|Posted by Shambler on 2005/01/02 08:31:57|
|Gaming technology seems to have progressed quite a lot recently, certainly this year has witnessed some impressive game engines being released. I think both graphics and physics are getting to a stage where they are generally very convincing (although there's a long way still to go).
So where do people think the next progressions are to be had?? Will technological progression stay in the same traditional area of improving graphics, or will developers start to focus on other areas given that good looking game engines are pretty much standard now??
And, for that matter, where do you think gaming technology SHOULD progress to next??
Physics - HL2 / Havok physics, enough said. Does like physics stuff. Combine this with a structure-damaging feature like in Red Faction, and you've got a pretty good physical world. (Although one thing Friction, I think, was talking about was more realisticly behaving fluids, like flow, viscosity and stuff - that could be a good progression).
Graphics - Combine Doom3 lighting / shadows and skinning, HL2 water and amount of detail onscreen, UT2KX detail textures, Far Cry terrain and foliage, and you're on to a real winner. Oh and add some good art direction too (e.g. HL2) All the main engines are great in their own right but tend to specialise in some areas and be weaker in others. If all the best aspects were combined, it could make damn convincing graphics.
Sound - blah. Sound is always good. Doesn't need to be much technological improvement there.
So I reckon current technology is enough to make a world that looks great and feels great.
I don't know where companies will go, but I do know where I think they should go - not necessarily my personal choice but the obvious weak point of gaming technology: AI. I think this where the biggest advances are still to be made, populating the currently convincing worlds with equally convincing beings. I don't mean enemy that take cover and sometimes through grenades - that is OLD. I mean enemy that vary tactics, I mean different styles of combat for each enemy, and different interactions between them. And I mean interesting AI behaviour outside of combat, beings that do stuff, that are programmed with variable behaviours, that will act from a list of possible options and behaviours in a way that gives them the appearance of personality.
It is an issue that I often wonder about, how hard it can be to give game beings a weighted list of behaviour options (including how they interact with other objects in the world), and then randomise what they do from that list, and let them get on with it. Obviously it's either damn hard, or games developers haven't put the effort in yet (at least not with FPS games). But I think it's the next big step to take (I think Stalker has been p1mping this??).
(At a slight tangent, I think more freeform gameplay will be another big next step in gaming, but that to me doesn't seem like a technological feat, more one of sheer grinding effort to have the game take into account a variety of possible user actions.)
(Although one thing Friction, I think, was talking about was more realisticly behaving fluids, like flow, viscosity and stuff - that could be a good progression)
Indeed, this would make some rather interesting ideas possible. Especially if the physics engine extends to the liquid dynamics simulation. Just think what happened recently over at Asia.
(Although one thing Friction, I think, was talking about was more realisticly behaving fluids, like flow, viscosity and stuff - that could be a good progression).
You might find Mercury interesting: http://www.eurogamer.net/article.php?article_id=57474
Stalker has me really interested with creature life cycles, social groups, higher level of social interaction among the AI all that stuff... I'd like to see that most of all, in other games. I think we've pretty much got some things down well, like physics and graphics.
I think a really interresting area of gaming that hasn't really been explored is control devices. How we interact with the game. A keyboard and mouse works well but we can do better! :)
I geuss what Im getting at is something akin to virtual reality machines but better. I was watching the extended edition of lotr: fellowship of the ring and in the extended "making of" feature it showed how Jackson had a virtual camera for filming the moria mines troll fight. basicly we could walk around a mo-cap studio with this camera and actually see all the characters doing their actions in realtime while filming.
SO translated to gaming, with the addition of a headset or something so you can see the gameworld, you basicly have complete control of your characters moves and stances. combined with a new type of per-polygon hit detection for character models and you have a totally nwe-generation beat-em up where u dont need to animate punches/kicks because the player does it all for you. And more etc...
Obviously its still far far away, and would require every home to have a "games room" with mo-cap technology (heh) but it sure is interresting to think about.
...has the most room to grow IMHO.
However, Warren Specter did make a good comment about AI. He mentioned that it isn't necessarily about making an AI that is realistic, it's about making one that is fun for the player and sufficiently convincing.
If you have to take out 100 enemies, you don't want them all to be as smart as the player.
What!? Non-realism!? Not In MY FPS!
By god man, if a first person shooter isn't realistic, how will we immerse the player!? Without physics that perfectly show how a toilet rolls down a hill that happens to have sticks poking out of it like a retarded plinko experiment in the wild, how will gamers truly feel they are in the world!?
Certainly, we could add more shader effects and AI and/or scripting and cinematics to our award winning engine to make the player play the game exactly how we decide he should play it. If we add enough perpixel lighting, they'll never relieze they're not actually 'playing' anything. And isn't that the point of games now? To be 'art' and 'interactive cinimatic wonders'?
While, I'm mostly going on about gameplay issues, I'm hoping you can kinda see the point... game (well, FPS game mainly) technology seems to be making this march toward realism that I feel is helping to brush aside the games we gamers fell in love with in the first place. Playing Doom and Quake back in the day, you didn't care that the architecture, lighting, or pixelled textures weren't perfectly real and awesome. You just wanted to play a damn game... go blow shit up, lasers rocket explosions. Sure you kinda mighta cared what the overall mission was (SAVE THE WORLD WIN THE GAME), but god knows you weren't sitting down to shoot 10 baddies then be lead through 20 minutes of beautifully rendere cgi cinimatics or in game scripting wonders before being realistically sent to the next mini-game of soundbyte action.
In a way, Doom and Quake's technology heralded their own destruction, making it very possible to start building worlds that could be seen as plausible entities... but not plausible enough that a runic castle, or techy base stlyed enviroments seemed 'cartoony' or 'fake' in their world. With all the advancements made since, and maybe this should be posted in the Quake V thread, I completely doubt such locales could be pulled off nearly as well. They're too vague to be remade as realistic realities that newer tech demands without losing some of the non-euclidean luster they had. A true Quake sequel will never be seen in the eyes of the Quake fan.
I feel we may be in for only a few more FPS games with non-realism combat... the future with all it's technology only holds Half-Life and World War 2 sims... battles against similarly equipped humans, complete with the ability to hide behind crates and push things over in scripted symphonies... until the PC market crashes and we're left with movie liceances and EA Sports games.
Lord help us all.
We had a good dialog with my friend a while ago about the same thing that we discuss here. My point about visuals/physics is that games' engines need realistic dust effect to look more realistic. All those texture effects (bump mapping etc) that exist today still don't give a feeling that I look at a real wall/rock, everything looks too ''good'', clean. All objects in game should be covered by dust that should interact with the player. It should raise up when player drops a crate, it should change when player moves near/touches things, it should move in a windy place, player/things should leave footsteps and so on. This dust can be used in some gameplay situations, e.g. when player comes back to the room he had visited before he can see that dust is reduced in some places, so he can realise that monster has come here and is hidden there.
And that dust should be not that decals or fog effect we have now, but should be something brand new.
This thing you mention about dust somewhat reminds me of a post by lead designer of Fable (a very big 3D RPG) on some forums. Basically, players wanted the world to have "dynamic trees" as in trees that would grow like a real tree would, get new leaves, drop old ones and die of old age or bad water and stuff. They tried to pull it off, but no matter what algorithms they use for the tree calculations, out of 100% CPU time used by the game, up to 10-15% would be used by those calculation ALONE, which was too big of a perfomance hit to take.
The Buzz Word For 2007 Will Be
Cinematography; after Unreal 3 based games come out, designers will realize that the harsh, ungodly lighting of the real world is not all that desirable.
An RPG based on Igmar Bergman's The Seventh Seal will be declared The Myst of the year 2008. It will be such a milestone, Roger Ebert will declare The Film as an artform dead and devote the remainder of his career to reviewing games. Quake mappers will be shocked to descover that the lead developer is none other than our own Wrath.
Later that same year, Carmack will declare it the end of the line for engine development. Quoted in Ipod News Daily, 'Oh for piss sake, how much more do you need? If a level designer can't recreate an accurate representation of Manhattan Island in 10 million polies or less, you know what I do? I show him the door."
There's a difference between something being realistic as in a real-life style, and something being realistic as in being convincing whatever it's style is.
You can have a fantastic enviroment that is nothing like a real-life enviroment, but is still convincing and "realistic" (i.e. a true representation of whatever it's trying to represent). In fact, it's crucial to do so - the more fantastic enviroments have some laws of existence that they obey, the more convincing they are, the more they make the player feel like that are in that place, rather than merely watching it on a screen. This was something that Clive Barker stressed in an interview about Undying - and one could say he's fairly experienced in fantastic enviroments AND that Undying did a good job of using it's technology to make it's enviroments convincing and all the more damn cool for it.
So I don't think the technology per se is a problem at all, I think it's a benefit, you can make fantastic-style games more fantastical and better because of it. The crucial point is how - or even whether - developers choose to do that. And this year, Painkiller and Doom3 did at least (Doom3 could have gone further into sci-fi enviroments, admittedly, but I really felt like I was playing through a sci-fi TV serious when I was playing it).
And of course the gameplay is a different issue...
You can have a fantastic enviroment that is nothing like a real-life enviroment, but is still convincing and "realistic"
I usually differentiate these two with the words realistic and believable. It's also worth quoting one of the Valve guys from an old interview about TF2:
realism is a tool, not a goal
The best example I usually cite in regards these definitions is the depiction of spaceships in movies: the Odyssey in 2001 is intended to be, as Ridley Scott put it, 'a Nasa object' i.e. gravity created through centrifuge, engines are off even while traveling at immense speed etc.
Conversely, the Nostromo in Alien looks far more familiar to us with its grubby platecrusted hull and industrial-strength interiors, even though such a vehicle would have a hard time actually working.
A quote from another Valve guy, Marc Laidlaw, is appropriate here:
"As far as the narrative goes, I lived by two rules: Poe's "Totality of Effect," which has to do with maintaining consistent atmosphere and tone, and another unnamed rule expressed by the science fiction writer, James Tiptree, Jr., which I'll quote loosely since I can no longer find the quote: "Start your story 500 feet underground on a dark day and then...don't tell them."
It doesn't matter what your diagesis is based on - whether it be period medieval realism, WWII or the gibbering horror of the Chthulhu Mythos - as long as all the design elements conspire towards convincing the player of the same thing.
Realism Is The Future! : /
Simply because the more a game world deviates from the world we live in the more it will cost to produce. A realistic title will be able to make extensive use of existing tools, code and content libraries to cut production time and keep the number of people directly involved in the project down.
I think realism is more common because it's easier to sell.
I think the trend seems to be towards semi-realism, i.e. near future enviroments that are fairly real-life, but with slight sci-fi / gothic tinges. E.g. Far Cry, HL2, Stalker, FEAR.
Not that I approve of that. More Undying / Wheel Of Time / Painkiller / Enclave / Q2 etc enviroments would be preferable IMHO.
One Reason Why Humanoid Monsters/characters Dominate So Much:
for skelatal animation, it is easier to rig bipeds to the same base armatures than it is to figure out the motion of several different monster types. Dragon fligh, versus copter hover versus scag flight; they all need a different set of motions.
heck, even switching between animating normal bipedal leg movement versus the extrajointed goat leg leg movement is a fair bit of change.
I disagree with Mr Shambler, there's a long way to go here. Sound can be very deceptive - I spent the first half of my life listening to cassette tapes and didn't care, but now I'm used to CD's I could never go back. You don't really notice poor until you've heard good.
Take Doom3. Knock a metal box over onto a metal floor... "clang!". Good enough? Try pushing it.
Where's the scraping that varies according to surface and speed? Run it over some rivets. Where's the "tink! tink!". Drop it in an airduct and then a cavern. Where's the different acoustics?
You could argue that solving this doesn't require much in the way of technical improvement - just more sounds for every combination. But I think that the implied demand there is exponential, especially when combined with advances in physics, and sooner or later people are gonna have to start managing it by doing some proper simulation. That's a huge and largely unexplored technical challenge.
The problem is that most people don't even realise there's a problem. In all the hype about next-gen stuff, I don't think I've seen a single mention of sound.
I have a hunch that the situation will continue stagnate for a while, probably well into the next-gen consoles, and then some backroom genius will suddenly come out with a system that blows everything else away.
In conclustion, I look forward to gently rubbing my (virtual) hairy flaming dildo over a (virtual) ewe, and hearing the exact combination of "fsssssh", "crackle!" and "BAAA-AAA-AA!" that I expect.
we're you also talking about sounds that are generated in realtime, or just having lots of recordings for different events?
Generated in realtime, although that generation might be just taking a bunch of recordings and combining them in some novel way to make an "original" sound.
Arguably, almost all game sounds are generated to some degree. There's almost always distance attentuation, and often reverb and occlusion.
I'd really like to see something more fundamental, where the surfaces involved and the way they're interacting all combine to generate an infinitely variable sound.
I vaguely remember some siggraph papers about this kinda stuff... will dig them up tomorrow.
what animation software are you using?
A spinoff of Ikka's AMAZING "Hall of the Shambler God" map, the episode basically deals with where Shamblers come from (the birds and the bees, eh?) Elder-worldish/other styles would probably be prominent, mixing in some ruins of temples, possibly Greek/Roman in style. Other ideas to flesh out the style as appropriate, no real solid idea as to how it should actually be. Generally more open-ended maps, as Shamblers are rather big.
The final boss of the episode would be the Shambler God itself, some sort of badass mofo no doubt. I have some ideas of what he would look like, and they're all of a huge twisted Shambler with a bunch of mouths coming out of different body areas (shoulders, chest, hands.) But then again, maybe not.
The thoughts popped into my head of what a Quake V episode could be.
I totally misclicked the wrong forum? I was reading Quake V stuff, and then poof. Disregard this, eh?
AI And Mirrors
I got to think about AI due to this scenario in Hitman 3:
Notice I am hiding in the bathroom stall, and there are guards ahead of me. They didn't know I was in the stall, yet you can see that they can see me. The problem is, the AI wasn't written to see the player or other people reflected in mirrors.
This should be done in Hitman 4 (among other games), it would make things that much more interesting/challenging.
Good point there phait!
New Engines Etc
While it will pretty much always be graphics that are the most noticeable improvement in each new batch of game engines, there is work going into other areas to make things more realistic, currently its physics and animation (hl2, etc) but we need more.
1) A.I ffs! Go stalker show them how it is! Phaits example is a classic one, A.I compared to where everything else is these days is just fucking shit! Its one area where HL2 fell down I think, the A.I just felt the same as in HL1 with maybe a few tweaks to understand the physics objects. It could have been so much better. What if combine actually threw grenades to try and dislodge piles of stuff to fall on you (unscripted) and actually pick up and throw objects at you, block passageways with debris etc. But no they stand there and shoot while YOU do all that. A.I really needs to understand the purpose of everything in the game world and be able to take advantage of that in a clever way. Its the next logical step (mirrors included!).
2) Animation, its gotten much better in last year or so, ragdolls and inverse IK make the character movements much more realistic, but its the basic character movements and actions that still need work. They still move very "digitally" if u know what I mean, always moving in straight lines, if u move into something they will just kind of "strafe" out of your way. What if u could actually barge past someone, what if when someone is hiding behind objects they actually take up a posture that fits the object they are hiding behind. This is actually a culmination of animation and physics I geuss. Also, while ragdoll deaths can be quite realistic most of the time, they just have no personality. I do miss the days of animated deaths, where people would spasm on the ground, or die holding their necks etc. This needs to be brought back and conjunction with ragdoll. so say, an animation is played in the arms of something flailing around while the rest of the body is ragdoll if u see what I mean.
3) Freeform worlds. Engines can do it now, developers should fucking use it. "on rails" shooters should start fading away and freeform worlds will be getting bigger. See Operation Flashpoint, Stalker, etc.
Nice One Daz.
3 good points there.
IK As A Technique
makes it easier for the animator to organize the bones of his model and move them about as such, but it doesn't necessarily mean that what the player will see in game is any better.
Realistic axial restraints coupled with a more accurate physics simulation of actions/reactions at a per pixel level of accuracy (instead of the good old bounding box) is what we are seeing as applied in HL2, Doom3, etc..
Except It Can Mean Exactly That
In Doom 3 for example, an IK solver is used to ensure that a characters feet don't float above the ground, but conform to the terrain. This isn't perfect, but it is an example of procedural technology addressing the fake appearance of game animations.
I dont see #2 happening any time soon -
thats really tricky to make model movement that acts according to both the pre-made animations and world geometry.
Now it feels like prefect moment to mention handless Gordon Freeman that drives the vehicles and moves barrels by the mean of TheForce (� LucasArts)
I like how you say 'see stalker', when its not out yet.
"On-rails" shooters have its appeal too. Not everyone wants to play big open-world game. And making such game good quality is a lot of work not every dev can afford.
Yeah, I Left That Part Out
so as not to muddy the point.
IK is typically used in modeler to rig a model that is, again, typically not dependent upon a particular engine to function (so you can use the same model in Quake, Q3A, Unreal Tournement with minimal changes). In engines with built in skelatal animations, this is less true.
I Was Referring To
what Blackdog said, about the characters legs etc conforming to terrain for a more realistic pose. You can see this in Farcry but im not sure if other games use it... Actually I think the spider bots in Doom 3 use it, bleh u know, sorry for confusion :)
Ah, It's My Fault
for splitting hairs that did not need to be split. 600 years in a Franciscan order arguing about how many angels can dance on a pin can do that to a man.
had feet adjusting to terrain in some occasions...
...i recall a point when some debris was flung at alex and she lifted her leg to prop her foot ontop of it while she fired her pistol back. It was a nice detail that helped the believability.
Isn't That IK?
it's a basic thing that has been in a few games for a long time. Hell, even Zelda: Ocarina of time uses it on the N64 (unless it's just the GC port that has it). Links feet always touch the floor.
I don't understand why more modern games don't have it.
It would be nice if characters looked more natural walking up stairs in more games though.
It can also be used for doing little shuffle steps and moving a character accurately, which is a pain to do with normal animations.
Daz, I've read that Max Payne did some stuff with combining animation and ragdolls - the example was someone hit with an explosion would play a windmilling animation on their arms while the rest of them was ragdolled. Also, Havok is pushing this kind of stuff as a big feature now, so it's likely more games will have it.
I was having this discussion with Lunaran in #tf until he accused me of being a pessimist and told me to get a job :(
Anyway, my thoughts were that I don't think that games are going to be graphically innovative for a while now, so the important thing is to use the technology we have in the most resourceful manner possible.
For example, games should use the current engines in entirely new ways (like Anachronox did with Q2's engine) instead of just cosmetic facelifts (like Quake 4 is doing with Doom 3's engine)
FPS especially will get stale very quickly if we just have games built on D3 and UT2Kwhatever without any sort of new gameplay or creative ways of using the engines/technologies offered.
Also, the trend of making things more realistic has been going on for the past 10 years, and Phait made a good point -- that no matter how realistic games get graphically, we'll always have little things like he mentioned that sort of ruin the sense of immersion in a realistic world. I think this trend is hitting a wall as games get to the point of being almost photo-realistic, and I think because of that, there will be a surge towards surrealism and more artistic uses instead of just a straight recreation.
To give you an example of being resourceful with the current engines, I want to bring up HL2. HL2's physics system is quite good (though not perfect). It was used minimally in terms of gameplay in HL2, but I think a mod team, or even someone else licensing the Havok engine could really come up with some interesting uses for it. Perhaps we could even have entire games based around solving physics puzzles -- coop games online where you need to work as a team to anchor things down or launch things using planks and see-saws or whatever.
Pope said he liked my metaphor when I was talking about this in #tf, so I'll give it again here.
The current state of games is analogous to the way Native Americans and European settlers of America used the buffalo, and nature in general:
Mod teams and developers who create an entirely new game from existing engines are like the Native Americans.
The Natives would kill the bufallo, eat the edible parts, make coats from the fur, tools from the bones, slippers from the testicles, (ok I'm getting silly but you get the idea) and just generally used the world around them as resourcefully as possible.
The big companies who are just looking to pump new content out ASAP, and uninspired mod teams who think a few new textures will suffice are like the Europeans.
The European settlers just killed the bufallo and ate it, and probably sold the fur to buy whiskey.
So even if graphic technology slows down for a while, we still have some really complex engines and current technologies to use. The problem is we need more people to be like the Native Americans, and less to be like the Europeans.
I think it worth mentioning that the main problem with the European settlers is that generally they didn't eat the buffalo they killed: the animals were skinned, the pelts shipped east and the carcasses left to rot.
Not that I'm arguing with your point.
OK Cheers Mike
(PS if you ever derail a thread like that again I will hunt you down and skin you)
Stop being such a pessimist.
Also, get a job you hippy.
Iv read some woes that we even need to stop the hardware race for a while and fully use what gfx technology we have now.
But that wont happen, and consoles will take over
Oh, and HL2 is great example.
Basically, quake engine with added features to make it cool and modern and nice looking.
Some call it 'hacks', but who cares -
game looks good, and runs smooth on pos machines.
Interesting post, well up to the buffalo bit whcih lost me really.
But I also agree with speedy, HL2 is a good example of evolving tech that gets used well.
Good post, I agree with you.
Good post, I agree with you.
Worst post evar, someone close this thread.
Good post, I agree with you.
So, uhhh yeah...games and stuff :} Well, I've been playing a lot of 3rd person action games on the Xbox over the last few weeks, and it's really struck me just how much more cinematic these games try to be, in terms of the number of scripted sequences/cinematics etc, compared to the FPS stuff I've been brought up on. I guess that makes sense, as it seems more logical that an "interactive movie" would adopt a 3rd person perspective. So, yeah...actually I don't really have a point here and I guess this isn't really a comment on gaming tech, so I'll shut up now :}
You must be logged in to post in this thread.
Website copyright © 2002-2019 John Fitzgibbons. All posts are copyright their respective authors.