Scott Warren: Building Light For Games

We’re happy to present you our interview with Scott Warren – senior lighting artist at Raven Software. With this article, we open a series of talks devoted to the creation and use of lighting in video games. In this post, Scott talked about the function of light, his workflow and the GI-revolution.

Introduction

Greetings! I’m Scott Warren and I’m a Senior Lighting Artist at Raven Software. I grew up and lived in Atlanta, Georgia for about 29 years before moving up to Wisconsin to join Raven earlier this year. My career has run 11 years so far and in that time I’ve worked on Global Agenda, EVE Online: Incarna, DUST 514, World of Darkness, Lichdom: Battlemage, and more recently, Call of Duty: Modern Warfare Remastered.

My original life plan was to graduate high school and then become a computer hardware engineer in college. That fell apart quickly when I realized just how much math would be required to accomplish it! Photography and visual art had always been an interest of mine, as well as a deep love for movies, so I dropped out of college at age 19 to take my own path into the game industry. It seemed like a fun marriage of science and art, and that’s what I wanted to do. In the year after I dropped out, I focused on bringing my digital art skills up to speed by studying Photoshop, Softimage XSI, Maya, and a few other tools to develop a deep understanding on how best to use them all together.

I was eventually comfortable enough with my portfolio that I began applying to several companies in Georgia, and Hi-Rez Studios ultimately helped me get my foot in the door to work on Global Agenda.

Function of Light

I would say that the primary function of light in games, and indeed in any medium, would be emotion. It’s there to push a certain mood or feeling. It’s the difference between a happy, sun-lit field of sunflowers and a dark, wet alley in 1950’s New York filled with steam, mystery, and silhouettes.

In a utilitarian sense, most casual games use lighting as a navigational aid. “This doorway is brighter, so come this way,” sort of thing. It’s an easy way to carve a path through a level that you want people to follow. Not unlike in reality when we stick to the side of the street that has street lamps; it feels safer and we can see what’s ahead of us.

On the other side of that, you have games that use light in a very noir, moody kind of way. There’s a lot of things that you can’t see and your brain fills in the gaps. It feels a little more mysterious and more thrilling that way. That’s the approach that I’m more a fan of, is lighting to show crucial information, but not to over-light so that things feel flat and you remove some of the mystery. Contrast can be very powerful in evoking emotions.

Something that I realized early on with lighting and why it’s so critical, is that it’s the difference between a good image and a great image. You can have the best ZBrush sculpted environment that’s filled with physically correct materials, but if the lighting is mediocre, the whole thing falls apart. Objects and materials look only as good as the light falling on them. Conversely, you can have a sloppy model or environment that’s built using hacked-together assets which can look gorgeous with carefully authored lighting.

I’d say that games need (good) lighting to bind everything together. It’s the X-factor glue that people don’t particularly notice when it’s done well (much like visual FX in movies), but if it’s missing or just plain weird, people will certainly feel it. Our job as lighting artists is to be invisible on screen so you accept the world you’re in.

Peculiarities 

Real-world lighting can be either an additive or a subtractive process. In a dark studio, you add lights one-by-one to build up the image to what you’re trying to achieve. If you’re outside, you do a lot of subtractive lighting to add contrast back, like adding large black flags next to actors to shadow their faces to get dimension back if they’re outside on a cloudy day, for instance. Games are almost always additive since every world starts pitch black. Because we’re free to add virtual lights anywhere without being limited by walls or ceilings, there’s a tendency to over-light things and loose contrast. That can feel very “gamey” since its apparent that you have light hitting the scene from somewhere outside of it. Personally, I like to stick to motivated lighting as much as possible, which is lighting that comes from a source that exists in the world somewhere. I’ll take advantage of window frames as light sources before I sneak some sources into the ceiling, or play up the bounce light from a pool of sunlight in the bottom of a cave.

Thinking about how light would move within a real space helps you keep a consistent approach to how you light a game environment. The big thing to keep in mind is that lighting is a universal constant. Its behavior doesn’t change regardless of the medium. It does the same thing in reality as it does in CG animated movies; bounces, diffracts, reflects, diffuses, and gets absorbed. I try to recreate those behaviors as a starting point, and then push and pull here and there for artistic license.

My overall opinion being: if I couldn’t get away with it on a sound stage, I wouldn’t do it in a game world. This mostly comes from my fascination with how movies are lit and I try to emulate some of those rules and restrictions for light placement to get a similar feeling with the visuals.

Global Illumination

What’s great about modern global illumination techniques is that they’re all starting to arrive at roughly the same final image, which is great for a lighting artist because it’s one less thing to have to worry about. You have Lightmass in Unreal Engine, VXGI in Cryengine, Geomerics Enlighten… there’s quite a few real time options. Half of them require pre-baking to render the results to
textures, and the others are totally dynamic.

The challenge with getting the most out of GI is that textures and materials need to be authored correctly to see the effects of light bouncing around the world. PBR materials and physically correct lighting values take care of this when done correctly, and tools like Substance Designer make it easier to get going quickly. The remaining challenges would include the struggle between baked and real-time solutions. Baking your GI usually gives you a smoother, more natural looking image with a lower performance overhead since you’re not computing the effect every frame. It also takes up more disk space since those textures have to be stored and then read into scene materials. Dynamic GI is great from a workflow perspective since you don’t have to stop to bake anything, but it does have a higher rendering cost which might alienate some machines from being able to run it.

I think as hardware gets more and more powerful, we’ll see everything move to dynamic solutions. Baking does usually look nicer, but I appreciate the speed of iteration with a dynamic solution.

Workflow

Reference! Everything starts with reference, and lots of it. Whether it’s concept art from the team or movie stills, or even photos we’ve taken ourselves. The look itself isn’t so much the focus with
initial lighting discussions. It’s hard to target a specific reference and say “We’ll make it look like that!” because it might not work for a particular story you’re telling or it might make gameplay difficult. Sometimes a concept image doesn’t work well when translated into the game world, and so you have a constant back and forth dialog about the mood as well as how it facilitates gameplay.

A lighting artists’ primary role in games is to help along gameplay. If you can’t see enemies or a story beat is too dark, the lighting is working against the goal of it being a fun game to experience. So you start with the designer’s goals for a particular environment, figure out where the major beats are within that space, and then work outward from there. A high-combat area will usually be more brightly lit than a transition area where there is no fighting, for example. Those in-between moments are fun to light because we don’t have to worry about player visibility, we can just make pretty pictures. There are plenty of opportunities to embellish parts of the world as you wish, but you’ll primarily be focused on keeping things readable, balanced, and interesting at the same time. And unlike a movie, we have to think about how things look from all angles since you’re able to roam around freely without being locked into a set camera path.

The difference between a single player and multiplayer experience can also be huge as each one has different visual requirements. Single player campaigns tend to be much more movie-like and contrasty. Multiplayer games have to split the difference between movie-like and fairness to everyone in the match. If you go too contrasty in multiplayer, half of the players won’t be able to tell what’s happening and they’ll be at a disadvantage than people with brighter displays. If you go too flat, then it feels dull and uninspired.

For a given environment, however, the actual lighting process is relatively straightforward: A) Dial in the sun, sky, and atmospheric values. B) Do a primary lighting pass through the world (any man-made sources, or navigational lighting). C) Add supporting bounce lighting as needed to soften contrast or add visual interest in flatter areas. D) Playtest the crap out of it to find problem areas. E) Fix problem areas and final overall tweaks. F) Color correction and post processing like bloom, AO, etc. Heavily simplified, but that’s roughly the process that happens from start to finish.

Advice

I’m not sure that there is a right or wrong lighting for a given situation or game. In the same way how there isn’t a right or wrong piece of music; there’s music that works better than others for something, but you’ll never find the “right” song. Lighting is strange like that and it requires that you play around with different ideas. You’ll know you’re getting close when you stop noticing the lighting and start becoming engrossed in what’s happening in the world.

For anyone who wants to pursue lighting as a career, I would encourage them to study photography and to become avid photographers themselves! Photography at its core is the recording of light at a moment in timу (or a chunk of time if you get into video recording as well.) Take tons of pictures. Spend a lot of time learning how a camera sees light since its very different to how we see it. Take lots of really bad photos and figure out what makes them bad, then re-shoot to make them better. Sit someone down in a chair and light their face using a bare bulb moved around to different positions to see how it affects the look of their facial features. Use translucent shower curtains as diffusion and see what happens when you shine light through it to your subject on the other side.

So much of lighting is experimentation and trial and error. And a huge dose of observation. Look around you wherever you go; standing in line at the store, walking through the airport, wherever it may be. Light is almost always doing something interesting and you get these ideas to try something based on an observation you made. Bounce desk lamps off bed sheets and different colored clothing to see how the colors bounce into the room.It’s such an interactive thing and I hope people play around with it and have fun! That’s how most of my learning happened, and continues to this day. There’s no rules to any of this stuff, which gives you the freedom to do whatever you like.

Scott Warren, Senior Lighting Artist | Raven Software