Techland's Katarzyna Tarnacka-Polito and Maciej Jamrozik told us about the art direction and the technical side of Dying Light 2: Stay Human and shared how the team stayed motivated when working on this immense-scale project.
The long-anticipated Dying Light 2 Stay Human finally saw the light of day on February 4. Before the sequel was released, we asked Katarzyna Tarnacka-Polito, Environment Art Director on the project, and Maciej Jamrozik, Technical Art Director, about the game's art direction, parkouring mechanics, and the technical side of the game.
80.lv: How many artists are currently working on the Dying Light 2 project? How is your team organized? How do different departments work with each other? Could you share some details on your team dynamics?
Maciej Jamrozik: It is hard to count how many artists are currently working on the project. At Techland, we have over 30 artists who worked directly on Dying Light 2 Stay Human. But we also cooperated with a lot of freelancers.
Katarzyna Tarnacka-Polito: Our teams are divided by domain: Concept Art, 3D Enviro, Level Design, Level Art, Tech Art. However, when working on specific locations or even groups of assets, members of individual domain teams form temporary interdisciplinary teams.
These temporary teams can operate for many months, during which the composition of the team itself may change as needed. The assignment to one or another team is usually determined not only by the individual predispositions/skills of a given person but also by his/her workload. By dividing our artists into teams, we use their strengths, talents, interests, and knowledge, e.g. architecture.
80.lv: Let’s talk about the art direction. How did the world of the first part help or interfere with creating the one in a sequel? What were your objectives and what has changed since you were working on the original game?
Katarzyna Tarnacka-Polito: When it comes to art direction we knew we needed to expand upon Dying Light 1 quite a bit. When developing the first game we already knew what potential there is, if we’d push it a bit further. We had a lot of things to figure out in order to depict a city that has been lost, not for 3 months ago (line in the previous game) but about 20 years prior.
One of our objectives was to develop what we’ve done in the first game, to go further. It was actually an adventure to explore that subject and see for ourselves what could happen to a world, to an urban environment if left for so many years in the world of Dying Light. Art direction is built on top of several pillars, like modern dark ages, broken world, hope, and contrasts. To me, the most interesting is the last one, because it derives directly from the title.
"Dying Light", aside from all the other meanings, for the game means basically the moment when the day turns into the night. So, the contrast between day and night (light and darkness) was crucial. But we took it a bit further and built on top of that. The day is for the people, so it represents life – something thriving, something new. The night is for the infected and those represent death, something lost and old.
We tried to find what it could mean for a city. Since it’s a game with parkour the player, as the protagonist Aiden, traverses mostly on the rooftops, while the infected roam mostly on the streets. So, roofs are for the people (life) and streets are for the infected (dead). That’s how we reached probably the most iconic visual aspect of Dying Light 2 Stay Human – lush, full of vegetation rooftops and dead, empty streets, that’s been lost for humanity.
We’ve put a lot of effort into pushing the visual division between those two worlds and used quite a few tricks to achieve that (including changing a color grading for the street level). It was a challenging design-vise and also technically – it’s not exactly easy to render both an urban environment and basically a forest on top of each other.
When it comes to the first game influencing the world design of the sequel we always kept in mind the differences between Slums and Old town. At some point, we joked around that the new survivor structures, built on the rooftops of city buildings, resemble Slums that were built on top of Old Town. It’s basically just trivia, but it’s been in our heads along the way.
Maciej Jamrozik: Also, it is worth mentioning that we didn’t want to make the biggest open-world video game on the market. We wanted to fill it with meaningful content and valuable activities for the player. One element that strongly evolved during the production was the size and nature of the map. Individual regions have evolved, merged into larger ones, adjusted to support parkour activities.
80.lv: Could you tell us how you used the procedural technologies? Did you create any tools that could help to create the buildings quickly or modify the materials? It would be great if you could share some of these tools.
Maciej Jamrozik: While targeting the original game size, we knew the map would be iterated multiple times, especially at the beginning, that's why we created a tool to support this process: City Builder (CB). The main task of this tool is to combine pre-prepared prefabs into complete buildings and arrange them on the map, based on a city plan prepared by LD/LA teams.
The city is divided by streets into quarters (polygonal plots), inside which the CB automatically puts up tight lines of buildings, decorations are automatically placed along the streets, or green areas are filled with vegetation biomes.
Apart from generators, CB also includes a mechanism that allows you to configure the appearance of the city at the level of individual buildings, without the need to manually set their individual elements. On the basis of CB, tools supporting the analysis and optimization of the map were also developed, measuring, e.g quantity and distribution of assets, or number/size of textures used.
After the overall distribution of the city stabilized, as production progressed, we gradually froze the generation in individual quarters, regions, and gradually shifted from automatic generation to manual finishing of individual locations. At some level of detail, simply a manual approach is just more efficient than extending the ruleset of the generator further.
Katarzyna Tarnacka-Polito: Procedural generation was a great tool when trying to find the base geometry, the gameplay rhythms, and general shape. Manual polishing and decoration were required in order to provide believability, smaller-scale composition, narration, environmental storytelling, and details that encourage exploration.
80.lv: One of your tasks was to create a city that would look seamless. Please, tell us about the main principles of urban construction. How did you build the areas that would look and feel differently?
Maciej Jamrozik: Seamless means no loading screens. If the city is too big to load, you have to divide it into smaller fragments and load them in full detail only around the player. As I’ve mentioned before, we divided our city into quarters and regions, thanks to that we were able to render efficiently at different levels of detail.
However, the division itself is not enough, because the game is also released on the previous generation consoles, which may mean streaming problems, e.g. delays in loading textures. One of the things that relieved the streaming a bit was the division of the city's regions into smaller districts with similar characteristics. In the first place, it helped increase the differences between the various parts of the city, it also reduced the number of assets needed at the same time.
80.lv: One of the base mechanics of the game is getting around the city by parkouring. We believe it is quite hard to create such a world where people can both get around quickly and distinguish the environment as well. How did these mechanics influence the art and how did you make sure that all of the environments were readable?
Katarzyna Tarnacka-Polito: It’s incredibly demanding and challenging to design an actual urban environment under the regime of measurements required by parkour. To put it simply: ledges need to be placed not only in given distances (in order to provide a way to the top) but also in a particular rhythm, which makes the traversal interesting and challenging.
It would be easy if we’d be designing something abstract that doesn’t exist in real life and can't be referenced based on actual experiences. A sci-fi setting, for example, would be less challenging. But in a regular city, buildings are far more difficult. The player knows exactly what a building looks like – failing to fulfill the expectations would result in a game that’s not immersive. The challenge is in finding the line between gameplay requirements and a sense of familiarity.
Parkour is not only a game-changer gameplay-vise but it also completely changes the development pipeline. A task to design a building, for example, means something much different in this title than it does in games without parkour. The artistic qualities are one thing, but the second part is to design them around a very particular gray box built by level designers. Sometimes, at the first glance, the prototype looks nothing like a building and it takes some thinking to find a design that works both for visuals and gameplay.
Thankfully our artists and designers are incredibly talented and the key to success is close cooperation between all the departments – concept art, level design, and environment art. The gameplay intentions need to be fully understood by all people involved, everybody in our development team plays the game a lot and understands the gameplay principles, including concept artists.
Besides that open world, FPP games with parkour require us to take into consideration that everything can be seen from basically every angle and perspective. All the structures are seen from afar, as well as from up close (basically as close as it gets). Players can see the back, the top, and often the interior, of every structure. There’s virtually no cheating, no smoke and mirrors. Everything needs to be accurate. Working within those constraints is really stimulating and actually a lot of fun, especially when you see the results.
Maciej Jamrozik: Also, the aforementioned dividing of the city into different districts was also intended to facilitate orientation in the city. After some time, the player learns that meadows are found in this part of the city and birch trees in another.
An additional element related to the parkour itself was a strict color code, especially the use of yellow to mark parkour elements – boards, beams, etc.
Of course, parkour itself had a huge impact on the city itself, forcing appropriate distances between objects, their heights or slopes, and the shape of the roofs. For instance, in the beginning, the roofs in the old part of the city were more soaring and pointed which was consistent with the references, however, they made it difficult to move around, so we changed them to a flatter one, allowing Level Designers to better control the parkour paths.
There were also a number of street-level elements, such as ramps, winches, and springboards, to reduce the time it took to get from street level to the roof. In the modern part of the city, characterized by much greater height differences, there are also vertical air drafts which, combined with a paraglider, allow the player to move efficiently vertically.
World Changes in the Game
80.lv: As far as we know, the player’s choices have much impact on how the world changes in the game. How did you approach this aspect and how did you create different scenarios? What were the main challenges?
Katarzyna Tarnacka-Polito: The main challenge for the development team was to make sure everybody fully and completely understands the game systems and dependencies. The topic is viewed in a very different way by game designers than the art department, for example. Providing a bridge between those two ways of thinking was the key.
Throughout the production, we used several methods to provide that understanding, especially between the departments. We often "played the game on the whiteboard", which was basically a room of people describing a system and one artist drawing a very simple comic on a whiteboard. People would ask: "Ok, so what happens if I do this?", and we’d draw the world changes on the board. Later, those comics would be cleaned up and shared with the entire team as a reminder or used to introduce the systems to new people.
The windmill settlement system is a good example – it’s connected to several other variables in the world. An abandoned windmill needs to be fixed by a player in order to provide electricity for people to settle around it. But it’s connected to the facilities system that controls the zone alignment. So, depending on which faction you decide to give control over the facility the zone aligns to Peacekeepers or Survivors, and all possible windmill settlements in a given area assigned to chosen faction.
But what if you fix the windmill before you assign the facility? Now that we see it in the game, it’s quite simple, but those were the questions that needed to be asked in the early stages but most importantly seen by people, drawn on a whiteboard before we all fully and completely understood the systems. Without a deep understanding, the artists couldn’t design it, but also couldn’t make good technical decisions on how to approach the topic, how to structure the hierarchy, how to divide it into modular systems, etc.
We actually made a simple point n’ click game at one point that simulated one of the systems in order to understand it well enough to plan the work wisely.
Maciej Jamrozik: Worth mentioning is that Dying Light 2 Stay Human is developed for different console gens. This imposed on us a lot of limitations resulting from the possibilities of the previous generation of consoles, and for us, it is extremely important that players have the same experience regardless of the platform.
The game revives life on rooftops and death on the ground. Thanks to this, the foundations of the map and buildings could remain mostly unchanged, and we focused on adding new elements to their roofs. Another aspect was the visibility of these changes so that the player could see that something has actually changed as a result of his decision.
In fact, the change of some elements on the roof on the scale of the city is difficult to see. It was a kind of "artistic challenge": how to make such constructions so that they were visible from a distance. Fortunately, DL2 is not a simulator;) so we could exaggerate a bit with their scale so that important objects protrude above the tree line.
More importantly, each of such locations was designed to have its own unique, distinctive shape, recognizable from a distance.
Work on Assets
80.lv: There are lots of weapons and equipment variations in the game. Could you tell us how did the work on these assets go? How did you manage to create such a huge variety? How hard was it?
Katarzyna Tarnacka-Polito: We kicked off the process of designing the weapons by asking the entire art team to work on sketches of weapons. That way we managed to generate a lot of very different ideas very quickly, which helped us find the direction we want to go. Those early ideas were the inspiration later on for a dedicated and crazy talented Weapons Team that took care of it for the rest of the production. We knew very early on what we expected from weapons in the Dying Light title. The Electric Machete is in Techland’s DNA.
Technical Side of the Game
80.lv: Let's discuss the technical side of the game. How did you work with skyboxes? What was your approach to lighting and weather conditions? How did you use RTX? How difficult was it to optimize the world both making it as beautiful as possible and allowing it to work on different platforms? Taking into consideration the amount of work and the complexity, what were the main bottlenecks?
Katarzyna Tarnacka-Polito: In the Dying Light title the sky can’t be just a background. In a way it plays a gameplay role – it contains and conveys important information to the player. When the night comes, the gameplay changes so we wanted to make sure the sky is easy to read when it comes to the time of the day. So, the key for us was to make sure the player will be able to learn and tell the difference between the first and second half of the day.
We made a baseline decision for the sunset sky to differ from the sunrise as much as possible – in colors and mood, besides obvious sun position difference. The sunset sky was the most important because, in a way, it’s in the title. We wanted to make it as iconic as it was in the first game. But the dusk sky was the one we wanted to look dangerous. The intention was for the players to be able to feel that the last rays of the sun are disappearing behind the horizon and the night is very close. Following this line of thinking the dawns and sunrise were meant to be as peaceful as possible. Dawn represents hope and sunrise represents safety.
Lighting and Weather Conditions
Maciej Jamrozik: We focused on selecting and refining key times of the day/weather conditions. In DL2 Stay Human the time of day is crucial for the game, we wanted the player to have as clear a message as possible, e.g. underlined golden hour, very different from other times of the day. And just before the end of the night, the sky is brightened to let the player know that it will be easier in a moment
Maciej Jamrozik: Ray Tracing is a technology that requires powerful equipment, and there is a gaping gulf between the latest graphics cards and old-generation consoles – these are different worlds. And as an efficient real-time implementation of RT is not an easy task in itself, hitting a game like DL2SH on PS4 or Xbox One S is a challenge. Therefore, most of our efforts had to be focused on optimizing the game on all platforms. Our small but efficient team was delegated to work on RT, which managed to implement the most important effects, such as soft sun shadows, ambient occlusion, indirect diffuse lighting, reflections, or indirect flashlight lighting.
Optimizing the World
Maciej Jamrozik: Optimizing the world to both make it as beautiful as possible and allow it to work on different platforms was challenging, indeed. As mentioned before, we wanted the game to provide the same experience for players, regardless of the platform, especially when it comes to smooth gameplay.
Nevertheless, our goal was to make the content on all platforms similar, so instead of drastic cuts in texture resolution or model accuracy, we focused rather on limiting technical parameters such as screen resolution, ranges, or the accuracy of individual lighting components.
We took a bottom-up approach: first, we optimized the game for the weakest platforms, and then improved individual effects on more powerful devices. This approach was also supported by the fact that the more the game is optimized on basic consoles, the more power, the better the effects we would get on more powerful devices.
80.lv: You reported that you guarantee the support of the game for years after its launch and the game has already been in development for quite a long time. How hard is it to keep being motivated when you’re working on a project for so long?
Maciej Jamrozik: at the last stage of video game development when all the elements start to click into place and a fuller picture of the whole game begins to emerge this gives us "the wings". We are also accompanied by stress and excitement about how the game will be received by our fans.
Katarzyna Tarnacka-Polito: There are different stages of the production when it comes to motivation, the most difficult is just before it all comes together. Then, when the team is able to play the game without major difficulties, everything works as intended, we can actually see how the game feels and everybody gets a motivational boost.
It’s not really difficult to keep the motivation post-launch. I’d say there are people on the team that thrive in DLC’s and post-launch support. It’s a different type of work, not so many problems to solve and some of our best work memories were made while working on DLCs. And most importantly by then, we get to see how people play the game, read the reviews and comments made by players, watch gameplay and that gives us a different type of understanding of what the game actually is when played by millions of people. This gives us the opportunity to provide more of what people actually want.
Teamwork and Working Model
80.lv: What is your approach to teamwork? And what is your current working model in the post-COVID world where lots of companies switch to remote or hybrid models?
Katarzyna Tarnacka-Polito: Our approach to teamwork is communication and deep understanding. Every piece, every asset that’s created during the production is a part of something bigger, is used by somebody else in a different specialization. A concept art piece is a tool for environment artists, a 3D model is a tool for level artists and designers, etc. It’s crucial for everybody to understand how their work, the tool they create, is being used by all the other developers, and to understand what it helps to achieve.
Maciej Jamrozik: We are working 90% from home, so rather a hybrid model of working. The pandemic has shown that remote/hybrid work is possible in our industry, we try to use the current situation to our advantage, e.g. we are looking for specialists abroad, without the need to relocate them. Of course, there is no contact –but we are still working on it.
Katarzyna Tarnacka-Polito: We adjusted surprisingly fast but the strangest part was working years on a game about pandemics and suddenly finding ourselves in the middle of an actual, real-life pandemic. I still can’t quite wrap my head around it.