Developed by Zoink and published by Electronic Arts, Lost in Random is a gothic fairytale-inspired action adventure where every citizen’s fate is determined by the roll of a dice. Released in September 2021 to rave reviews across multiple platforms, the game features a unique stop-motion aesthetic, inspired character design, and explosive dice gameplay.
We recently featured Lost in Random and the Zoink team as part of our Creator Spotlight series on Twitch. In case you missed it, you can watch the full stream recording below.
The Kingdom of Random is a shadowy fantasy world where danger and excitement lurk around every corner. We sat down with Zoink’s Environment Artist Leo Brynielsson, Creative Director Olov Redmalm, Art Director Victor Becker, and 3D Artist Juno Johan Holm to take a deep dive into the development process and stunning art design that makes up Lost in Random.
Redmalm: Zoink [now part of Thunderful Development] has always been an art- and innovation-driven studio, starting new projects with a strong visual and conceptual hook – fueled by our passion for exploration. When I first joined Zoink in 2016, we had just signed our first EA Originals project, Fe, which we worked on alongside Flipping Death. These games feature completely different styles, atmospheres, and tones, [but share] a longing for exploration in game design and for telling great stories in gorgeous, handcrafted-looking worlds.
After these two games, we went on to make Ghost Giant, a VR experience that really tested our ability to deliver a great-looking game – [in] a fairly realistic cardboard, dollhouse world – without losing frame rate.
Our next adventure, Lost in Random, brought with it a larger budget, as well as loosened technical restrictions. This, in turn, opened [up] a variety of artistic paths we could take, which became somewhat overwhelming, so we began by setting some sticks in the sand:
After the cute and colorful Ghost Giant, we wanted to make a dark and epic fairytale, [with] the setting, style, and gameplay revolving around giant tabletop games, including dice, cards, and playing pieces. We wanted to combine the mystery of Fe with the funny and smart dialogue writing of Flipping Death [by Ryan North], plus the deeply emotional storytelling of Ghost Giant.
Rather than create a hyperrealistic game, we [decided] to use this new freedom to recreate a realistic claymation world – something that I think Klaus Lyngeled (our CEO and creative head) had been wanting to do for a long time but hadn’t been able to until now. As to how we went about making this vision a beautiful reality [that] runs smoothly on PC and all other platforms, our Environment Artist, Leo Brynielsson, is here to enlighten us.
Brynielsson: When we started working on the art for Lost in Random, we knew we wanted to go out of our comfort zone and push our limits. We wanted to create a world with high-quality graphics without losing the Zoink vibe from our previous games. To study and recreate realistic materials and lighting was a new challenge for us as a studio.
The world of Random is crafted by materials like clay, wood, and cardboard. To really sell this claymation look, our materials needed to appear believable. We even made stuff in real clay, just to see how it behaved and what it looked like up close, so that we could recreate it in Unity.
Another part of our art style is the exaggerated shape language – I don’t think you can find a single straight line anywhere in the world of Random. Buildings bend in weird ways, some are huge while others are tiny. Art Director Victor Becker and I would always stress the importance of contrasting shapes and materials, as well as playing around with crazy color palettes to create interesting contrasts that push the stylized look further.
Brynielsson: We knew from the start that it would be challenging to create this detailed universe and still have it run smoothly. On the environment side, thinking about performance early on is very important and something we always have in mind when coming up with the workflow for creating assets and materials.
We decided to go for a kind of modular, kit-based approach, where each kit contained a set of assets that shared the same material. For instance, we created a kit with clay assets that included a few blocks and pieces assembled in 3D software to create houses, bridges, and other assets.
This process meant that we could create an unlimited number of assets just by combining basic pieces in different ways. It also lent itself really well to our art style, since we wanted our world to feel handcrafted, almost as if it was patched together.
This workflow not only looked great, it improved performance as there were fewer materials being used than if we’d created many uniquely sculpted assets. This, in turn, led to less texture memory usage and fewer draw calls onscreen because most of the larger assets could be merged together and still share the same material.
One downside to this technique is that the polycount needs to be quite high when patching together bigger objects with smaller pieces. We solved for this by creating alternative, heavily-optimized assets for the objects that were further away from the play area.
In the early stages of development, we used the High Definition Render Pipeline (HDRP) to really push the visuals and leverage the realistic lighting, volumetric fog, and so on. But we realized that this approach would not work in the end, since we needed the game to run smoothly on all platforms – and HDRP focuses more on high-end PCs and consoles.
We [ultimately] decided to switch to the Universal Render Pipeline (URP). This was a hard yet necessary decision to make, and I’m actually glad we started out with HDRP because we used those visuals as a benchmark for what we wanted the game to look like in URP. So we put a lot of work into getting the most out of URP, figuring out our own solutions and using the assets available to us, like implementing volumetric fog, decals, planar reflections, and so on.
For an environment artist, it’s always a battle between art and performance. During the production of Lost in Random, we faced many challenges in this area, and had to rethink and rework many parts of the game to keep it looking as great as possible, while keeping performance on top.
It can be a bit frustrating to be limited by performance, but it’s also very satisfying when you succeed in creating something beautiful that runs perfectly on any platform.
Brynielsson: One great challenge while creating a game set in a big city [with] a dark atmosphere is the lighting. We were dependent on using many spotlights to lead the player and light up the darker areas of the game.
Artistically, it’s fun to color the world with lanterns and torches, but again, it [must] balance with performance, [so we] tried to keep the number of lights to a minimum. At one point, we experimented with a method called lightbaking. With this method, you bake the lights to a texture and then apply it to a separate UV texture on the objects in your level. The downside to this method is that it requires a lot of memory and can’t bake the specular highlights of the objects. This was a problem for us since our texture style depends on the roughness and specular highlights of clay, metal, and slimy creatures, [among other factors]. Another issue with doing this for very large levels is that it takes time to bake the lighting, and the memory impact is big. So we decided to go for a fully dynamic approach instead. We had a number of solutions to correct the performance issues that came with it:
1. We implemented our own light culling system. Each light has two parameters – one for the distance of fading in the light, and one for when the light is at full intensity. With this system, we could manually control which lights to enable, and drastically decrease the number of lights active at the same time.
2. We turned off all shadows on the environment assets by default and activated only the most necessary ones. Whenever possible, we used shadow proxies – a simplified mesh to cast the shadow – instead of having an actual object cast the shadow. In this case, we used simple cubes.
3. On the less powerful platforms, we turned off the shadows on all lights except the sunlight. To compensate for this, we implemented a classic blob shadow underneath the main character.
4. Many of the shadows on the characters were also turned off. This was possible because many of them stand in the shadows [or] dark parts of the levels, and some of them are only used in action-packed sequences. There’s so much else to focus on – some of the enemies, for example. It’s barely noticeable in the bigger scheme of things.
Redmalm: The big bang of Lost in Random was when we found the painting of a girl and her dice companion walking through a gigantic board game. From that place stemmed the culture, history, and belief systems of Random, as well as the gameplay. It was such an inspiring theme to us, and all of the game’s elements grew and intertwined naturally from there. Passing the word on to our Art Director, Victor Becker here! What do you say?
Becker: Well, you’ve simplified it quite a bit, so let me elaborate! Through extensive painting and bouncing many ideas off each other, we located some key objects that could work as reference points for the world of Random. The houses you can see in Two-Town [for instance] paved the path for future workflows and creations.
As Leo [mentioned], we also spent plenty of time establishing our foundational materials, or our kits as we call them. These consisted of metal, wood, clay, and such. We would apply them to different shapes or objects. This really helped us maintain the art style. Key words were another tool we used: Some of these were “flow,” “contrast,” and “wonky.”
Lastly, we would usually paint objects and characters into their respective locations. When they were being created in 3D, we would always test them early on, inside the engine, which [clarified] the improvements that could be made in terms of shape, material, and scale.
Becker: We aimed for our characters to be very expressive, creepy, and cute at the same time. We wanted to capture the sensation of something being “off” with every single one of them! To achieve this, we were on the lookout for a hand-sculpted vibe and painterly look, which makes the characters come to life in a very different way. We drew inspiration from the work of Laika Studios and Borislav, who would later join our team. Western cartoons and Japanese anime were also inspiring in terms of achieving strong expressions.
My favorite character is the Nanny. She’s pretty much the embodiment of Random [who] has seen it all. She’s always caring for the Queen, through thick and thin, [and] remains a brutal and loyal servant. The things we do for love!
Redmalm: Funny that you mention her indomitable Majesty, the Queen, because she’s my favorite character! She’s definitely the most imposing character in the game, who took the most design work, with a dress made of dark crystal, broad menacing shoulders, and a face like an owl’s. I’m not alone in my opinion here – Kotaku even set her against Lady Dimitrescu in an article.
Brynielsson: Keeping the Zoink vibe was important to us. I worked on games like Ghost Giant and Flipping Death, and those games were strong influences on my work for Lost In Random. Ghost Giant, in particular, has this highly crafted, patchwork look that is really Zoink-esque – that feeling of a crafted world is something that we really wanted to embrace in Lost In Random as well. Simply put, we wanted Zoink’s foundation to be there, but [with] some extra spice added to it, with more realistic lighting and materials.
Redmalm: I just love working in a studio where we don’t immediately turn to hyperrealism just because we can. We always ask ourselves “but what’s new about this? What’s going to make someone go ‘oh, it’s that dice game’?” We’re always pushing ourselves to think outside the box, to see if we can flip the pancake just one more time before calling it ready to munch. Even though our studio has joined Thunderful Development (together with Image & Form and several other studios), we still work to hold on to our Zoink philosophy.
Becker: I totally agree with both Leo and Olov, and firmly believe that these are the reasons that brought us to this company. I wanted to inject a fair dose of Zoink into the art of Random, so I would spend quite a bit of time mimicking the style of Olov and Klaus to understand why Zoink is Zoink. I mean, this idea is as Zoink as it can be, so I zoinked it up as much as Zoink let me Zoink!
Brynielsson: Try to keep the polycount as low as possible while still maintaining the desired visual result. Do this for every asset before calling it finished. It’s lots of work if you do this later on, when hundreds of assets are already in the game.
Don’t use the same assets for backgrounds and play areas. Background assets are assets that you will always see further away, so they will never need the same level of detail. Use very simple assets for this, sometimes billboards work perfectly. We painted in a lot of details in our backdrops to make sure the background blended in nicely with the actual 3D assets.
Next, turn off any unnecessary shadows. Usually there will be lots of meshes in the shaded parts of a level. Those meshes don’t need to have a shadow caster since there will be no visual difference if it’s on or off. Take the time and turn them off.
If you have really complex meshes that need shadows, it might be worth creating a separate, simplified version of the mesh for the shadows. Especially if it will be used many times.
Try to be as modular as possible when building your levels and preparing your assets. Focus on assets that can be used in different ways and duplicated many times. This way, the engine can instantiate many assets in the same draw call. What this means, for example, is that, if you have four duplicate meshes in the screen view, these can be batched together as one mesh during runtime, making the overall rendering faster.
Holm: We realized we could lower the texture size quite a bit without losing fidelity. We let the normal map stay at 100% size to get crisp reflections, but then we lowered the base color to about 50%, and put roughness and occlusion at about 25%.
So if an object had a normal map that was 2048x2048, the other maps could be lowered to 1024, 512, or 256. We lowered them until we started to notice any loss in detail.
If something doesn’t have a lot of detail in the base color map, you can probably lower it quite a bit before it [starts to] look off. In some cases, you can put the textures on even lower resolutions.
Brynielsson: It depends on the asset. For generic environment assets, we usually block out a small part of a level with temporary assets first, to try and find the shape language we’re looking for. Same [goes] for figuring out what kind of assets we need for a specific level.
Then we paint some basic concepts from these shapes and start working on the final sculpts. It’s a very iterative process, so we usually go back and forth many times until we’re done. It’s important to see the assets in the context [that] they’ll be used, so that they’ll work with the rest. I also usually ask myself some questions before starting [to] work on an asset: How important is this asset? Can we create this from a material we already have? Can we make use of this asset in ways other than its original purpose?
It’s about being smart with what you have. Very often, it’s not necessary to create simple props from scratch if you already have a foundation of assets, or kits, that can be used in different ways. For instance, we use ZBrush for sculpting and Substance Painter for texturing.
Brynielsson: Don’t get stuck on individual details. Focus on the big picture and make sure that everything works together. Some areas might benefit from having fewer details, while others need more to give the player something to focus on. Make sure to test out the visuals in the context that they’ll be used in. It’ll always be an iterative process; don’t be afraid to scrap something if it doesn’t work.
Additionally, don’t forget to put effort in the background of the levels. They play a big role in the overall look, and can also work as beacons for leading the player.
We have plenty of fog in Lost in Random, and this is something we leveraged to our advantage to make the silhouettes pop, and create a more interesting color palette.
To give the fog some extra life, we made a shader that we call the Fog Plane, that’s basically a mesh plane with a faked fog. It looks like 3D fog because it takes the depth of the level into account, and it fades at the edges of meshes. When you get close to the plane, it will slowly fade away. While it can be used in many different ways, we mostly used it to give the levels some extra color and make the gateways stand out.
Brynielsson: It’s important for me that the levels are beautiful, but it’s just as important that the play area is interesting to explore [and] that players know where to go. I always try to think of this aspect when building level art. This includes everything from bending the shapes and silhouettes to lead the eye, making gateways pop with extra fog, and adding lights to specific locations. I also try to be smart [about] where I add in hero props and elements that stand out from the rest.
Working with colors is a great way to differentiate areas; one area might be green, while another is red. This is especially important for larger, open levels, where you have to go back and forth between different areas. There are many tricks you can use, but it’s hard to know what works until you’ve had it tested on other players.
Redmalm: Just like with any intertwining disciplines, work closely with the artists to see what art tech is available to help clarify your vision, and see if it’s possible for them to create the thing you’re missing. Run through your levels and try to imagine how much of each activity you’ll be doing, and for how long. Walk through the streets of grey-box Two-Town, pretend to talk to characters that aren’t there yet, and fight the battles that might be there at the end.
Getting things down on paper is great – but the sooner you can try it out, or at least roleplay your way through an imaginary final level, the easier it will be for you to visualize that level.
Redmalm: Dare to take risks with both art and game design. Bring an open beginner’s mind to the blank canvas. Players are thirsty for new stories, new characters to tell them, as well as new environments to tell them in. Double-check your inspirations and intentions – why are you drawing this character in this particular way, and how can you make it more interesting? What’s out there in terms of style and storytelling, and how can you bring something new to it, while still building on what others have already created?
When you bring a weird idea to the drawing board and people raise their eyebrows, don’t immediately surrender that idea. Keep pushing, keep trying, keep pitching. If we didn’t, there wouldn’t be any living dice (with arms, legs, and one eye) in this world!
We hope you enjoyed this deep dive into Lost in Random’s creative process. Go check out the game today.
Now, we leave you with this Halloween Musical, and look forward to seeing you again in random places... Odds are we will.
If you’d like to see more artwork from the creators behind the game, visit their Instagram pages.