My first look at GameCon was reviewing the level that was blocked in by our game designer, Josh, during my first month with Proletariat. I knew it was going to be one of our most challenging levels; not only was it modeled after a large convention space like Boston Convention and Exhibition Center (BCEC), but it would contain tricky interior lighting from the hall and accent lights, motion graphics for large screens, poster graphics, physics objects that could be kicked around or knocked over, and tons of various greeble in the corners and occupying display cases and counters. Creating a balance between the production of the level and maintaining performance while being mindful of a tight deadline meant that I had to work as efficiently as possible, and that required doing as much pre-planning as I could.
Usually, we start by compiling reference images in Pinterest to get a high-level idea for what would be expected in the level. This allows us to collect various types of machinery or building references and review them together. However, in the case of GameCon, we had the opportunity to visit the BCEC firsthand (Michelle Konnath from MCCA was kind enough to give us a behind-the-scenes look between expositions). I knew the potential challenges beforehand and where I needed more detailed reference images to help aid in my pre-planning. The tour allowed me to gather the much-needed reference images to help fill in the blanks of the level and to capture the essence of the convention center.
While touring the space, our art director, Damon, and I pointed out various architectural elements that would make great focal points for GameCon. A usual step in the pipeline is for Damon to take three to four screenshots of the level that show the most objects or important elements that could be fleshed out, along with a link to the reference images in Pinterest, and send them to our environmental concept artist, Sylvan. But for GameCon, Damon created the concept images himself.
For concept images, we generally focus on bold profiles, shapes, and color palettes that could make the level unique and make the objects stand out. We often apply three layers of color separation: the backgrounds receive almost pastel colors, game play areas receive the next level of saturation (while still keeping walkable areas brighter), and finally the characters (and, ultimately, the VFX), so gameplay objects—such as point-collection—always visually stand out the most.
While the concepts are in the creation process, I begin to breakdown the blocked-in level from Josh with a more detailed eye. Drawing inspiration from the image board, I begin to export the unique shapes out of Unreal 4 and, at the same time, write down what these objects might be in an Excel sheet. I look to divide the assets into three major parts to help me tackle the amount and complexity of potential objects: Primary (structural), Secondary (vehicles, large set pieces), and Tertiary (deco/decals). These objects are imported into Maya and I begin to set up my scene. I organize objects by their various types and assign them to individual layers, which helps me keep my work organized.
Whereas the Secondary and Tertiary objects are fairly straightforward and often don’t require detailed geometry planning, the Primary or Structural tasks nearly always require a lot more planning. Oftentimes, these are large pieces that make up the backbone of the scene and are tiled and repeated; corners, reverse corners, straight walls, floor tiles, and ceiling pieces with their variants are planned out. You can almost think of the pieces as a Lego set. Trying to find the most efficient way of approaching these assets can be time-consuming, but spending more time planning at this stage can save a huge amount of time in the long run. Aside from being the most efficient way to create these assets, breaking them up into these pieces has another benefit: Unreal Lightmaps.
What are Lightmaps? To help aid performance in Unreal, static objects won’t require a dynamic shadow like a character would. Because these shadows never move, they’re baked onto the objects themselves, utilizing a second UV channel located on the object itself. By making these pieces as smaller static meshes, I can assign a lower resolution light map to each of the objects. If I find that an interesting shadow is on one of the pieces, I can duplicate that mesh and increase the light map for that one area, allowing me to strategically raise the light map resolution in just the area that really needs it rather than waste the memory on areas that don’t require it.
The next pre-planning stage for GameCon was the least fun, but producer types love it: entering all these assets from my Excel sheet into a product planning system. In this case, we used Redmine. This helped me organize my thoughts and assign priorities from what I felt was the riskiest and therefore should be addressed first (usually the Primary [structural] tasks), all the way down to the Tertiary (deco) tasks. Time estimates were then assigned and checked against the schedule to see if this would fit within our timeframe and to call out possible issues or tasks that I perceived would be more complicated that might actually not be as important to the designers. This step, while tedious, accomplishes a very important part of game development, which is communication. It announces what I plan on creating, gets everyone to see my priorities and lays it out for the team to see. Many times, issues are caught and priorities adjusted before I begin working. It’s a step that gets overlooked too many times.
With the concepts created by Damon complete and the preplanning grunt work set, I then begin to block out the shapes. This is a lot more detailed than the simple blocks Josh uses in laying out the level, but they’re not quite the final model either. It gives me an idea of the real shape and dimensions for what will be needed and usually is the basis of what the real model will be built on. A simple material shader is applied to get a rough color to match the palette, and a first pass on the level lighting is completed and checked into the branch for testing.
After additional feedback and adjustments to the lighting and the shapes, the real work begins. I start with the structural objects by finalizing their shapes, applying UVs and using a checkerboard texture to make sure the texel density is consistent.
For Unreal 4, we’re using Physical Based Rendering materials, which means that the textures are often broken down into four main elements: an Albedo, a flat color texture void of any shadow or details; a Roughness map that defines the microsurface of the material (how rough or smooth it is); a Normal Map to create depth; and a Metalness map for a more accurate reflectivity for metal. With so many components needed for these materials in GameCon, I introduced a Dye Map for many of the materials. The Dye Map has a mask on its Red, Blue, Green, and Alpha channels, allowing me to either target a certain section of the map to change its color, or all or none of it. This helps create a much greater variety of materials while utilizing the same textures and keeping the memory count low. From the very beginning, I knew GameCon was going to be a very expensive map memory-wise due to all the unique pieces required to recreate a game convention, so making sure the textures can be reused and that they’re tile-able means I can get the most out of them and keep GameCon manageable from a memory and performance standpoint.
Once the textures were created, they were assigned a Material Instance to keep plugged into. This helps cut down the amount of unique materials and greatly increases performance. Most of GameCon uses Material Instances from five master materials for the whole level.
Once the structural assets are textured, I’ll create another lighting pass and submit this for review. I try to break up the lighting passes at the end of every major stage of the level creation because, as more textures are added, it changes how the light behaves. I’ll then begin the next stage of Secondary objects, another lighting pass, and then the Tertiary objects and another lighting pass, all the while, submitting and receiving feedback. Finally, the level reaches a point where it is ready for an internal play day and can be judged while multiple players are in the map, running around and collecting points. Throughout the game play tests, I’ll receive feedback, adjust textures, and send minor mesh changes and lighting through again.
Probably the most fun part of creating GameCon was developing the various motion graphics and fake game posters and advertisements found throughout the level. During the production cycle, we sent out a form to ask the team for various game name suggestions. Most didn’t make the cut, but getting a chance to subtly poke fun at the industry and sub-culture we’re part of definitely cracked me up and lightened the mood of creating this very challenging level. Using a mix of After Effects and Premiere to create the various screen graphics, it gave me a chance to incorporate additional light and color through motion graphics and add life to an otherwise static level. These took the form of either an .mp4, flipbook texture or material shader, depending on the content required.
Lastly, I created the physics objects and scattered them around the level. These introduced more opportunities to inject humor into the map by creating the actual game boxes for our made up IP’s, soda cans, coffee cups, chairs, and laptops. This allows players to slowly but surely trash GameCon over the course of the three rounds, adding even more life and variety to the level.
I’d like to say that this would mark the end of development for GameCon but, as always, there are endless tweaks, changes, and adjustments to be made throughout the production cycle. We will revisit a map time and again to make further performance changes as we find them and to tweak the look to help maintain consistency. The end result can sometimes be dramatically different from where you started. Your perceptions of what you initially thought would look good sometimes don’t pan out, or a quirky idea implemented in a moment of inspiration can have dramatic effects that influence the rest of your artistic decisions. It’s these unexpected and rewarding experiences that keep me motivated throughout my career and was no doubt present for GameCon.