You have worked on a number of titles since you starting working in the games industry, but what do you feel have been the main changes and improvements over the years to PC titles from a developer's perspective?
PC hardware and game engines are constantly evolving. From a developer's point of view this gives you a lot more possibilities and options, but it also increases the scope of your project. There are certain features customers expect from a full price game today that were not achievable five years ago due to technical constraints.
Normal maps on every asset, dynamic lighting, breakability, large scale multiplayer and a more or less open world are standard for shooter games these days. Six years ago a developer would focus on one of these areas, whereas today you need to implement all of them to be able to compete with the other products on the market.
While these features certainly make your game a lot more interesting, they will also vastly increase the time and money it takes to build it. For a game like The Guild 2 we needed about four artists to create the entire environment whereas today you would need around ten to twelve people just to build the environment assets themselves.
More powerful hardware and newer engines like CryEngine3 make working on current games a lot more fun and interactive. Being able to move objects around or to adjust the lighting in real-time and then to jump in-game with the click of a button without having to wait for the level to compile is just great, and allows for much faster iterations on your game.
You have mentioned areas in which games have evolved over recent years, but how do you see them changing over the next six years or so?
The new DX11 graphic cards are becoming more readily available and the new console generation is not too far away either, so the next big feature we can look forward to that actually impacts content creation is real-time displacement mapping.
At Crytek we are currently working on a DX11 patch for Crysis 2 that will add this feature to the PC version. It's really stunning to see the geometry that you sculpted displayed with millions of triangles in real-time. This is going to be a big step up in visual quality and can be used to generate some truly stunning environments.
I'm also hoping to see new features regarding breakability. It's very time-consuming to create destroyable assets and I'm hoping for a procedural solution to make the life of artists a bit easier. The PhysX technology that Nvidia is working on looks very promising and could be a big help when it comes to adding breakability to your game.
In the long run games will reach a level of visual quality where they will hardly be distinguishable from movies. Thus they will have to stand out based on their features and not by the quality of their visuals. Things like artificial intelligence, storytelling, creative gameplay or style will become even more important and I hope to see major improvements in all of these areas.
Crysis 2 has been very well received by both the media and the public alike. When given an environment to build what are the typical approaches you take, and what constraints affect the way you go about organizing the model components and texture templates?
We usually start out with a whitebox model of the environment that contains all features relevant for gameplay in this area. Things like cover height, door sizes, scope of the area and different paths to access it are defined.
After that the scene gets passed on to the concept art department, which creates a concept painting based on the whitebox and the vision of the art director. They will define the mood of the scene and the main features. Depending on the object this can be a quick overpaint of the whitebox or a fully realized concept of the entire area with shots and paintings from different angles.
We then export the whitebox geometry from CryEngine to 3ds Max and start building the asset based on the existing block-out. Depending on the size of the scene and the time available to build it, I usually start out with splitting the model into modular pieces with tiling textures so that I can create the asset as fast and efficiently as possible.
The amount of different modules and textures used for an environment are defined by the scope of the area. Indoor areas can usually be a lot more detailed than outdoor areas because there are fewer objects and a lower view distance.
We try to reuse as many textures and models as possible in a scene so that we can keep a smooth frame rate on all platforms. CryEngine has many different profiling tools that show the areas where your environment is over budget. It all comes down to keeping a healthy balance between the amount of characters, lights, objects, physics and view distance in a scene. If you want a really high res asset you will have to tone down the number of characters or unique objects and vice versa.
When you say you start out with a whitebox model, does that mean that the environments are actually designed in 3D initially and not created by the concept art department? Does this mean that artists like yourself have input?
Yes. We usually start out with a whitebox mesh using primitives in CryEngine and then use these block-outs to create concept art if needed, or just build the asset without concept art if it's based on a real-life object and we have enough reference material. Regarding the input, the initial block-outs are usually done by the level designers or game designers themselves in order to make sure the area plays well. After that, every artist can implement his own ideas into the mesh as long as it does not interfere with gameplay and fits with the mood of the scene.