Breakdown: Come On Studio on Creating the Cozy Sandbox Game Hozy

What inspired the core concept behind Hozy, and how did the idea of restoring a forgotten neighborhood take shape early in development?

What if we made a furniture arrangement simulator, but not like all the others —one with a realistic, tactile-feeling process?  What if we also added cleaning mechanics like in PowerWash Simulator – not just washing, but repairs too? And everything should feel, react, and sound like it does in real life.

Imagine starting the game in an abandoned, dirty attic, and finishing in a beautiful space you quite literally created with your own hands. And there won't be any timers!

Since it won't be just one or two levels, they need to be tied together by something and someone. So we went through ideas ranging from a flooded city to time travel, and eventually settled on restoring a neighborhood, since that fits Hozy's spirit best.

The game focuses on cleaning, decorating, and restoring spaces at a very granular level. How did you approach designing these mechanics to feel satisfying moment-to-moment?

For many years, we designed UI and interfaces, in which the key element was the pleasure of using them. Hozy is an interface within a game environment, where we set out to create a constant sense of enjoyment from every action in the game.

For example, the mop doesn't just move back and forth 0 it turns in the direction of movement, tilts depending on speed, leaves behind a wet trail, squishes, and makes other sounds. We hope you can believe that's a real mop.

Many cozy games lean into low-stakes gameplay, but Hozy seems to remove timers, scores, and failure states entirely. What guided that decision, and how did it impact overall design?

We just wanted to create a game that does not require a correct answer from you. Internally, we called it a curated sandbox experience. People can enter the flow and do whatever they want – even throw away most of the objects from the level, if they want to. We thought about adding some 'beauty points', but removed that concept early on.

Can you walk us through your environment art pipeline, from initial concept to final in-engine scene?

Before our concept design workflow solidified, we ran several rounds of R&D because there were so many tricky, non-obvious questions we needed to answer. Which layouts work best with our camera angle? Can we do multi-level spaces? Can we use stairs, and how well will they mesh with our cleaning mechanics? How do we show walls between rooms? Where is the line between a satisfyingly dirty space and a downright disgusting mess?

One major roadblock was figuring out how to handle the outer edges of the room, where the rest of the house would logically continue. We experimented with making them monochrome, flat, high-vis safety colors, or jagged cutouts, and we even tried a massive iteration using stylized paint strokes. Ultimately, we realized the boundaries needed to be dynamic – slightly rugged, with small bits of floating building materials. A chipped brick here, a stray roof tile there, a pipe sticking out. These little details instantly made the spaces feel alive.

Naturally, the design process for any location starts with a core concept or theme. For our very first apartment, we wanted something universally cozy. We kept picturing a typical Brooklyn apartment – a historic brick building with a quirky layout, massive windows, and warm sunset light washing over everything.

Once the overall vibe is locked in, we put together a moodboard. Two brief tangents here: First, thank God for AdBlock blocking ads on Pinterest – otherwise, it's hard to get any actual work done! Second, right around the time we started heavily scouting Pinterest for Hozy references, the platform got absolutely flooded with AI-generated content. It felt like over half the interior design images were AI-generated, which honestly breaks a designer's heart. But, of course, we eventually adapted.

Once a location is finalized as a mood board, we move to the concept phase. In our case, we do this straight in 3D. It gives us a much better feel for the scale of the room, the lighting, the materials, and exactly how many objects we need to scatter around to create that signature cozy feel.

It's also worth mentioning that early in development, we had this grand ambition to tell environmental stories through the messy states of the levels. You can definitely see this in our first few maps – scribbled-on walls, random gnomes, and themed trash. However, we soon realized this approach often bummed players out. They hated having to throw away items they actually wanted to keep, or being forced to paint over something that looked genuinely cool. Because of that, we pivoted to a much more straightforward, streamlined approach.

We also spent some time figuring out what the backgrounds should look like. We eventually landed on two key principles: first, the background shouldn't distract from the level itself, and second, the levels need to be self-contained enough to work flawlessly against any backdrop.

We really wanted to implement a detailed environment visible through the windows, but it turned out to be quite a headache in Unreal Engine. Ultimately, we chose not to burn our team's limited, valuable resources on it. Next, we take the concept and test it in the engine, checking whether the space ended up too small or the ceilings too high.

At the same time, we break the concept down into individual objects in Miro. There, we also expand the narrative, add references, and track the production status of each object. In the end, we get a catalog of finished objects with screenshots and names, sorted by category:

What tools and software are central to your workflow for modeling, texturing, and assembling environments?

At the beginning, there was chaos: we were inventing the visual style and constraints on the fly, and most objects were made with unique textures. Later, we built a set of basic tileable textures: wood, metal, plastic, and so on. Right now, we use two methods for creating props:

• High-poly model → low-poly model → baking in Marmoset Toolbag and texturing in Substance 3D Painter.
• Mid-poly model with tileable textures applied in the engine.

These two methods are often combined within a single prop. Statues were sculpted in ZBrush. With rare exceptions, Substance 3D Designer was used to create tileable textures, and Marvelous Designer was used for cloth-based props.

Object animations were created in 3ds Max, Marvelous Designer, or Unreal Engine. We used three animation formats depending on complexity:

1. Baked frame-by-frame animation (Alembic) for the most complex sequences and simulations.

2. Morph Target for simpler cloth simulations where two frames are enough.

3. In-engine animation using Blueprints for everything else.

How did you approach creating props and clutter assets that feel personal and story-driven rather than repetitive?

All locations are different, objects do not repeat, and every object is a surprise. On the Bar level, where there are lots of chairs and tables, there are no matching pairs – arranging identical objects ended up not being fun, since we are not an interior design simulator.

All locations are tied together through portraits, personal belongings, and character photos, from which you can imagine relationships between them and the story behind the location.

What techniques did you use to achieve the game’s soft, cozy lighting and color palette?

The game is built in Unreal Engine 5, and we used Lumen, so we do not bake lighting and do not even use Ambient Occlusion maps in the shaders. We used strong, warm sunlight in almost all locations – it is simple and works very well for creating a cozy atmosphere. We hardly used any additional light sources, always relying on a standard setup to achieve the most natural look possible:

We are currently working on nighttime lighting and would really like to add different weather conditions. Each location has its own color palette. The main rule is that it should look like a game and not an interior design catalog.

The game emphasizes tactile interaction with tools and objects. What technical or design work went into making those interactions feel intuitive and responsive?

Tactile feel is rooted in object animation: how an object is picked up, how it is put down, how it rotates, moves, and so on. This is handled by a special component that contains a set of other components within it:

For example, there is a component responsible for moving an object while it is being held. Here, we configure how heavily the object responds to movement: large furniture moves more slowly and with minimal tilt, while smaller objects react to the cursor faster and tilt more noticeably.

For convenience and realism, paintings automatically turn toward the wall, objects do not clip through walls, and large pieces of furniture cannot be placed on top of one another.

Hozy is often compared to games like Unpacking or renovation sims. How did you define your own identity within that space while still embracing familiar elements?

We moved the focus onto the gameplay flow. We have a narrative, yet it's more of a background bonus, not the main driver. We have renovation and putting furniture on the level, yet there is no economic benefit for it, and it's more about the process itself than full completion. 

Can you talk about the approach to environmental storytelling, especially as players restore spaces tied to people and memories?

We wrote a huge number of details – the characters, their history, circumstances, mood, and even their attitude toward the player. Yet all of that is mostly a background people will never see. We made those details to support the asset creation and not overwhelm the player with extra information they might not be interested in.

What were some of the biggest challenges or surprises during development?

Not a challenge, but a sad fact – normally, players interact with around 30-40% of the content and hidden mechanics we have in each level. 

Looking ahead, do you see Hozy evolving with more systems, content, or player-driven features, or is the goal to preserve its focused, minimalist experience? 

We actually have a lot of plans. We see players' reactions, and we know how we can expand everything, while still keeping our golden rule: tactile and fun to play. Right now, the two main areas we are focusing on are new levels with unique stories and a new type of experience sandbox.

We will try to develop these as different parts of the game, so that one influences the other, yet neither is mandatory for players to complete. If they like sandbox, maybe we will allow them to have all the content available from the start without completing the levels.

Come On Studio