Good ol’ Teal and Orange

I wrote this as a reply to somebody in a forum recently and I thought it was worth re-posting here. The OP was asking about why we always see the same color palettes – Teal&Orange and Brown – over and over again. Here’s what I replied:

The reason a lot of movies are color graded Teal/Orange is because skin tones are in the orange range of hues. If you take everything that is not faces or in a similar color space to skin and push it in the opposite direction (complementary colors, natch), you get teal. Now, your eye is naturally drawn to look at the faces and not the unimportant scenery. Warm colors draw your eye more than cool.

This works for movies because the colorist can key the grading off the faces in every single shot. If the face is lit by blue skylight in this shot, that’s ok, just key the grade off the face. If the face is lit by warm incandescent light in this shot, that’s ok, just key the grade off the face.

This technique is tougher to do in games where the color grade just gets baked down to a single LUT for all lighting situations. That’s the answer to your question. Instead we usually go for a more general color grade – one more balanced, or one more warm and appealing (the brown look).

Another reason teal/orange works for movies is because the faces are almost always the most important thing in the composition. Faces are warm and warm colors draw the eye, thus teal/orange. In games we often care more about an entire character wearing a costume, or an inanimate object. Often in games we have a lot of dirt.

In movies, not every object in the background has been colored specifically to fit within a palette. In games you can texture everything to a specific palette (this is not always done in games, but it is possible). The teal thing is more necessary in movies to unify the palette of the scene. I really don’t care about the doors behind Iron man in the shot above, so make them all teal to blend in with everything.

When games do teal/orange, it’s more often an attempt to ape the aesthetic of a movie and elicit the same tone.

Why don’t we see more color variety in games? Why does it have to be just two colors (teal/orange) or one (brown)?

Lots of the games you reference go for an aesthetic of high detail. High detail is often targeted in AAA art direction as a shorthand for MOAR GRAPHICS – we have the best tech, the highest fidelity, the most future in our codes. Sometimes, this is actually true (Gears of War, Battlefield). Often, this is weak art direction.

Lots of these games also have high value contrast, like Saving Private Ryan, to give more of a ‘wartime’ feel. They are super explodey combat games, so this is appropriate.

If you have high detail and high value contrast, adding in a complex color palette would be a mistake.

It’s a matter of contrast. Artists use contrast to draw the eye. There are lots of forms of contrast:

- Value contrast (black and white)
- Color contrast (lots of different hues, strongly complementary colors up against each other)
- Edge contrast (lots of detail)
- Temporal contrast (any of the above over time)

If you have high contrast of all types across the whole scene, you have visual shit.

So, the games you see with more colors are usually the ones with less value contrast or less detail.

The Complexity Wall

There is a complexity wall you hit during development. Even if you are on a small team, there is a point early in development where the complexity of the game makes it too cumbersome to add new systems. Either there are too many systems already there, or too much content built on those systems. What this means is that the initial period of development is a golden time. It’s the one time during the making of a game when it’s easy to change something.

It’s hard to understate how important this idea is toward making good games. Game design is more about process than theory, much like John Carmack would say it’s more about execution than the initial idea, and much like how Steve Jobs said product design was more about process than the initial idea.

Getting the right pieces in place before you pass beyond the complexity wall is what makes the difference between a good game and a bad one.

You have a choice on what to spend that early time on. You can spend it on making a new graphics engine. You can spend it on prototyping a fresh and exciting new core loop. Or, you can spend it on animation systems and cameras for narrative storytelling. Fairly quickly, you will pass the complexity wall and that thing you chose to work on at the beginning will be the de facto defining element of your game.

The foundation of the Cerny method was that, for action adventure games, you should spend that golden development time  on the 3 C’s (camera, control, character). Mark formalized preproduction as a way for teams to stay outside the complexity wall until they had a good foundation based on those three elements. If you didn’t have a great “First Playable,” or “Vertical Slice,” you did not get greenlit and move on to production. Staying out of production meant staying on the cheap and flexible side of the complexity wall.

There are many studios who are repeatedly successful because they are good at focusing down on one thing early in development, one thing that is good to work on outside the complexity wall for their kind of game, like Bungie with its sandbox. I like to belive Bethesda Game Studios has this quality.

If you choose to focus poorly, you can end up stuck with a bad foundation, and then there’s no amount of time, money, or talent that can turn that project around incrementally. The inertial resistance to change beyond the complexity wall is too great. At the beginning of development your game is a small, nimble go-kart. Beyond the complexity wall, it’s a thousand ton train and you ain’t turning that thing around, even if you know it’s headed to shitville.

I’ve been on a few trains to shitville (not where I currently work), and it’s a terrifying, helpless situation.

If you scale up a team too soon, as is fairly common for teams up against a looming release date, you’ve taken a voluntary pass on the golden period outside the complexity wall.

Now, there is one way to tear down the complexity wall, and it’s relatively common thing to hear about from the best games ever made. Throw out the bathwater and start over. 

Yep, just toss it out. Reboot. Halo started out as a Mac RTS and ended up as a console FPS. Large portions of the original game were thrown out and started from scratch. Half Life was famously thrown out and started over, as was Half Life 2.

The sunk costs fallacy (and budget politics in big companies) might lead you to believe that throwing out the current version of the game is an expensive idea. It is surely daunting, but it’s the only way to regain that ability to change your game significantly. If the current game sucks, it’s often cheaper to throw it out and start over with a smaller team with a smaller burn rate.

Luckily, throwing out the whole game and starting over is not necessary if you choose the right things to focus on early in development. Most great games don’t need to get made twice before they are released once. Usually, there’s a proven previous game that gets sequeled or made greater (Mario 64, Battlefield 1942, Uncharted 2). Or, there’s a great initial prototype or mod to build on (most indie games, most Valve games).

The complexity wall is important to recognize because it’s the most common downfall of producer-driven development. The fallacy of the producer-as-creative-director is the idea that he can steer the big ship, make directorial decisions mid-development that direct the team back on course towards success. Producer-driven development is the hallmark of big publisher owned teams, and it’s why you don’t often see innovation there.

People think creativity is inherently lacking at the big companies. That’s not the case. Creative risks get taken all the time on big teams. The problem is when the creative idea gets adopted on a big team, it needs to be shepherded by a producer. Guess where that producer lives? Behind the complexity wall. No matter the level of selflessness or skill that producer has, she will have a epic challenge opening a hole in that complex game to fit in the creative idea.

The complexity wall is why we can’t have teams assemble on the fly like hollywood movies. It’s why a stable tech foundation works to a team’s advantage. It’s why you have to have your core loop understood before anything else, and it’s why you need to understand the tone of your game before you grow your team.

The complexity wall is why game design is more process than theory.

Game Art Style Presentation

I gave this talk at BAF Game 2012 and I thought it was worth recording before it left my head. It’s about art direction in video games, or at least how I think about art direction in video games. Since it’s a fairly big subject, this is just my take on it.

In 3 parts on YouTube:

Valve is coming to the living room

Gabe confirms.

I wish I hadn’t made my post on the future of the living room so huge, burying Valve on page 4. I suddenly have the desire to update the picture with a companion cube submarine emerging from the depths between the Apple and Google ships. Here’s why we should all take Valve seriously: They are insanely efficient at leveraging their small size.

Valve has the sharpest outlook on how a software/services platform should be run in the game industry. Despite being a small company at around 300 employees, Steam is dominant on the PC. Michael Abrash’s recent post illuminates how Valve is optimized around delivering maximum value per employee. Valve’s 300 is more directly effective than, say 300 of the people working on the next Xbox (efficiency #1).

They also fly the flag of open platforms and direct access to customers. They believe (as I do) that connecting developers to customers as seamlessly as possible will create the best experiences and ultimately the best platform. Oppressive certification is a waste and a platform hindrance. TF2 is a massively better business on Steam than it ever was on the 360. That direct connection creates an efficient platform, which requires less scale on Valve’s part (efficiency #2).

An open platform device doesn’t have to be as complex and monolithic as an Xbox or an iPad. The graphics hardware race is largely irrelevant on anything that’s not mobile these days. Valve is capable of taking PC components, putting them in a box, and using Steam as an OS* (efficiency #3).

An open platform device doesn’t need a huge launch with a massive marketing push. Forget retail. Valve is big on market experiments and I can see them putting out a Steam box in a bare-bones version 1.0. From there they will learn, iterate and scale up (if the market supports it). It’s an antithetical strategy for a hardware platform launch, but it’s gospel for a software platform launch (efficiency #4).

The thing about the next generation is, it’s going to be much more about software and services than hardware. Valve is in a great position to disrupt the big players purely by virtue of their dedication to open platforms. Not to mention, they understand the importance of a good controller to the living room experience, something I’m not sure even Apple gets, what with their undying devotion to the all-powerful touchscreen.

I once made the mistake of expecting Steam to take over and disrupt the PC market overnight. I was right in my predictions but utterly wrong on the time scale. If anyone can do this, it’s Valve. Just understand, it’s going to happen on Valve Time.


* Most games currently on Steam use DirectX, which is part of Windows. Valve (including Gabe himself) are working on a Linux port of Steam – Why? There is probably a solution to the DirectX issue either by either launching their Steam Box without the DirectX titles, or by using a networked Windows PC (already running Steam) to render those games over the network back to the Steam Box.

Next Gen is Terrifying?

I don’t think Naugty Dog has anything to worry about. Next gen will be less about a graphical leap than ever before. Sure, the hardware will be much better, but the initial graphical leap in game tech and art will be imperceptible next to current PC games like Battlefield 3.

So what’s the selling point? Certainly Sony can’t put out a $900 PS4 and expect anybody to buy it on the back of “MOAR GRAPHICS.”

Next Gen will be defined by the declining retail market more than anything else. MS/Apple/Sony understand that they are making a platform for digital content delivery, not just big AAA games, but also small XBLA games, free to play games, social games, but potentially more importantly: TV and Movies. A new way of interacting with your tv and games, that’s the selling point.

Apple is doing this which means MS and Sony need to strike preemptively or be disrupted. Believe it or not, there are lots of people that would happily never own another “game box” if their TV played Wii Sports and Just Dance. The PS2/Wii era market penetration is the realm of iTV.

I work on AAA games so please don’t remind me that AAA blockbusters aren’t going away. I know, I know. I like those games too. Yes, we will always have them and they will get more uber in the future. It’s great.

But that’s not the highest end of the market. The highest end is $100/month TV & Movie subscriptions that people currently pay to the likes of Comcast and Verizon. The highest end of the games market is not even Mondern Warfare 3′s Billion Dollars, it’s WoW’s Several Billion Dollars.

The next gen is about making and owning the living room software & services platform. Look at how well Apple and Google have done on the backs of iOS and Android. Everybody wants to do that in the living room. The competition is not between Xbox 720 and PS4, the competition is between:

- iTV
- Xbox Live
- Google TV
- Steam (which is remarkably well positioned to be relevant in the living room)

Platforms. That is what matters and is what will define the next gen. The level of uncertainty around the future of those platforms is far more terrifying to me than “how much RAM will the box have?”