Making great art, quality software, or engaging gameplay… each of these is challenging enough by themselves. So, it stands to reason that combining these disciplines into a single medium is not only vastly more compelling, but also vastly harder to do. Juggling the competing priorities, desires, and dependencies of all the smart and talented people brought together by a game production is made even more complex by dramatic differences in interpretation, language, and philosophy. These difficulties give rise to unique disciplines in game production, such as Tech Art.
My name is Raymond Stewart and I am the Lead Technical Artist at Xaviant. My role on Lichdom is to help the Art Department fulfill the vision of Art Direction within the constraints of Engineering’s technical requirements, and demands of scope driven by Production and Design. This requires enough awareness in each of those disciplines to facilitate communication, best practices, and management of dependencies that lay at the core of integrating Art production into what is really a software product.
Technical Artists help the Art team in a variety of ways. We automate the tedious tasks that make the work of the artists unnecessarily difficult, and support the artists when processes are not working right. We write tools to make putting art into the game easier, and make recommendations on best practices to help manage dependencies within the production pipeline and over the course of the production life cycle. However, my favorite aspect of Tech Art is helping the artists make the most of the technology provided by the game engine.
Making art for a game engine is one of the most abstract processes that confronts a 3D artist. 3D Art as a discipline has existed for some time and is host to so many techniques that it is not easy to know from one game to the next which apply, or if the techniques required even exist yet for real-time rendering. Many of the features are often in development at the same time as the art. It all comes down to the specific features of the game engine and the art direction. To make the process viable, tech artists will prototype art to present the feature, and define the workflow required to get results.
On Lichdom, we are using CryEngine. This is a very powerful engine with lighting and rendering techniques on the cutting edge of game development. So for us, exploring its features and fitting them to our art direction is a major undertaking with many challenging new insights for artists, both in terms of visual fidelity and performance.
For instance, most games have made use of Diffuse, Specular, and Normal texture maps for many years. Now, High Dynamic Range rendering, Ambient Occlusion, and Secondary Lighting techniques are becoming standard fare. However, not only are these techniques higher quality, they can also be a drain on graphic performance. Today, an object in-game might have to be drawn on screen as many as five times to make use of these advanced techniques. So, finding ways to improve the performance is a make or break issue in getting cutting edge quality. And the solutions often fly in the face of previous lessons learned on last generation games.
Consider this image…
CryEngine provides many powerful parameters to control things like Specular Reflections across the surface of an object. In most cases this is sufficient. In the case of complex models, like this character, a fixed parameter can be a problem. Surface details can be too small to be modelled efficiently in 3D, but still affect the apparent quality. However, if these small details are not represented in the geometry, they can’t have separate surface materials. The result is that surface material parameters like Specular Fresnel get shared across the whole object and negatively affect the quality of the art. Major discontinuities in how these surface details appear get ‘glossed over.’
In the image and video above, the character on the right shows how one set of material parameters for the whole object can lead to odd results. On the left we see what happens when we vary these parameters with texture maps. Its much easier to get the specific characteristic you want from Metal, cloth, leather, without having to have separate surface materials defined for each. Here are the textures that produce the variation of Fresnel and Shininess on this character.
Admittedly, they don’t look much like Art when viewed like this. What makes these textures so bizarre is that they are not really ‘image’ maps in that the colors you see don’t mean anything in terms of color. They are actually an array of data that is used as inputs into the parameters of the surface material. They drive obscure values in lighting equations that have nothing to do with anything you directly see as ‘color’ on the object. Here is an example of ‘color coded’ values that we apply to these parameters, what they are intended to mean, and an example of their application to a character…
The result is that light behaves differently across the surface from one pixel to the next according to the texture values. Its the creation of textures like these that present artists with very novel and counter intuitive problems that challenge their understanding of the medium. Helping the artists meet this challenge of understanding the relationship between the images they create and what ends up on screen is one of the most rewarding parts of this little slice of game production.
We have an incredibly talented Art team. It is amazing to see what happens when we squeeze all of the power out of CryEngine and put it firmly in their hands. So, when you look at the art of Lichdom, know that its not just about the tech. The tech doesn’t change the art: it changes the artist.