3D环境艺术家详解《Silent Ruins》中巨型机器人与教堂废墟的制作流程

3D Environment Artist

Interviewed by

29 October 2026

Jueon Kim shared the creation process behind Silent Ruins, inspired by Studio Ghibli's Castle in the Sky. The artist explained the work with UVs and showed how to make damaged structures look more convincing.

Introduction

My name is Jueon Kim, and I am an aspiring 3D Environment Artist based in South Korea. I have always been fascinated by how spaces can create unique experiences, and I wanted to create environments that could offer those experiences to others. This interest led me to study architecture at the university. However, I eventually realized that designing real-world buildings was different from the kind of spaces I wanted to create.

I began looking for a field that would allow me to explore and express a wider variety of environments. Having loved video games since childhood, I felt that game environments would be an ideal medium for bringing my imagined worlds to life. This led me to pursue a career as a 3D environment artist.

At the university, I developed an interest in 3D through programs like AutoCAD, SketchUp, and Lumion. I later enrolled in a specialized 3D art academy, where I spent approximately two years studying the fundamentals of environment art production, including modeling, retopology, UV unwrapping, and texturing. I also incorporated AI as a supplementary tool in response to recent developments in AI technology.

The main software I used for this project included 3ds Max, ZBrush, RizomUV, Substance 3D Painter, Substance 3D Designer, Photoshop, Unreal Engine, and Marvelous Designer.

Silent Ruins is my first environment art project, created while studying game environment art. Centered around a cathedral and a giant robot, the environment features assets and structures that I modeled and textured, as well as damaged architectural elements that I created and placed throughout the scene.

In this interview, I will walk you through how I developed the initial concept and brought it to life as a 3D environment. 

Silent Ruins

The idea for this project began with my fascination with the overwhelming scale of monumental architecture. During an exchange program in Europe, I had the opportunity to travel to several countries and visit numerous cathedrals. I was deeply impressed by the scale and atmosphere of these spaces, and I wanted to recreate that feeling in a virtual environment and share it with others.

I knew from the beginning that I wanted to build a cathedral as the main setting. However, I also felt that a compelling portfolio piece needed a story behind its environment, not just a visually impressive space. I therefore began looking for a hero prop that could serve as the centerpiece of the narrative.

One of my main inspirations was Castle in the Sky by Studio Ghibli, particularly the scene featuring a giant robot lying dormant on the island. I was drawn to the contrast between the enormous, inactive machine and the peaceful environment surrounding it, and I wanted to capture a similar atmosphere in my own project.

To bring the cathedral and the giant robot together in a believable way, I developed a setting in which the remains of a massive machine had been left within the cathedral for so long that they had gradually become part of the environment in a post-apocalyptic world. This helped me combine the grandeur of the cathedral with the presence of a long-abandoned robot within a cohesive fictional world.

I gathered most of my references from ArtStation, Pinterest, and Google. When I could not find suitable reference images showing the angles or forms I needed, I used image-generation AI to create additional visual references. I also explored the cathedral's interior using YouTube's 360-degree video feature, which allowed me to look around the space from different viewpoints and better understand its architecture and atmosphere.

To create an environment where a giant robot had fallen inside a cathedral, I first built the cathedral and its main architectural elements before positioning the fallen robot. I then recreated the structures in the areas where the robot intersected with the cathedral as damaged and broken pieces, making it appear as though the robot had collapsed into the building.

To establish the cathedral's overall structure, I referenced the floor plan, elevations, and cross-sections of Chartres Cathedral. Based on these references, I planned the layout and proportions of the main architectural elements. I also adjusted the dimensions of major structures, such as the walls and vaults, in 10 cm increments to keep the measurements consistent while maintaining proportions close to those of the original cathedral. I then applied these measurements in Unreal Engine to establish the overall scale and structure of the environment.

Once the blockout was established, I refined and placed the main architectural elements to complete the cathedral's basic structure. I then positioned the robot and created additional damaged structures around the areas where it intersected with the surrounding architecture.

When creating the damaged structures, I primarily used Unreal Engine's Fracture Mode and Chaos Simulation. I used Anchor Fields and Bomb Fields to control which parts of a structure would remain intact and how other parts would break apart. By running simulations and reviewing different results, I selected the outcomes that best matched the direction in which the robot had fallen and the surrounding architecture. 

I also used Chaos Cache Manager to record the desired simulation results and convert them into Static Meshes for placement in the environment. This allowed me to produce a variety of damaged structures without having to model every broken piece from scratch.

To make the exposed cross-sections of the damaged structures look more convincing, I adjusted the Base Color of brick meshes sourced from Fab and placed them along the broken surfaces. I grouped nearby bricks into Packed Level Actors so that I could select and edit multiple bricks together more efficiently.

Mesh & UV Workflow

Mesh Workflow & Optimization

Since Nanite in Unreal Engine 5 can efficiently handle high-polygon geometry, I did not need to focus solely on reducing polygon counts as much as in traditional game asset workflows. However, I still wanted to manage the overall scene complexity while preserving the necessary details, so I used five different mesh creation and optimization workflows depending on each asset's shape and purpose.

1. TurboSmooth (Low → High)

I mainly used TurboSmooth in 3ds Max to create decorative patterns and ornamental shapes. It allowed me to modify the base mesh directly while checking the subdivided result, making it useful for assets that required intuitive shape adjustments.

2. Crease Set & OpenSubdiv (Low → High)

I primarily used this workflow for hard-surface assets in 3ds Max. By adjusting edge crease values, I could control the sharpness of edges and the shape of curved surfaces without adding extra support edges.

3. ZModeler & Dynamic Subdiv (Low → High)

I mainly used this workflow to create hard-surface assets in ZBrush. I first built a low-poly mesh using ZModeler, then refined the shape while previewing the subdivided result with Dynamic Subdiv. Once the shape was complete, I applied Dynamic Subdiv and adjusted the subdivision levels to obtain meshes for both high poly and low poly use.

4. DynaMesh & ZRemesher (High → Low → High)

I primarily used this workflow for organic assets in ZBrush. I first blocked out the overall shape using DynaMesh, then duplicated the mesh. I applied ZRemesher to the duplicate to create cleaner topology, subdivided it, and projected the details from the original mesh to restore the original forms. I then continued sculpting additional details at higher subdivision levels.

5. Decimation & ProOptimizer (High → Low)

This was the workflow I used most frequently. I first optimized completed high-density meshes in ZBrush using Decimation, then applied ProOptimizer in 3ds Max to further adjust the polygon count.

I also used this workflow when reworking AI-generated assets, refining their shapes with DynaMesh before optimizing them. Additionally, when exporting meshes from Substance 3D Painter after applying displacement, I enabled the Keep UV option in ZBrush's Decimation Master to preserve the UVs during optimization.

Rather than applying the same method to every asset, I selected the most appropriate workflow based on its shape and intended use. This helped me preserve important details while managing the overall complexity of the scene.

UV Workflow

I primarily used RizomUV for UV unwrapping. Its editing tools were particularly useful when working with complex and irregular meshes, such as those optimized with Decimation. However, I laid out the UVs for trim sheets directly in 3ds Max.

Texturing

I mainly used Substance 3D Painter, Substance 3D Designer, and Photoshop for texturing. Depending on the asset, I either created tileable materials in Designer or textured individual assets in Painter.

In Substance 3D Painter, I generally started by layering two or three Base Color layers to introduce subtle color variations across the surface. I then used Curvature and Ambient Occlusion (AO) to emphasize the asset's forms and structural details. Depending on the material, I added further elements such as rust, dust, cracks, wood grain, and emissive details.

For the wooden assets, I sculpted the surface details in ZBrush before texturing them in Substance 3D Painter. I aligned the UVs in a consistent direction to ensure that the wood grain followed the intended direction of the wood.

I created the tileable materials for the cathedral's brickwork and roof in Substance 3D Designer. This allowed me to create reusable materials for the roof tiles and the surface details of the brick walls.

For the trim sheet, I sculpted the required shapes and details in ZBrush before texturing it in Substance 3D Painter. I then reused the same trim sheet across various assets, including the cathedral walls, stained-glass frames, and fences. This reduced the need to create separate textures for individual assets while maintaining consistent details and materials across the environment.

For the robot, I first established the base metal material, then layered multiple Fill Layers to create the painted surfaces and rust. I added streaks of rust around areas where the paint had chipped away. I also used Anchor Points and Levels to distinguish the inner and outer edges of raised and peeling paint, helping me create a more convincing impression of paint lifting and flaking away from the metal surface.

Scene Assembly

After completing the cathedral's main structure and positioning the robot, I placed the smaller assets while considering the overall density of the environment and the flow of the viewer's eye.

Around the fallen robot, I placed brick meshes and piles of debris sourced from Fab to make the damaged areas blend naturally into the surrounding space. I also used Unreal Engine's Foliage tools to distribute vines and certain brick assets across larger areas. This reduced the need to place every instance manually while allowing the elements to spread naturally throughout the environment.

I then reviewed the remaining empty spaces in the main area and created and placed additional assets, including tombs, choir stalls, and chandeliers, to reinforce the cathedral's character and scale. Rather than simply filling empty spaces, I focused on using these details to emphasize the identity of the location and convey the atmosphere of a place that had been abandoned for a long time.

Conclusion

The project took approximately two years and three months to complete. It took longer than expected because I was learning several new programs along the way, including ZBrush, Substance 3D Painter, Substance 3D Designer, Marvelous Designer, RizomUV, and Unreal Engine. However, this also made the project especially meaningful to me.

I handled the modeling and texturing of the main assets, as well as the environment construction. My instructor helped with the final lighting, rendering, and post-production as we brought the project to completion.

One of the most challenging parts of the project was baking textures when the cage failed to conform to the intended shape. This was especially difficult with complex assets, where a single retopology method wasn't always enough to achieve the desired result. To overcome these issues, I experimented with different approaches and repeatedly tested which methods worked best for each asset. Through this process, I learned the importance of understanding multiple techniques and adapting flexibly to the needs of each situation, rather than relying on a single method.

The most important lesson I learned from this project was that there is no single correct answer. Different artists may approach the same object in different ways, and even the same artist may choose different methods depending on the situation. By studying other artists' workflows and learning from various tutorials and courses, I discovered the value of exploring different approaches rather than simply following one established method. Developing the ability to consider multiple possibilities and find solutions suited to each situation is essential.

Although I have not yet entered the industry, I would like to share one piece of advice with those who are just starting out in environment art: learn to love your portfolio project. Looking forward to seeing the final result and feeling excited about what it could become can be a powerful source of motivation. I also believe that this excitement can help you come up with new ideas and find creative solutions as you work through challenges.

Finally, I would like to thank my instructor for the guidance and support throughout the past two years, including the help provided in bringing this project to completion. I would also like to thank Emma Collins at 80 Level for allowing me to share the process behind my project.