Building an Abandoned, The Last of Us-Inspired Mediterranean House with Custom Materials

Introduction

Hello everyone, I'm Barış, a 3D Artist from Turkey, currently working as an Environment Artist at Misa Games on Misa Online, an upcoming MMORPG. I have a background in Interior Architecture and have been working in 3D for around five years, with the last few years focused primarily on game development and environment art.

I've been playing games and drawing for as long as I can remember. When I was choosing what to study, game development wasn't yet a very accessible or clearly defined path, so Interior Architecture felt like a natural direction for someone interested in drawing, design, and spaces. During my studies, working on architectural projects in 3D was what first made me realize how much I enjoyed building environments digitally.

That eventually led me toward game development. I started teaching myself the tools and workflows used in game art alongside my studies, and after a brief period working as an interior architect after graduating, I decided to pursue games full-time. Environment art was where all of this came together: my understanding of space and design and my passion for games and world building.

My studies taught me to pay attention to how spaces are constructed and how materials, scale, composition, and history shape the way a place is experienced. But beyond the technical side, what drew me to this field was the idea of leaving something behind: creating places that players don't simply walk through and forget, but ones that stay with them long after they've left.

Project Goals & Inspiration Behind Mediterranean Echoes

My main goal was to take my skill set to the next level and become a more well-rounded artist. Most of my previous work was hard-surface and sci-fi, so I wanted a scene that would push me in areas I had little experience in, particularly foliage, sculpting, and the more technical side of environment art, all at once.

I knew from the start that the scene would be inspired by The Last of Us, a game I grew up with and that left a lasting mark on me. I had also always been drawn to abandoned spaces where nature slowly takes over the manmade world. My first starting point was Danar Worya's concept, The Last of Us: Part II – Fungus Exploration, which shows a bathroom in that universe, but the scene soon grew beyond it.

Around the same time, I came across Katelyn Johnson's work on the Santa Barbara section of Part II. Its overgrown, bougainvillea-covered corridor had everything I wanted to explore: abandoned architecture, strong organic elements, and a believable relationship between the two.

Since foliage was my weakest area, it was also the challenge I needed to break through a plateau. I was fortunate to connect with Katelyn once my blockout and first base materials were in place, and feedback from someone who had worked on the original scene gave me a whole new perspective and pushed my motivation even further.

The result is a personal reinterpretation rather than a recreation. The original atmosphere and visual language remained my reference point, while the layout, corridor, bathroom, and many other elements were redesigned or built from scratch based on my own decisions.

References

Once the concept was settled, I moved on to references. The concept art and Katelyn's Santa Barbara scenes were my main sources, but I also needed detailed real-world images of the elements in them, plus some material that would let me expand the scene with my own ideas.

I broke the concept down into what I would need, created a category for each element, and gathered everything under those headings. This never really stopped. Since I worked from large to small, I needed more specific references as I moved deeper into the details.

Katelyn also changed how I approached this stage. She suggested mixing game references with real-world ones and interpreting both in my own style, which made me realize I hadn't used references much before, and when I did, I wasn't really processing them.

Blockout & Composition

I started with the bathroom. Since my starting point was a 2D concept, matching its perspective and working out real-world dimensions was the first challenge, and this is where my architectural background came into play.

Once a very simple blockout locked the angle, I blocked out the remaining objects to enclose the room. After deciding to lean into the Santa Barbara architecture, I stepped away from the concept's interior and carried the visual language of Katelyn's corridor into the bathroom. Studying Santa Barbara's bathrooms was decisive here.

Then I expanded the level into the corridor and, as soon as the base model was in, placed my main camera. To get the scale right, I went into the game, worked out Ellie's stride length from her height, and measured the space in-game, which was as crazy as it was fun.

With the main objects and angles set in both spaces, I wanted to add something of my own to the corridor, so I knocked down one of its walls to strengthen the sense of abandonment. I then added mid-sized objects to test the storytelling while still in blockout. I paid close attention to human proportions throughout: even without any technical knowledge, you can feel when a door handle or a light switch sits too low or too high.

Building both spaces together as one seamless environment was deliberate, since the goal was always to merge a reinterpretation with a space of my own design.

The blockout was also where I planned production. Looking at it alongside my references, I chose a mostly modular approach with unique treatment for hero assets and decided which surfaces would use trim sheets or tileable textures. Knowing early which models and texture types I needed kept the rest of the production clean.

Modeling & Technical Setup

I chose Unreal Engine to present the project in accordance with the current industry standards. Plus, I knew it would let me dig into things I wanted to learn in depth, such as material shaders, RGB masking, and vertex painting.

I modeled everything in 3ds Max and used Maya for retopology on the few assets that needed it. ZBrush handled sculpting and decimation; with Substance 3D Painter, I did the texturing, and then I used Substance 3D Designer for the materials. Toward the end, I added a few Megascans and Fab assets to support the storytelling, but they make up only around 5% of the scene.

I used the Datasmith plugin to build the whole scene in 3ds Max while testing it in Unreal at the same time, which made iterating fast and cut down a lot of import/export time. For the final stage, I converted everything to FBX so I could properly optimize the scene with tools like Packed Level Instances.

Modular Kit

With the blockout finished, I turned its architecture into a modular kit. I had kept its real-world measurements in whole numbers, mostly multiples of 2, 3, or 5, so the walls could be split into grid-friendly pieces without drifting from the blockout's boundaries. Straight, corner, and arched variations gave me enough variety.

The Broken Wall

The broken wall is the biggest departure from Katelyn's original scene: it's where the overgrown exterior starts to invade the interior. I wanted it to follow the layer logic of a real wall, from the brick and wood underneath to the plaster and paint on top, so I found a similar wall in the game and then drew an illustration that showed every piece I would need before modeling anything.

I set up the base in 3ds Max, sculpted the concrete, bricks, and wooden pieces in ZBrush, and decimated them so the silhouette of the low poly would still read as high poly.

The catch was that baking the high-poly detail onto a wall this large couldn't hold the texel density, so I took a hybrid approach. I built the plaster and brick materials in Substance 3D Designer, and for the wood, I sculpted each piece and laid them out as alpha cards in a trim sheet that I reused across the scene. I then baked the high-poly data onto a second UV set in Substance 3D Painter and exported RGB masks.

Everything comes together in one master material, which I'll cover in the texturing section.

Trim Sheets

I love working with trim sheets. They become one of the backbones of a scene and can carry a huge share of the workload. I ended up using two here: one for most of the wood in the scene, and one for the paintings and a set of props.

I believe a good trim sheet comes from the planning at the start: anticipating the scenarios it will need to cover before anything gets sculpted.

I first decided which objects would use the trim, then laid out their sections on a 4x4m plane in 3D, moving things around through trial and error until I knew how much space each one needed. For the wood trim, that meant wood ends, damaged decals, broken planks, wall and ceiling beams, ceiling planks, and a polished wood strip for the doors and the vanity.

Once the layout worked, I gave each section its own ID, took the plane into ZBrush to sculpt it, and then baked and textured it in Substance 3D Painter.

The second trim sheet covers the paintings and their frames, along with the marble and metal surfaces used on several props.

Assets, UVs & Texel Density

Apart from the broken wall, I only sculpted the wooden trim sheet, the curtains, the sofa's fabric and foam, and the foliage leaves. Everything else was low-to-mid poly modeling with clean topology, with most of the surface detail coming from the materials.

The biggest lesson here was about silhouettes: low-poly edges that show up in a silhouette break realism immediately, so I spent my polygons there and kept triangle counts low everywhere else. Since many of the hero assets were hard-surface, modeling ended up being one of the fastest stages.

For texturing, each asset got the approach that suited it best: trim sheets with RGB masks, unique textures, atlases, or tileables with vertex blending.

I also wanted to work as if I were restrained by a studio's budget. Apart from Lumen, I stuck to a traditional workflow and didn't use Nanite.

I set the texel density to 512px/m on 2K textures and kept the same general target across the project. Where a piece truly needed it, I allowed a small deviation and scaled those UVs down without hurting the visual balance.

2K textures were the norm everywhere across the project, including hero props, and I only went higher for close-up shots. Smaller, less visible objects got atlased and lower-resolution textures to keep the texture budget in check.

Texturing

Material creation was one of the biggest challenges of the project, and deliberately so. Substance 3D Designer was an area where I felt weak, and the idea of building my own materials in exactly the aesthetic I wanted made even a program with a painful learning curve feel doable.

I started with the materials that would take up most of the screen: the floors and walls. Before going into detail, I built a base version of each to make sure every material in the scene spoke the same visual language, because I didn't want to spend days on a material only to find it didn't fit in the end.

Once the set worked together, I detailed each material and exposed the parameters I needed so I could control them however I wanted, which is one of the things I like most about Substance 3D Designer.

All of this comes together in the master material. It combines tileable textures, unique masks, baked data, and vertex painting, and its shader complexity still stays in the green.

On every asset that needed it, the RGBA mask follows the same logic: red drives mud and dirt, green is responsible for edge damage and wear, blue controls roughness or color variation, and alpha handles the moss layer.

Each channel's color and roughness are exposed as parameters, so the same surface can be pushed into very different looks. It took a lot of trial and error, but in the end almost everything in the scene uses this single material.

Here are three examples of how the system works in practice.

Hero props were where Katelyn's feedback made the biggest difference. My level of detail had stayed fairly basic, and she showed me how to push it further and how to add the kind of details that actually sell realism, efficiently. That changed my whole texturing process, not just the hero props.

Foliage

Foliage was something I tried to solve from the earliest stage, yet it kept changing until the very end. Understanding the system was overwhelming at first, and Katelyn walking me through how the original scene's foliage was built and placed changed how I understood game-ready foliage. Still, having it planned in the blockout helped shape the environment.

The most valuable thing Katelyn taught me was that controlled foliage placement comes from building it modularly. I first tried SpeedTree, but when its procedural tools couldn't give me the shapes I wanted, I went back to a traditional workflow.

I started by sketching the atlas, modeled the branches to match it with subdivision modeling, sculpted the leaves, and textured everything in Substance 3D Painter.

From there, I made low-poly cards from the leaves and lined them up along a long, straight branch that tapered toward the tip. I then traced the branch shapes from my sketch as 2D splines in 3ds Max and used the Path Deform modifier to wrap the straight branch, leaves and all, onto those guides. Every branch fell into place quickly, and I could still adjust the rotation and shape of each one in 3ds Max.

Up to this point, the ivy and the bougainvillea shared the same process, but I finalized them in two different ways. The ivy kept a separate mesh card for every leaf, while for the bougainvillea I baked the finished atlas in Marmoset Toolbag and cut it back into low-poly clusters, which went into the scene as baked pieces. Together, they formed an ivy and bougainvillea pack I could combine however I wanted to dress the scene.

The ground foliage took a different path. I first built it to match The Last of Us, but once the rest of the assets reached their final state, it started to look slightly stylized and stood out next to them. Rather than lose more time on it, I switched to the Megascans ground foliage, a change you can spot in the corridor progression GIF further down.

Scene Assembly

Once the foliage was finalized, I hand-placed it into the scene cluster by cluster. For every plant, I thought about how it would actually grow and how that kind of growth appeared in my references, and I used it both to strengthen the atmosphere and to guide the eye through the frame.

That left only the set dressing and lighting to complete. All the props you see in the breakdowns are my own, and I spread them through both rooms to strengthen the storytelling and give the space a lived-in feel. Watching everything finally come together was when the scene started to feel close to finished, and that made the last stretch the most exciting part of the project.

Lighting

Unlike Katelyn's original scene, I wanted mine to have a golden-hour atmosphere, close to what you see at sunset. Mediterranean architecture usually carries a warm, welcoming feeling, and I wanted to show this abandoned version of it being embraced by nature in that late-day warmth.

Inside, the goal was to let the greenery spill its color into the rooms in a controlled way, while atmospheric fog made the storytelling a little more dramatic.

The setup itself was simple: a Directional Light and an HDRI, with fake Rect Lights at the windows to make the interior lighting feel more natural. Lighting interiors with Lumen can take more effort, so I based my values on real-world light intensities and aimed for lighting that was both natural and optimized.

Katelyn guided me here as well whenever I needed it, helping me get the tone and colors across more clearly.

Rendering & Post-Processing

Everything was rendered with Lumen. I kept post-processing to a minimum because I wanted the image to rely on physically based lighting rather than heavy adjustments: the Post Process Volume only handles exposure and a slight saturation tweak.

The final touches were done in DaVinci Resolve.

Final Thoughts

Mediterranean Echoes took me around five months to complete. It wasn't a focused rush: there were many areas I had to figure out through trial and error. Looking back, the biggest challenges were the areas I chose on purpose: material creation in Substance 3D Designer and foliage. Both were overwhelming at first, and both kept changing until the very end, but pushing through them is exactly what made this project worth doing.

A few lessons stayed with me. Planning pays off: deciding in the blockout which assets would be modular, unique, trim-based, or tileable kept everything that came after much cleaner, and building base versions of my materials before detailing them kept the whole scene speaking one visual language.

It also pays to know when to let go. Swapping my ground foliage for Megascans once it started to stand out gave me back time I needed for the rest of the scene. And small things decide realism, whether it's a low-poly edge breaking a silhouette or a light switch mounted at the wrong height.

If I had to give artists starting out a piece of advice, the first would be to choose a project that targets your weakest areas. The second would be to look for feedback from people whose work you admire. Reaching out to Katelyn early in the process completely changed this project: from how I used references to how I built foliage and detailed hero props. It's fair to say that her guidance shaped almost every stage. Don't be afraid to ask questions.


I'd like to thank Katelyn Johnson for her time and guidance, my friend Idris Talibov for his help with the peeling paint material, everyone who shared feedback along the way, and Emma Collins and the 80 Level team for the opportunity to share this project.

A special thank you goes to my partner for her patience and support through all the late nights. If you've made it this far, thank you for reading. I hope Mediterranean Echoes turns out to be one of those places that stays with you a little after you leave it.

Barış Kılıç, Environment Artist

Interview conducted by Emma Collins