Very, very early in the history of motion pictures, movies generally just looked like stage plays. The camera was a fixed, unparticipatory entity set up front and center. No cuts except between scenes, no camera moves. Later the "no cuts" thing was relaxed when it turned out that, for bizarre reasons that we still don't entirely understand, the brain is totally okay with it when your viewpoint suddenly and drastically changes. But still, for a time, cuts were a purely utilitarian thing. When your main character moved off the frame, you cut to a different viewpoint so you could see him again. It's unsurprising that this was the case, because the guy behind the camera was not an artist--why hire an artist to run a machine?--but a technician. The technician's job was to make everything work efficiently and well, and that included putting the camera in as "useful" a place as possible. We have Edwin Porter, an early and hugely influential director, to thank for bringing this ugly situation to an end. It was Porter who realized that in the new medium of film, the natural way to delineate a film was by shot rather than by scene. Porter paved the way for later directors, such as Orson Welles, to control shot composition and continuity in a way which heightened emotion and artistic effect rather than just faithfully recording them. In Porter's hands, the camera was not just an observer--even an active one--but a performer, perhaps the most important performer at certain points. Fast forward a bunch of decades. Are we game developers the unartistic technicians behind the camera? Obviously in first-person games, the decision of shot composition is more or less forced upon us, but what of everything else (games which I will refer to as Third Person Camera, or TPCs)? Our RTSes give us a maximally utilitarian view which we control via scrolling, and our third person platformer cameras orbit frenetically around our characters, ducking stalactites and reluctantly swinging to a head-on view when we back Ms. Croft into a corner. We have excuses. First of all, as anyone who played an early TPC (and even some recent ones) can attest, getting a camera to act even in a purely utilitarian manner is no mean feat. It's surprisingly difficult to keep an unoccluded and undisorienting view of the character without frantically weaving and zooming. Recent tools like GPU occlusion tests have made things a little easier on us, as have the experiences of games gone by. Also, of course, a game camera has more duties than a film camera. Players are expected to not only watch the game, but to control a character as well... and to do that, they need a fair amount of usable information regarding the area around the character. This need is reflected in most TPCs' "mouselook" mode... but who the hell wants to be ducking into mouselook mode every five seconds? Grim Fandango taught us that dramatically placed cameras could be rather disorienting, as the angle of view became disassociated from both the character's angle of view and the global reference frame. And Prince of Persia: The Sands of Time showed us that view-relative control could be just as disorienting, as the prince ran into a new shot and suddenly turned on his heel and ran back due to an opposing new view direction. Clearly, then, we cannot improve on the camera without simultaneously improving, or at least modifying, our control schemes. Lucasarts and Sierra point-and-click adventure games such as Full Throttle worked fine with cinematic camera angles, since control was both view-relative and position-exact. It's unclear whether this would work in an action game. But if we can surmount this difficulty, we have decades of knowledge about cinematography to draw on. Things like the line of action and the 30-degree rule were designed precisely to keep things from getting too disorienting, and many of them can work without modification in computer games. Best of all, most such rules really are expressed as rules rather than wishy-washy "things to keep in mind", meaning that we can code our game to do most of the cinematography for us just as it does most of the camera control for us right now. What do we gain? Well, dramatic effect, of course. But that's not all. While replaying Sands of Time the other day, I noticed I could classify every single disorienting camera move and cut by the well-known rule of cinematography it was breaking. Left suddenly becoming right? Breaks the line of action. A few bewildered milliseconds figuring out where the heck the Prince was after a cut within an unfamiliar setting? Rule of thirds. Wishing I could see if the Prince had anywhere to jump to, or was about to leap to a temporary but annoying death? Balancing the shot while accounting for the weight of the gaze. Film directors have been all over this stuff, for a long time now. They've gotten really good at it. It's not going to translate to the video game camera without a bit of work, but isn't it time we started listening to them?

Aw, don't be like that. Disagree with me, or we won't have anything to talk about. [grin]

Well said mate. However, it is not so simple as you make it sound. "Hey, let's just emulate movie cameras!"

You wan't to know how movie cameras get those epic shots? I will let you in on a little secret. Something that has been kept from the general public for far too long.

Those beautiful shots...those moments of epic triumph and absolute despair...those shots that capture your emotion...are...SET UP! *Shock* *Gasp* The horror! I know!

Now, I will cut the crap and actually throw some meat out there. You are absolutely right to say that epic view can not take place without an absolute change in control schemes. Think of action movies you have seen. What common scenes are there? Looking down the barrel into the eye of the sniper...the camera moving with the legs of soldiers as they run, dodging bullets. How the hell can I control what I want to do in these views? Yes, they give me perspective into the man at the end of the barrel, but how the hell can I aim my rifle when I am looking at myself.


It is obvious that cutscenes have tried to fill this role, but they can only do so much.

You said that first person shooters don't have much of a choice, and I agree. At the same time, I don't feel there is much choice in other games without both sacrificing control and user interactivity.

So, I propose a broader question: should movies and games serve the same purpose? In my opinion, no. In the grand scheme of things, they both aim to entertain: there is no doubt in that. But a movie's goal is to make you forget where you are and tell you a story. I would argue a game's goal is to make you forget where you are, and have you forge a story.

Games are supposed to be interactive. Imagine this. You are the dark prince. You stalk in the shadows, throw your whip over a ledge, swing, grab a banister, jump to another banister, slide down, and half way off, jump, whip out your blades, and land on the head of an enemy, decapitating him. Sweet. Now, how would a movie set up the camera? I bet in the shadows, they would focus on the face of the prince, then have him run at the camera, to a sudden switch where he jumps over the ledge. Next, you see a shot straight up, looking at his giant leap onto the banister. Perhaps then a brief glimpse of his face. Another leap, but this time over his shoulder, and you slide down the banister with him. As he jumps, you switch to a shot under the enemy, looking up at the prince as he comes down upon his prey.

Sweet.

But how does that translate into a game? Quite simply, it doesn't. You could never "play" a movie. Something epic is always lost in translation from movie to game. You don't quite get the same "woah" feeling without cut scenes and loss of playability. A change of controls seems to also mean a loss of play. Even if the entire scene was like Prince of Persia until the last strike, I might still be slightly disoriented and disatisfied. How long does the camera change last? What happens to the guard that I didn't take out while I can barely control the prince because of this funky camera angle. Does time freeze and...shock...I watch a "set up" movie? Yes, that seems to be what happened. The game "recognized" what was going on and made it into a movie. And at what cost? Well, maybe none. Maybe this would be a cool feature. I can't tell, but perhaps it would just be disorienting and frustrating after the 10th time.

So while yes, I agree, cameras need to take a thing or two from cinematography principles, I don't believe that everything can be applied.

Quite simply, games and movies were never meant to be the same, and thus, not all principles apply.

Don't have too much to add, but I wanted to point out that if you're going to attempt a cinematic camera for a 3rd person game, you really owe it to yourself to play/study Ico and God of War. Usually I don't even notice the camera - but those two games stood out to me as the correct way to do it.

Quote:Games are supposed to be interactive. Imagine this. You are the dark prince. You stalk in the shadows, throw your whip over a ledge, swing, grab a banister, jump to another banister, slide down, and half way off, jump, whip out your blades, and land on the head of an enemy, decapitating him. Sweet. Now, how would a movie set up the camera? I bet in the shadows, they would focus on the face of the prince, then have him run at the camera, to a sudden switch where he jumps over the ledge. Next, you see a shot straight up, looking at his giant leap onto the banister. Perhaps then a brief glimpse of his face. Another leap, but this time over his shoulder, and you slide down the banister with him. As he jumps, you switch to a shot under the enemy, looking up at the prince as he comes down upon his prey.

Sure. If I may presume to reinterpret you, in the movie the viewer is basking in the badassness of the dark prince as an "other". They're thinking, what is he going to do? He looks angry... WHOA, he's jumping off the--oh no! he's--DUDE! He totally kicked that guy's ass! In the case of the video game, the player knows what the dark prince is going to do because they are the dark prince. And so a shot sequence designed to illustrate the dark prince's personality and current state of mind would be redundant at best, and annoying at worst (if the player's actual state of mind were different).

But that's only one use of cinematography. Continuing with PoP as an example: Occasionally when vaulting over the head of an enemy, the game would switch to a top-down view and give you a slow-motion shot of the action. This served two purposes: making the Prince seem more more badass, and (sometimes) decreasing the disorientation of an axis switch. After the tenth time it did get less effecive at the badass reinforcement thing, but I wouldn't say I ever got sick of it. After all, since I am the prince, I certainly don't mind the game patting me on the back for being so badass.

Also, keep in mind that not all TPC games are PoP. What about games like Resident Evil, which are combat heavy but still manage to use clever camera angles to heighten tension? It got in the way of combat occasionally, but probably much less than Alone in the Dark's decidedly un-cinematic TPC. Your point that it's not a smooth transition is well-taken, but I really don't think we're doing enough right now.

game != movies
movies != games

This works both ways. Im sure you have all seen video game converted to movies titles. The movies are horrible. As you have heard, interaction is what makes a good game. This is why rail shooters died out so fast and were just a phase. Sure it was fun the first couple of times, because of the graphics and cool movie shots, but they lacked interaction, which is what a video game is about.

Now take a movie that has been converted to a video game. The same is usually true, a very linear story with some degree of freedom added to make it a game. But most times it come up short, since the selling points are the characters and the past movie experience. Most would conclude that the movie was way better in the end.

Video game wise, camera shots can be incorperated into cutscenes. Putting these into a game (too much) makes the game unplayable and ruins the illusion of control. However I think what you speak of is important to note, because they have given us things like slow-motion and instant replays, which I have enjoyed in games. Im sure there are a few other just waiting to make it into our virtual worlds that would enhance our experiences.

Movies are missing the vital element of control and interaction. Due to this simpler model, it has been easier to master the elements of film. Most things are static which as many probably know, is what makes game-making a hard medium to develop for.



So yes it would seem that there is a lot to learn from movies (I agree). The problem arises mainly when you introduce input. And since this input is the main motivating factor of a game, it should be at the root of the element tree. All other factors are supporting elements to the input. For movies I would argue that story is at the root of the development tree, and that is why you find supporting epic camera shots. I think the main question is how to incorperate strong/popular movie elements combined with input, essentially making them dynamic? It is easy to make a cool cutscene, but to what extent can you make that same cutscene interactive enough to be entertaining? I think the rail shooter phase was the closest to "movie magic" cameras as you can get and in hindsight, they were not that much fun. So develop for what makes games fun (interaction), and *support* it with film elements. So far slow-motion and replays have been adapted to games. But I think the next big step will be in interactive stories, where you actually affect your surroundings in a very dynamic way (not just mutiple paths).

Interesting topic, heres some thinking out loud: The basic problem with camera cuts in a game is, as mentioned, control and the moment of disorientation.

Disorientation isn't too hard to combat, providing you don't move the camera too much or you give the player a second or two with no action so they have time to adjust. Not knowing most of the camara techniques mentioned I can't comment on them but it sounds like they fall into the first method. Regardless of game you should always be able to add in a breif moment of non-action if you really do need some kind of strange viewpoint shift. It doesn't even have to interupt the flow, eg. you could force all enemies to miss for the next two seconds after a camera cut or similar.

Control is the real kicker. Most games (particularly third person) are with screen-relative controls. This is nice because it's more intuitive to play with, but obviously makes a cut even more dangerous - that moment of confusion can send you over a cliff edge. Character relative controls would reduce a lot of that (as with the first tomb raider, where left/right changed your characters orientation). Unfortunately this is somewhat less responsive and tends to result in characters which have a turning radius (ick!). The interesting thing is that with this control method what the player tends to do on a harsh cut is to 'freeze' (or at least an experienced player will). They'll just keep the controls in whatever state they were before and hold them like that until they've got a grip on the new surroundings again.

You could easily take this over to screen-relative controls, and many games like Eternal Darkness already do - you'll walk out one door/exit and automatically walk into the next room for a few paces which gives the player enough time to adjust. Resident Evil 4 is another nice example as it has lots of dramatic camera work, but generally it only comes into play when you've already completed the action and it's giving you a nice dramatic view of the (already decided) outcome.

I guess what I'm implying is that if you're going to mess with the camera other than a nice smooth pan/track/etc. then you'd better only do it at a point where the player isn't actually in direct control (even if only for a second).

OT.

Ahh, a topic close to my heart. Please forgive the jumble that is this reply, I'm supposed to be writing my thesis :).

I would agree that control mechanics is one of the main issues before cinematic camera systems become usable. Pascal Luban has an article on gamasutra that covers some alternative control styles, however they don't really resolve the problem (not to my satisfaction anyway).

The primary issue with incompetence in a camera system in a game is due to the methods used to derive the camera position. They're computationally efficient, but ghastly at providing good control. In order to rectify current camera systems, I personally believe that you need to look beyond the polar/spherical coordinate methods with damping, quaternion interpolation, steering, and proportional controller methods, and look at something with better expressive power. Of course, with expressive power comes computational cost :(. Applying cinematography rules (they're more guidelines than rules) will not resolve (m)any of the issues that currently plague games.

There is a lot of buzz about cinematic camera shots for games, but I personally can't understand why you would want to. For games like 'interactive stories' I can see a point, however for almost every other genre, there is no benefit. In many cases, doing so makes the game almost unplayable.

As the anonymous poster said: games != movies. I think the similarities between theatre and movies is a lot closer than games and movies/theatre. Applying these rules is not really the appropriate option, deriving game-specific rules is.

Having a camera that can do cinematics in real-time and interactively is a nightmare (although possible). Limiting its use to cut-scenes completely negates the purpose of it existing. If you have to encode what is happening in the scene (from an action/emotion standpoint) to tell the camera how it should shoot the scene, why not just manually animate the camera? It's a lot easier.

I don't really have a "this is how you should resolve the issue", but I would really like for game developers to REALLY ask themselves why they want such a system. The last thing we need is for cinematic camera systems to become the new lens flare: used for the sake of it.

EDIT: In summary, designing new control schemes to rectify a problem introduced by adding cinematic cameras seems like an incredibly stupid thing to do. Designing new guidelines that take into account in nuances of games would be a damn sight easier and more practical in the long term.