make it do the things they want first, like R6 and RS without restriction. Not saying graphics are not important, but these plug and play game engines are not so simple when it comes to TacFPS's. Swat4 seems to have done a good job with unreal tech, but I still think they are better off going from the ground up.
I know there is a seperate "engine" thread, I'm just saying this here because I think the overall engine has actually contributed to some limitations in what they wanted to do or could do with the games. Especially in the realm of optimizing it.
Senior Member
There seems to be limitations with Unreal with SWAT4 as well. One of my biggest frustrations is that the AI does not reconize the difference between an open and closed door, and will always double stack, exposing them to fire as they cross the door.
Posting about this I found out that the map maker needs to setup stackpoints for every door. The AI is clueless on how to do it unless it's part of the map, thus the stack points are the same for an open door vs a closed door. Better solution than RvS where they just stand infront of the door, but still limiited.
Senior Member
TedSmith brings up what seems like an age-old conumdrum game developers still haven't gotten right in team-based/sqaud based shooters and that is balancing the Team AI to be independent enough to do their jobs competently, but not so independent the player can't control them.
A prime example is when Ghost Recon 2 for the Xbox came out.
I remember watching the Developer Diaries on Gamespot and they were talking about the improved AI of your Ghost team mates and I specifically remember one of the devs said during testing that they made the AI so good at eliminating targets, taking cover and acting like "real" soldiers (within the confines of the game) it didn't leave much else of the player to do other than just tell them where to move, or to hold, etc.
In a way, I kind of wish MORE games took this "extreme" approach because this is the level of AI where I think (*hope*) we are headed where the AI simulates being as alert and competent as a real human where we don't have to babysit them as much as we have to now... Especially, in CQB squad games where being alert and aware of your surroundings is sometimes even more crucial than in combat with large, open environments.
I also understand the counter-argument people play games to PLAY games and not just watch inhuman AI shoot other AI and leave them with nothing to do, either.
However, it is this balance which no game has ever truly reached in my personal opinion, and I am just hoping we eventually get there because as Ted said, it doesn't have to be some complex routine that sucks up CPU cycles and brings the game to a crawl like most devs think "realistic" AI has to entail. It is just a matter of implementing priorities to where they at least appear to be more "alive" (alert) and can make small decisions like the above example of securing suspects after a period of time passes and there is no input from the player.
Also, what I would like to see AI do is something I have only experienced in the console versions of R6 (R63; Black Arrow; Lockdown) and that is once they properly clear a room, they take up strategic positions around the room and do exactly what Ted said and that is cover potential threat areas.
For example, in R63, once Price, Weber and Loiselle take a room down and yell, "Clear!"... They all move to appropriate cover spots and take a knee. Price will move and kneel behind a desk if one is there and train his fire sector on anything in front of us. Loiselle will take a knee by the door we just came through and cover it in case a Tango sneaks up behind us. Weber will kneel and cover whatever angle is left depending on the room's configuration.
It is partially scripted, but also dynamic enough where whatever team member is nearest the door we just came through, for example, will immediately take that role and cover it when they move for cover. It isn't just Loiselle who covers the door every time and depends on who is nearest when the room take down is finished.
What I think the console versions use are some kind of "cover point nodes" the AI automatically reads once they finish the room clearing routine and they automatically head toward those nodes and it presents the illusion of covering different areas and fire sectors.
If this kind of system is what it takes to make the AI more "intelligent" I wouldn't mind it at this stage because at least the nodes are strategically placed to where they are in a position to mimic the jobs done by real fire teams and actually provide a real sense of security and not just the illusion of security...
And this is on a 1999-era PC (Xbox) with a Pentium III 733 MHz CPU and 64 MBs of RAM!
It still baffles me as to why they can't "port" the AI from R6 consoles to the PC because even as anti-console as people are... If you have played the R6 games on consoles the AI is the one aspect which actually SURPASSES the PC versions, hands down, and is basically one of the only reasons I even got the R6 games on Xbox, let alone just to have as part of my R6/Tom Clancy collection.
Senior Member
I know the cover nodes are in LD, no idea if the map makers used them properly though.
From the NED editor documentation:
CoverPoint
CoverPoints are placed throughout a mission anywhere you want the AI to consider as valid cover during a firefight. When placed, it displays an adjustable arc. During gameplay, if an AI senses a threat within the angle of that arc, they use that location for cover. CoverPoints have a number of properties that allow them to be adjusted for different heights and angles. Due to the number of CoverPoints that need to be placed in a mission, you may find it easiest to open a mission included in the game that uses the map you’re scripting on and use the File menu bar options to export/import CoverPoints.
Maybe they were so rushed to get it out the door they didn't fully script all of the missions?
Senior Member
