Take a look at this trailer for Star Wars: The Old Republic, nevermind the fact that it is better, in every single way, than episodes 1, 2, and 3.
Many video game trailers are done with amazing cgi, but of course the game looks nothing like the trailer. Can this be done or is it a couple of generation consoles away?
Seeing as how both technologies keep advancing, I don’t think it’s a matter of waiting so much as choice.
Of course what you won’t see is that level of animation for anything except premade cutscenes.
In game graphics in modern games are light years ahead of cutscenes and trailers from even five or ten years ago.
The difference, as CutterJohn points out, is in the animation and small details.
No, because the cutscenes used for trailers are not gameplay. They aren’t limitted to what the player can reasonably control or an environment and models that have to be rendered on the fly from just about any angle. And they’re always going to try to show you things that will sell the game and that means better graphics and styling than you can possibly get in actual play.
Just remember that whatever trailer or screenshot you see before release is marketing. They’re not going to show you the identical corridors with fuzzy textures that comprise the 75% of the game between the giant monster fighting arenas. They’re not going to demonstrate the camera motion that obscures the action.
I’m very aware that you can’t judge a game by its trailer. I wonder, though, if at least a linear action game could be done with the cgi graphics, or maybe it would be too expensive to make?
Can a game look like that? yes of course, some already do. Can it play like that though? no way, not for a very very long time. Theres some very realistic looking games out there, for example this is a screenshot from age of conan which looks about as good as that video on high settings. When you start playing you start to deal with things like the UI and the camera following you rather than following the action so it stops feeling like a movie, also theres no game which gives you complete freedom to act and move like someone in a movie.
Actually, we’re not too far off.
Pre-rendered cutscenes has reached a plateau of visual fidelity the last year and a half. Some games’ engines (like the CryEngine 2 & 3) are pretty much up to the standard of the average cutscene - aside from the cutting edge companies such as Ubisoft’s. Crysis played at the Ultra qualitiy is almost indistinguishable from the “CGI” trailers produced for the game. (The caveat is, of course, that the Ultra quality setting was designed for computers years ahead of it’s release date.)
So, at a visual fidelity standpoint, we’re pretty much there.
That leaves two factors: animation and technique.
Animation is going to be a limitation, seeing as most cinematic trailers are highly contextual. That is to say, characters are not using stock attacks, but rather acting independently. Computer games have stock attacks because you need to have a predictable attack that can be bound to a button. Which means you’re going to have repetetive animations and, at best, variations on a theme.
However, “custom” input configurations, like for instance the new motion detection scheme Sony unveiled at E3 yesterday will be able to blunt this. When your controller scheme isn’t limited by a finite amount of input commands, developers will be able to approach animation in a different way, linking the skeleton of the actor up to the direct, physical input of the players. Instead of pressing A to do a left-right swing with his sword, the player will swing the widget from left to right, making the skeleton replicate his exact move.
Which would be contextualized animation, if I’m making sense.
Technique is the last pillar. I couldn’t think of a better word, but what I mean is, basically, how the game treats the camera and what’s happening in the game. Newer games are rapidly integrating “cinematic” camera into it’s actual gameplay. The best example, off the top of my head, is probably the Bourne Conspiracy game, released back in summer 2008.
Here’s a video to showcase what I mean: http://www.gamespot.com/ps3/action/robertludlumsthebourneconspiracy/video/6176997/robert-ludlums-the-bourne-conspiracy-gameplay-movie-1?tag=videos;title;17
(NSFW: Violence)
What you see is that the camera is reacting to what’s happening on screen. If your guy gets uppercutted into the air and knocked down into a crate, the camera will twist to get a cinematic view of that. If an enemy strikes from a predictable angle, you get a split second to make a “counter” move (in this case a Quicktime event) and instead of grabbing you by the hair and slamming you against the railing, you grab his wrist, pivot and throw him to the ground, or break his wrist. While these are stock fighting game tricks, the camerawork is what draws it closer to a movie, or a pre-rendered trailer.
(For the record, the Bourne Conspiracy wasn’t a great game. It was good, but it’s cinematic action style was the real potential.)
So, to conclude, it’s my belief that when contextualized input (i.e. direct player control), contextualized animation (i.e., direct 1:1 transfer of player input) and the skills to code contextualized camera reaches the level visual fidelity is currently at, you will not only have reached the level of pre-rendered trailers, you will have exceeded it.
Well frankly, given an infinite amount of processing power anything is possible. The point is that there is a segregation between “media” and “gameplay”. What you’re seeing in trailers is “media”. It’s being cut using camera angles and film making techniques to highlight what the producers/developers want to show you.
“CGI graphics”, are produced right now, today, in high end action computer and video games. (I’m thinking of titles like Ninja Gaiden 2 or Devil May Cry 4), they just don’t look like trailers because gameplay is being considered above presentation.
The trailers, and more the specifically the cutscenes themselves, are in fact using the in game graphics for many titles these days. Games like Metal Gear 4, (and the aforementioned NG2 and DMC4), aren’t using pre-rendered cgi. They use the in game engine and are animated and cut using higher presentation techniques.
I haven’t been following the development for the KOTOR MMO but I wouldn’t be surprised if the trailers are, in fact, using the in game engine. The game will look like the trailer, just from a standard MMO perspective. (From behind the player’s character at all times.)
Gukumartz went into much more detail than I did. Just read his post. 
It’s not so much the models as it is the lighting effects. Certain effects like diffused lighting (light bouncing off other objects) add realism but require a lot of processing power and take a long time to render.
Some day, yes. Soon, no.
Consider this: Even the most modern games are graphically inferior–by a noticeable margin–to Toy Story, released 15 years ago. And modern CGI has advanced considerably since then.
Inferior? I think you’re just looking at the wrong games.
There are very few movies that use non-cartoony characters well (think: Final Fantasy Spirits Within, and more recently Advent Children) because it’s near-impossible to make human-size humans look believable, tho the gaming industry is remarkably closer than the movie industry is. If you think Ratchet & Clank Future is noticeably worse than Toy Story, I’d love to know how.
Other good examples of game trailers that use in-game rendering instead of pre-rendered CGI, to great effect, are Splinter Cell: Conviction and Heavy Rain.
I work in the game industry–I play games all day. Ratchet & Clank is a great looking game–I don’t mean to take anything away from it. But it’s not perfect. Toy Story is virtually seamless–no visible polygons, textures that seemingly never pixilate, perfect reflections, and of course a rock-solid frame-rate (okay, so that last one is slightly tongue-in-cheek). All of which Ratchet & Clank simply can’t match, even if it does get closer than most games.
All the good bits are cut scenes, the actual game play is noticeably less polished. At least in that trialer.
The cut scenes weren’t. The bits where they’re flying a fighter ship around is positively clunky and the bit where he’s running around destroying boxes with some sort of glowing whip (?) have nice effects and all, but their animation is nowhere near the fidelity in the cut scenes.
That appears to be 100% cutscene, except perhaps the ‘run down the hallways shoot the window and jump out’ scene which was horribly animated and clunky looking. That may be a cutscene too, though, as the scene where spec ops folks drop through the glass roof also has some pretty piss poor lighting effects on the glass as it initially shatters.
Some pretty crappy lighting effects at the beginning (well below Pixar standards, that’s for sure). and it rest looks to be an extended cutscene, too. The snippet of in-game footage where he’s walking around a room with some sort of circular cog-flags is much less finely animated and the surfaces are, for lack of a better term, much less ‘glossy’. The dance scene at the club is either a very poor cutscene or in game footage without the processing power to render top rate lighting effects.
It’s not a question that, sooner or later, processing power/programing will catch up and we’ll see games where the in-game play is as beautiful as today’s cutscenes. But we’re simply not there yet on most games aside from some processor-melting Cry stuff, maybe.
My editing window expired. Anyway, to expound further, here’s a trailer for the original Toy Story:
Even with the video’s crappy quality, one can already see how games simply aren’t yet up to par. Look at the reflections on Andy’s floor as dozens of toys (with no visible polygons) zoom around, or how Andy’s sheets depress realistically as Buzz Lightyear’s box is dropped on them.
and i thought that was hyperbole…
Film techniques for CGI are indeed a considerable step above what we can currently do in rasterization 3d graphics. And I’m talking about what we can currently do on PC’s in games, obviously consoles, being 3 to 4 generations away are even further from the mark.
A lot of such cut scenes use very calculation heavy processes that take even hardware specialized computers, heck, farms of them some time to render a single frame of the scene. Obviously, this doesn’t matter when you have plenty of time to put frame to film, but it’s no acceptable for a game, where you want to be rendering at least 30 per second.
When CGI cut scenes and movies using rasterization, they don’t have to limit polygons, they can use many passes to achieve great post processing effects, reflections, shadows, physics simulations, super high resolution and HDR textures. They also don’t have to use rasterization. Some will use other techniques like ray tracing which can produce very life like light interactions.
We’re still no only far from that, but as film techniques continue to improve/have more processing muscle to achieve better effects, They will still remain ahead of what games can render simply due to the fact that they don’t have ot render several frames per second. They can take their time.
Edited for legibility and clarity a bit, but I missed my edit window. Otherwise, it’s what I already said.
Film techniques for CGI are indeed a considerable step above what we can currently do in rasterization 3d graphics. And I’m talking about what we can currently do on PC’s in games, obviously consoles, being 3 to 4 generations old are even further from the mark.
A lot of such cut scenes use calculation heavy processes that take even hardware specialized computers, heck, farms of such systems, some time to render a single frame of the scene. Obviously, this doesn’t matter when you have plenty of time to put frame to film, but it’s no acceptable for a game, where you want to be rendering at least 30 per second.
When CGI cut scenes and movies use rasterisation, they don’t have to limit polygons, they can use many passes to achieve great post processing effects, and they have time to render great looking reflections and shadows, simulate amazing physics (like the cloth on the cloaks in the OP’s trailer), as well as use super high resolution and HDR textures. They also don’t have to use rasterisation at all. Some will use other techniques like ray tracing which can produce very life like light interactions.
We’re still no only far from that in in-engine game 3d rasterisation, but as film techniques continue to improve/have more processing muscle to achieve better effects, They will still remain ahead of what games can render simply due to the fact that they don’t have to render several frames per second. They can take their time.
It’s widely believed that R&CF’s cutscenes are in-engine, but I can’t find a reliable cite, so I’ll give you that one.
I’m going to have to pause to point out that a lot of the expressiveness and fidelity of a cutscene or a movie is the fact that the camera isn’t stuck twenty feet behind the main character. This is, however, considered an advantage for a videogame.
Damn - I linked to the wrong trailer. Here’s the gameplay I was thinking of.
… You’re pretty used to this sort of argument, aren’t you? 
