Sean Ian Runnels talked about the production of Days Gone, tackling large worlds and content creation as well as gave some game development advice.
Introduction
Hi everyone, I’m Sean Ian Runnels, a Senior Environment Artist at Sony. My primary focus is content creation for environments and hero props but I have had the pleasure of wearing multiple hats through my career. I have been in the game industry since 2012, however, my roots go back even further to my days in modding communities. In fact, modding helped kickstart my career since one of the mods I helped out with was picked up by the devs and added into the main game. Since starting my career I have had the pleasure of working on many titles with talented teams of all sizes. In recent years, I have worked at Ready at Dawn on their Lone Echo franchise and at Sony on their recent title Days Gone.
Days Gone Production
I joined the Days Gone team towards the end of production as a Senior Environment and Lead Prop Artist. Even though production was near the end we still had a lot to do in the nearly two remaining years. As the saying goes, 90% of production is done in the last 10% of development. My goal was to focus on hitting the most important elements first such as the cover assets because the players would spend most of their time behind a cover during combat and the camera would be closer to these assets. Other priorities were areas of the game that were still in blockmesh, the cinematic props, some interactibles, collectibles, etc. But I wasn’t alone, and a small team of artists all dedicated to different aspects of the game helped accomplish our individual and team goals.
For art direction, I pulled lots of inspiration from film, television, YouTube, and photography. Speaking of film and television, inspiration came from such titles as War of the Planet of the Apes, Dawn of the Planet of the Apes, 28 Days Later, The Road, The Walking Dead, and Life After People. On YouTube, I spent a lot of my time pulling ideas from bushcraft videos which are super useful if you’re looking for ideas on temporary shelters. I recommend checking out Joe Robinet’s channel. Finally, photography – there are different books and websites but I strongly recommend checking out work by Rebecca Lilith Bathory who specializes in photography of forgotten and abandoned places. The main direction and question for us during the production of Days Gone was what Oregon would look like after just 2 years of neglect. The answer for us was that some vegetation would start to creep in, lots of trash, and messed up roads.
Working on a Massive World
The sheer size and scope of the world were certainly a bit intimidating and we had to overcome many challenges such as performance and population of a world so large. To help us manage the workloads, the Art team was split into different sub-teams. To simplify the explanation, we have teams of people who create content such as props, environment art, VFX, characters, etc. and then we have the level art team who is responsible for building out the world using the content created by other teams while collaborating with the mission/open-world designers to achieve the overall goal for a POI (Point of Interest), Region, Mission, etc. Because we are such a small team, most people had lots of responsibility and ownership of major chunks of the game. For instance, we had one person dedicated to all the vegetation, one dedicated to textures, one weapon/vehicle artist, etc. Each person was essential and responsible for major and multiple art pillars shifted around during the course of the production. In addition to that, tech art helped the team by creating auto-population tools and identifying areas that were going to have major impacts on performance.
The last few months were spent focusing on optimization, worrying about streaming, LODs, the number of unique assets and texture sets, etc. First, we focused on global wins such as condensing the number of assets and swapping unique texture sets for tiling materials, reducing geometry for vegetation and the translucent card overdraw. But some areas had specific issues, for example, the militia walls were originally gnarly rough-cut planks that allowed for players to partially see inside the forts. It drastically influenced performance because we were paying for the cost of the content inside the fort while the player was outside its walls. That’s why I reworked the militia walls sealing up all the gaps which significantly boosted performance in the crater lake and diamond lake POIs. Efforts like these made by the entire team allowed us to tackle a larger world.
Content Creation
I always approach content creation with the idea of being flexible and iterative. I worked very closely with the Level Art team to determine what content they needed and how they would use it. When creating any type of modular kit I work on the grid creating straight pieces, corner pieces, ramps, stairs, etc. while always adhering to the metric system. I typically try to keep the kits small and manageable, with most of the kit simply leveraging tiling materials, vert blends, and weighted normals for faster iteration. A lesson I learned when working on other games is that having a massive modular kit can quickly become a burden especially if you have to make large art direction changes. Most of the time only the same small number of pieces would actually be used in a robust kit. So my suggestion would be to initially keep your modular kits to the essentials and then add to them when changes are required by the project.
Also, a valuable thing to know is that the grid is your friend until it’s not. What I mean by that is sometimes you’ll encounter unique scenarios where forcing pieces to work within your modular kit might lead to increased asset bloat, funky results, and time lost, so rather than forcing everything to work within your modular kit it can be very beneficial to instead have one-off pieces that allow you to quickly get through the problem area. If you’re looking for a deeper dive into learning modularity, definitely check out GDC talks by Bethesda – they are experts in modular kit creation and usage.
Advice
When working on huge worlds, my suggestion is to try to keep the number of unique assets to a manageable number and get variation through material swaps, color swaps, blends, and world-building instead. When I first started working on Days Gone and getting familiar with what content the team had, I quickly came across many instances where there were multiple variations of the same kind of asset. For example, there were about four different dumpsters all with their own unique meshes and texture sets, with only minor variations. There were also over six different barrels, again with unique meshes and textures. So one of the first things I began to do was consolidate these assets because we didn’t need six barrels that were all unique and yet looked similar. Instead, we needed at most two barrels with variety achieved through cheap color or material changes. Be efficient and smart about how you build your content to keep your library of content small and flexible.
Re-iterating on the idea of being flexible, I highly recommend using weighted normals and tiling materials with vert blends for as much content as you can. This methodology was ingrained in my workflow while working at Ready at Dawn. They have presentations which about the work on The Order: 1886 which I recommend checking out – there, they go in-depth over their workflow. Basically, this workflow offers the highest flexibility because you can make changes on the macro or micro levels very quickly. For example, you have a furniture set that is baked high-to-low with unique texture sets. In case you have to change the length of the tables, at best, you make a change and get minor texture stretching, at worst – you’ll have to re-layout UVs and rebake, plus your texel density is probably not that great or consistent anymore. But if you use the RAD workflow, you eliminate all these problems and can iterate way faster as you can drag some verts, relax your UVs, maybe re-apply the weighted normals and you’re done. Then, to take it a step further you can make hardware fixtures such as drawer handles or hinges – those assets can be uniquely baked onto shared texture sets. Take those assets and make prefabs for your furniture sets out of them. What you end up with is a furniture set that looks highly detailed and is very flexible because if you decide to change the wood color or the style of the handles it all applies to all the furniture. Game development is a highly iterative process and the more flexible your content is the easier it will be to make those macro and micro changes.
One of the last bits of advice which can be overlooked is just how important it is to embrace blockmesh for as long as you can. The longer you can remain in the blockmesh phase the more flexible you are when complications appear or the direction is needed to change. On many occasions, I have seen requests for content that were taken to the final art stage before prototyping. This a huge mistake which can lead to wasted time and money because almost nothing works on the first go. Before taking an asset to the final stage, make sure to provide people with a blockmesh so that they can get check out the content. This way issues can be spotted or even cut entirely before any more of the time investment is made.
Game development is constantly changing and every cycle is different from the last one, which just part of the fun for me. If you have any questions left, please feel free to contact me on ArtStation – I’m happy to elaborate!
Sean Ian Runnels, a Senior Environment Artist at Sony Interactive Entertainment
Interview conducted by Kirill Tokarev