Call us boastful but Burnout games have always looked great. So when faced with the task of creating a true next-gen open world our artists were keen to ensure that yet again their work would shine. Here Scott tells us just how the art team built Paradise.

Looking back on the production of Paradise City, I think our biggest challenge on the world team was that of discovering what "next-gen" actually means. From the outset, we knew what we wanted to achieve from a gameplay perspective. We have the large world where you can drive to anything you can see, we have more shortcuts and stunt routes than any other open world, and we run at 60 frames per second. But what does a "next-gen" world actually look like? More specifically, what does Paradise City look like? In the space of five Burnouts, the world team has modelled well over a dozen US cities. If Paradise City were just a high-resolution version of those, it would have been a failure. We started Burnout Paradise roughly two years ago, and one of our earliest tasks was to answer these very questions.

Initially, we looked to other games. However, two years ago, the only games that were present were launch titles, and the next-gen launch titles had a bit of a common theme. Most of them were a bit brown with the bloom effects turned up. Many were just PS2 titles with high resolution textures and over-emphasised normal mapping on everything. We covered a lot of these themes on Burnout Revenge 360, but for us, that game was very much the warm-up act. Paradise is the show. Because of this, we couldn't take our cues from other titles. The "launch-title" look would not stand the test of time, and everyone (ourselves included) would expect more from Criterion's first true next-gen game. We didn't intend to be the best-looking open world racer around. We intended to be the best-looking open world racer around in two years time when the game would be released. This sounds like a very obvious and aspirational goal, but in honesty, it's very tough. Being the one who sets the bar rather the one who reaches it means that you never have the luxury of stability. For example, for the first time since Burnout 1, the code team rebuilt the tools and the engine from scratch in order to make the best possible game without carrying legacy code. At the same time, we were making the world. The engine was constantly changing, and as such our art budgets were constantly changing. No one could tell us how many polygons we could have in an area, or how many physics objects we could place, and the question of whether or not we could do normal mapping or ambient occlusion arose quite a few times.

The parking garages. Initially included as a potential gameplay test, they are now prevalent throughout the city.

For the first year and a half, we had to just throw art at the game, and wherever the game crashed or the frame rate dropped became our limits. There was a point where pretty much every art asset became its own mini-R&D exercise, where we would just try things to see if they would work.

Nonetheless, this didn't present too much of a hindrance to the world team. Even though historically, Criterion has always been a technically-based company, the world team makes art. The tech is just a medium, not an end in itself. This is why the "launch-title" look didn't survive. In the past, the pressure to be "next-gen" led the industry to an emphasis on technical showcases (i.e. "how many shiny things can we have on the screen?"). In Burnout Paradise, we already had our technical showcase. It's a 60Hz open world, we just had to make it look good with art. That said, we did have some new tech to work with. For example, we could do real-time lighting, dynamic shadows, and we even had shaders that were written for specific purposes. For example, there was a shader designed entirely to make the roads look good, one to make the water look good, one for the trees, one for the buildings etc, a far cry from our last-gen material system.

The Paradise City road shader - 50% crispier than the roads in Burnout Revenge 360, at around half the rendering cost, with textures that do not visibly repeat.

All of this notwithstanding, the world team made Paradise City look "next-gen" by making it believable. In previous Burnouts, buildings were placed in a way that looked good, or that fitted the reference images we had. In Burnout Paradise, every building had a purpose. The game is set in a single, self-contained city, and the player would expect to find the elements that make a realistic city. Every city has a city hall, or a police HQ, or a residential area, and so on. I remember during the pre-production phase of the project, one of the world leads said that every landmark in the game should have an unofficial story attached to it (why is it in the world, when was it built, why is it shaped the way it is etc). For example, we have a car factory in the world called "Nakamura Automobiles", which is also the name of one of the fictional brands that we have on the cars. This gives the car factory a context and a purpose.

This sense of believability is extended in other aspects of the world. Every street has a name, and the entire city is split into separate neighbourhoods (the latter of which are actually signposted in the world).

Also, we have billboard advertising in the world, and all of it actually advertises the fictional landmarks in the game. For example, you will see advertisements for the Rayfield Hotel or the Film Reel cinema, both of which are in-game landmarks that you can actually drive to.

Multiple versions of the same shops that are not just duplications of the same buildings

Similarly, Paradise City has chains of restaurants and shops that you will see throughout the world (e.g. the "Caffeine Hit" coffee shop, "Daley's Donuts" et al).

Also, there is a sense of scale in the world, and the feeling that the world still exists even when you are not directly affecting it. We have always simulated this in previous Burnouts, but in Paradise, we have taken this idea to the next level. For example, if you were to park the car at any point in the world, you would see the traffic driving by you as before. You would also see the trees swaying beside you, and the normal-mapped waves of the distant river rolling by. You would even see other traffic cars on overhead freeways where you couldn't immediately drive to them. In the higher points of the city, you can see most of the world at once. For example, from the wind farm on one side of the city, you can see the downtown area of the city seven kilometres away, with the full knowledge that you could actually drive to that area and explore it. Achieving this wasn't easy. Our streaming and frame rate budgets certainly didn't allow us to draw the entire world at once, and we quickly learned why most other open world games are just corridors of buildings. But we achieved it, not through any special next-gen trickery, but through the same fundamentals that we have always used.

Fun with draw distances

This all sounds very trivial, but the result is more than the sum of its parts. If we made a world where every street just repeatedly had the same five buildings in a row, the player would notice, it would become distracting, and Paradise City would just feel like another bland video game environment. Conversely, if we made a world that satisfied the player's visual expectations (however subconsious those expectations may be), then the world would become immersive. It would become next-gen.

Gallery

Related Articles

Making The Tracks
In this series, several key members of our world art team will take you behind the scenes, track by track, and show you how the world of Burnout Revenge came to life.