Why Game Trailers Fake Things

Much to the chagrin of game developers, most game trailers need to be made before the game is finished. A lot of people who watch game trailers aren't too pleased with this fact either, because it means how the game looks in trailers sometimes looks different than how it looks when they're finally playing it at home. A few infamous examples being the splashy trailers made for Kill Zone 2, No Man's Sky, and Watch_Dogs.

The case of Kill Zone 2 was quite an exception because it was a pre-rendered video intended for internal purposes and wasn't meant to be representative of realtime gameplay. Unfortunately, much to the horror of the developers it was revealed as the latter; they ended up scaling up the production to meet that high standard.

No Man's Sky unfortunately bit off more than it could chew; many things shown in early trailers didn't make it into the game at launch. After years of free updates, the game now looks very comparable, has a very dedicated fan base, and has earned back a lot of good will. Fortunately, the early success of the game gave them the latitude to be in a position to gradually bring the game closer to the vision of the early trailers.

Watch_Dogs' differences from their E3 demo and the final game are mostly pretty subtle if you're not someone who pays a lot of attention to the graphical fidelity of things like reflections, textures, shadows, lighting, etc. But people who care a lot about games with shiny graphics which show off the power their latest PC build are very vocal, and can feel cheated and distrustful of the publisher, developers or game advertising as a whole. (just read the comments of this Watch_Dogs graphics comparison video)

The internet LOVES to compare the retail version of a game and trailers which might be over a year old. I truly feel for games which receive this level of scrutiny O_O

The other side of the coin which unfortunately doesn't get as much talk are when games look BETTER than their early trailers. Take a look at this comparison video for Uncharted 3's trailer and its finished version.

There are myriad reasons it is simply not practical to wait until the game is finished before the first trailers get made. The marketing of a game has to start early to build interest and hype. Also, not selling a finished game means money is being burned, but no money is coming into the company. This is a problem for game companies of all sizes, but especially at large AAA game companies which can have hundreds of jobs at stake.

The reality is: game trailers are almost always made when the game is unfinished. Trailers can take weeks or more to make, and games nowadays are generally worked on up until release (and after) which means even a trailer put out in conjunction with a game's release is comprised of footage captured from an older build.

Making a trailer for an unfinished game means entire sections might not yet be built (especially when early in production), the game hasn't been optimized for performance, artwork is not finalized, and parts of the game may still be undergoing enormous changes. This means the version of the game being used to make trailers can be very different from the one that ships into stores.

Unfortunately, games which look better when they ship don't seem to get as much attention. I think because the internet is just calibrated more for outrage -_-

So what are the ways you can compensate for the game's unfinished state when making a trailer early in production? The main thing you need to do is concentrate production on the parts of the game which will be featured in the trailer such as: level and character art, animation, visual effects, sound effects and voiceover.

Here are a handful of options for creating game trailer footage with increasing level of risk:

• Use editing to shorten or dramatize

• Video editing filters and plugins to touch up visuals

• Script sequences which cannot be controlled via player input or the game's dynamic systems

• Pre-render in-engine footage

• Animate manually

Editing

The beauty of making trailers is you have the power of editing at your disposal! So if for example, there's a lever in the game which opens a door (but that function doesn't exist in the game yet) all you need is the ability to capture a shot of the lever opening, and a separate shot of the door opening. Cut together, people should be able to understand that the two are linked. Regardless of whether or not you're "cheating" to get this footage in-game it's sometimes good to use this sort of cause/effect style editing to make game concepts very clear. 

The trailer just needs to communicate a lot of "If, then..." concepts from the game, but if none of those things are wired up in the game, editing will allow the audience to understand they're connected. This means you can break up some of the game's mechanics into individual shots, and as long as you can get the game working well enough for those shots you should be good.

Video Filters and Plugins

If you're planning on adding visual effects to the game like: light bloom, light rays, chromatic abberration, color grading and particles, a lot of these are fairly trivial to do in editing software like Premiere or After Effects. There is risk of course, because you might find out later that those effects couldn't all be implemented in the final game.

But in the grand scheme of things, this sort of visual difference doesn't seem to get quite as much scrutiny in game graphics comparison videos. Based on comments I've read over the years, people seem to put up the most fuss when the differences are in lighting, texture quality, explosions, etc. 

This scene and the others in Firewatch's E3 trailer were color graded in After Effects by art director Olly Moss. Also, the color scheme in the finished game ended up being orange rather than green.

Scripted Sequences

I'm not a game developer by trade, but here is my understanding of how "scripted sequences" work in games. Say you want to get a shot of a person getting into a car, then driving it into a telephone pole which then collapses onto something explosive, and then that thing explodes. Building this as a scripted sequence means the game doesn't need to have certain systems built like physics for the car hitting the telephone pole and knocking it over, visual effects for the explosion being dynamically triggered, and maybe even lighting thrown off by the explosion. Since it's not a ton of systems running in realtime behind the scenes, it's easier for the computer to render.

As far as I understand it's kind of like how Star Trek's automatic doors are just opened by stage hands rather than going through the effort of making real automated doors which detect the actors The first option is much easier to do on a budget, but looks the same. 

This approach is much riskier because you're assuming much more of what you'll be able to implement into the game by the time it gets released. The more dynamic systems you simulate with a scripted sequence, the more potential points of failure you're adding. 

Just hope this isn't what happens to players when they get the game :P

Pre-rendered in-engine footage

Next is is pre-rendering footage out of the game's engine. Basically, this means creating images using the game's engine like they do for feature films. For example, a single frame of a Pixar film can take DAYS to render out because of all the dynamic physics and lighting which the computers need to calculate per frame. The main problem with games is they need to generate 30-60 images PER SECOND.

But there is the possibility of running all the game's dynamic systems which the computer cannot yet run at 30-60 frames per second, but instead having the engine render single frames which then can be put back together in editing software so it's running at the proper speed (like a digital film strip). I don't know this for sure, but I think this is the subtext of trailers which say things like "All footage capture in-engine." The caveat is that footage might've not been running in real-time, which of course it will have to when you're playing the game at home.

This and the scripted options are both incredibly risky because it means when the images were captured, the game was not running in real-time with all the systems implemented. You might find out later that the game is never going to be running well enough to include all those effects which were scripted or rendered out of the engine for the trailer. This is likely how games like Watch_Dogs ended up looking different from their trailers and early demos.

This last shot in the Firewatch trailer was animated manually by James Benson. This moment doesn't actually happen in the game, it's actually a hybrid of two different scenes (though I never saw any complaints.)

Animate manually

Lastly, there's the possibility of creating images using no game engine or programming at all. Just using 2D or 3D animation programs to create the images and hoping that later on the game will look like that. I've seen this used most when it comes to 2D assets like UI, or if a game is mostly 2D images like in a visual novel. This is probably the least risky way to use manual animation for a game trailer, but I'm sure developers could get themselves into trouble if there are too many effects layered upon a scene which then cannot be implemented later.

I think to a lesser degree it's done for 3D images, but I don't know this for sure. If a game is going full on 3D animation then this technique is better used for internal pitch videos like the Kill Zone 2 trailer rather than in a trailer released to the public, because people signing on to provide funding or other services need a more general sense of the game's feel, but players watching a trailer on YouTube or a Steam page are going to want to see something as representative of the finished product as possible. 

Unfortunately There is no one way to do compensate for a game's unfinished state, because every game is different and also locking down design or art assets of the game at an early stage might be detrimental to the final product. So every developer has to do their best to make early trailers look as close to what they think the game will look like when it ships.

Also, the level of scrutiny on each game's graphics will vary wildly depending on its audience. For example, high end PC games are largely sold on the fidelity and improvements of their graphics, but indie games tend to sell based on their art direction, unique game mechanics or personal stories.

If you're making a trailer for a game which is very unfinished it is possible you'll have to use one or more of these techniques. I hope it helps you navigate the difficulties of announcing a game very early in development, or just makes you more cautious about getting too hyped from early marketing materials.

I also want to stipulate that I believe the vast majority of game developers and publishers are not in the business of willfully deceiving potential customers by making trailers and demos look as good as possible and then "downgrading" the graphics in the finished versions of the game. There's already so much distrust around differences which occur because of technical limitations, poor planning, and myriad production realities which can occur despite the best planning. Considering the vitriol and hate spewed on the internet, it would be absolutely unthinkable to try to "fool" the audience. It is on the marketing people to be as honest as possible, and it is also on the consumer to be make informed decisions about their game purchases!