Crafting Elden Ring-Style Cathedral Environment Inspired By Real-Life Architecture

Introduction

Hello everyone! My name is Horia Mocanu, and I'm currently a Senior Environment Artist working at Ubisoft. I've most recently worked on The Division 2's Battle For Brooklyn DLC and Star Wars Outlaws. I'm mostly self-taught, with a background in Industrial/Automotive Design.

A couple of samples of my professional work:

Sometime during university, I lost interest in car design, but I did enjoy the 3D modelling component I was taught in one of the modules. Games have always been a big part of my life, so I decided after finishing my bachelor's to try and make something out of my love for 3D and games, eventually landing on Environment Art.

Ancient sketches I managed to dig up from my uni days:

I've gone through a variety of 3D software, starting with Alias, Rhino, and Cinema4D, then Blender, and finally Maya. Nowadays, I mostly stick to Maya for my modelling needs, hopping back and forth between ZBrush, if need be. It's kind of crazy looking back at how far I got in C4D without having a clue about UVs, materials, or proper modelling/topology, just vibes.

It's difficult to pinpoint how I got the skills I have, as I think an Environment Artist's most valuable asset is their taste and their sense for composition and art direction. The tools you use and the techniques you learn along the way to get you to the final look are important too, but for me, composition takes priority. Collaborating with all the neighboring job families and learning about what Level Design, Concept, or Tech Art, etc., look out for gives you a more well-rounded perspective on what a good environment should feel like, and I think this helped me grow as an artist a lot.

I feel that the discipline of cinematography is very similar to game art. You have Set Dressers, Lighters, VFX Artists, Storyboard Artists, Directors of Photography, and so on, all of which have analogs in the games industry. Because of this – and my love for cinema – I often find myself studying the way films are shot and trying to understand the artistic intention behind compositions. I'm naturally a very curious person, so this inquisitive mental pattern applies to photography, paintings, sculptures, and music as well. The latter are just more difficult to apply to your work.

To name a few cinematographers I’ve drawn a lot of inspiration from – in no particular order – I recommend looking into Jessica Lee Gagné, John Alcott, Dante Spinotti, Christopher Doyle, Autumn Durald, Franz Lustig, Jarin Blaschke, Michael Bauman, and Jingsong Dong.

Some of the incredible physical set design for Alien Romulus (2024) by Neeman Marshall and his team: 

The Marika's Cathedral Of Gold Project

There were a lot of factors that led me to start the Marika's Cathedral of Gold project. I have a deep admiration for Europe's older churches, and earlier this year I took some time off travelling and visiting a lot of them. I would often find myself staring at the architecture inside and outside these cathedrals, trying to wrap my head around how they were built centuries ago (this still breaks my brain sometimes), and how I would create something like that in 3D.

I took a lot of reference photos, sketches, and notes during these trips, and as they kept piling up, I eventually decided to try and put everything together, in some sort of architectural church study in UE5. At the time, I was just coming off replaying Elden Ring's Shadow of the Erdtree DLC and wondering if I could put together a scene that matches the game.

My earliest Inspirations for the project's vision. The framing of Vicar Amelia's cutscene really stuck with me:

At the same time, while working full-time, I felt I was becoming a bit out of touch with Unreal Engine, so I wanted to use this project as an opportunity to do a deeper dive into the newer systems I wasn't as familiar with, such as Nanite, Lumen/MegaLights, and see how performance holds up. There were a bunch of geometry-heavy asset creation techniques I wanted to put to the test as well, but more on this later. 

I'm a huge Elden Ring nerd, so a lot of the research you usually do when starting a new project was already stored in my head; I just had to go in-game to all the locations and gather screenshots. I always tell people – especially on personal projects – to not try and do everything at once, and so while I still exercised some art direction with this project, the game's already established visual language did a lot of the heavy lifting. My job on this front was mostly to try and tastefully incorporate this language into my scenes' compositions in a way that made sense.

I filled out my reference board with both Elden Ring screenshots and pictures I took of churches on my trips, as well as some miscellaneous references I found useful along the way:

Blockout

Elden Ring's color palette tends to be quite monochrome, which was a bit of a blessing and a curse. I didn't have to worry as much about colors and more just about values and making sure all the materials look like they belong together.

Mostly monochrome spaces that rely on lighting and local values to create contrast and clarity:

I wanted the environment to be an actually playable space, so I spent most of my blockout time trying to figure out the cathedral’s proportions and scale. It was very interesting to observe how space perception changes based on the spacing between the pillars and their height. The example below is similar to Stephen's Dome in Vienna, but it turned out way too big for what I was looking for.

Scale perception in-game vs IRL analog:

More deliberate composition work only came later. I'll list below a few composition considerations I kept in mind when shaping my scene. These are usually at the back of my mind at all times when working on a scene, and I gradually address them as the scene progresses.

Illustration of different forms of contrast. Using these, you can plan out your focal points and somewhat guide the eye across the scene in a more controlled way:

Below is an early blockout example where shape and scale contrast are already in play to some extent. The top-down lighting really wasn't working in the composition's favor either, a lot of the scene being flat and drowned in similar values. More on this later.

Shape contrast example:

Anticipating how complex and noisy spiral shapes can be, deciding what my pillars would look like was my main objective during the blockout phase. It was also very important to place my models in a level and see how the pillar repetition holds up from different angles.

After trying out several different silhouettes, I would go back and forth between the trimsheet and the spiral, adding and modifying the trims to reduce noise and improve readability. This part has to be done fast and loose, and ideally before moving on to sculpting the high-poly. The more iterations and trying out different shapes and trims, the better the final result will look. I learnt a lot about proportions and scale during this process.

Differences in the readability of the space between different pillar shapes. Lighting and perspective/FOV should also be accounted for.

Pillar Study

I get headaches just recalling my struggle figuring out my asset creation pipeline and shader setup. The brief was relatively straightforward: Trimsheets for most surfaces, with a baked normal and mask map overlaid on top. Use unique texture bakes as a last resort.

In practice, however, I was really struggling to keep track of the steps required, so I decided to take my pillar asset and put it through the pipeline, all the way to final quality, almost like a benchmark. This gave me confidence that the plan worked, and helped me adjust steps along the way as I found mistakes. I figured a lot of the unknowns would be answered by going this route. 

I think, despite the more complicated pipeline, trimsheets helped me save so much time in the long run. On top of the fact that you can't really texture big assets without them, once you have them, texturing any architectural piece becomes so easy.

Once the pillar was done and I had my master material shader set up, I moved on to spending a bit of time on the Marika statue. I had to figure out its scale, pose, and how to stage it in the scene before I did any serious sculpting. 

The original setup, inspired by the Basilica Collegiata in Sicily, ended up being too crammed, taking away from the statue's importance. I decided to scrap it altogether and try something simpler. At this point, I had a couple of angles I was happy with, so as I removed elements from the scene, I would check how the silhouette reads from those angles. The initial lighting setup featured a light cast from the dome opening in front of the statue, but it never ended up looking right, no matter how many elements I would remove beside the statue. That's when I made the tough decision to remove the back side of the church and let light in.

Progress on the arch and statue framing, as well as silhouette iteration:

Topology

My retopology ambitions for this project were rather modest. I wanted to push Nanite to its fullest extent, so most of my sculpted assets were simply passed through ZBrush's Decimation master and then cleaned up in Maya to save time. Unwrapping would also initially be done in ZBrush and then cleaned up and organized in the appropriate UV channels in Maya.

I like to work within the confines of a brief, similar to how you would in a studio environment. This means time is a valuable resource, and assets are made with specific uses in mind. Details are added deliberately to guide the eye and create focal points. Conversely, details are removed in order to create eyerest. For instance, the main pillars are very detailed and dense assets, both visually and technically. They are the supporting focal points that you see first, after the statue. I want you to look there and be rewarded for doing so.

Other assets, like my small props kit, were made to provide scale reference for the pillars and the statue. They are reasonably detailed, but they are all fairly basic assets; I want you to see them, but I don't want them to distract you. This sort of intentionality I apply to all stages of my work, to keep the composition focused and to save time.

It was a bit of a challenge figuring out how to tackle the small props, since I didn't know how many I needed, how big they would be, or what they would look like. I started filling out the texture one trim at a time, starting with the benches.

Small and medium props laid out and assembled:

What I found working out in the end was making somewhat of an "asset request" list, 
modelling a rough blockout of it, and making a mental note of the UV space it would take up. I also looked into how much effort it would take to produce the prop out of trims or make it unique. This way, I could keep an eye on when my atlas sheet would fill up, and I'd have to make do with the props I got.

The drapes were an element that was part of the main design from the project's early beginnings. Its purpose was to fill out the otherwise empty top part of the scene, direct the eye towards the statue, as well as continue the drape motif found in other places in the game. Despite the Leyndell banner being originally in shades of yellow, due to readability, I opted for the alternate red color, also seen in some other banner variations.

Drapes and rugs references:

I modelled both the drapes and the carpet in Marvelous Designer and then cleaned the meshes up in ZBrush. Texturing the carpet was really fun, because it was the first time I set two textures like that up. The idea is to create a middle texture and then a variation for both ends and align the UVs along it. It's not perfect, but it gets the job done.

Example of the drape UVs, with torn cloth cards placed by hand along the rips over a masked trim. The tassels are also placed over an alpha masked trim for both the rug and the drapes:

Texturing

Texturing and modelling large assets was a fun change of pace from the other assets. Because of the Statue's scale, using a traditional prop texturing method was out of the question, so I made use of a baked normal map and a packed RGBA mask to keep my sculpt details. These would go onto UV1, while UV0 would have the appropriate texel density for the tileable materials.

The chalice was made in a similar fashion. Modeling the high-poly proved a bit tricky, requiring a mask mapped over the UVs of a base version of the chalice. I used this mask along with the Inflate modifiers in ZBrush to get the details I wanted.

Chipped Edges Trimsheet Technique

Some time ago, I came across an asset creation technique by the talented Raul Aparicio used for God of War (2018), which used the geometry of a trimsheet, along with the texture, to create world modules. In simple terms, it is very similar to the more common edge chip damage normals, but with a bit more geometry to hold details.

I prefer this method for surfaces close to the player camera, as the geometry holds details much better and also creates much better lighting results. I'll use this Edges Trimsheet I made a long time ago as an example of the workflow.

The sheet is set up for 512x1m texel density, then sculpted in ZBrush and baked like you would normally. The final asset is then sliced and used along the edges of your surfaces. The goal is to keep a UV channel for the baked edge normals, and the other for the tileable texture that goes over it. If done right, the seam is barely noticeable. Below is somewhat of a step-by-step of the process on a floor pattern asset that got cut in the end.

Using this method is how I approached a lot of the bigger surfaces in my scene. The walls were all made with the first two trim sheets, and the floor and round pattern with the last two. I used bend deform and curve warp for surfaces that weren't straight.

Presentation

To enhance and sell the Souls-like look, I went the extra mile and added Elden Ring's UI to my gameplay clip. I put together a couple of transparent cards and timed them with the footage in Premiere. I also used the Gothic knight character by TalkingDrums and tried to match the FOV and camera height of the original game.

Over the years, FromSoftware has become an expert in scale and describing large structures.  Below, I tried to break down some dressing approaches that stood out to me. I tried to apply a lot of these, along with the use of repetition, depth planes, and other compositional principles, to decide where details go.

Lighting

Lighting was one of the skills I wanted to work on the most with this project. Trying to understand a light source's function and its impact on composition, as well as its impact on performance, has been very fun to figure out. I'm by no means a Lighting Artist, so most of the notes below are just personal findings based on composition fundamentals and instinct. I started by breaking down lighting into a list of primary, secondary, and tertiary functions.

Primary function. This is the scene's main light source, in our case, the sun casting from outside into our cathedral.

• Provide value contrast between the statue and the background, ensuring the statue is the scene’s main focal point;
• Provide highlights and complement the pillars, the scene’s secondary focal points;
• Provide shadows and help ground the elements into the scene.

Secondary function. These are fill lights that help light the scene up where the sun doesn't reach.

• Provide highlights and shadows for the surfaces it casts on;
• Serve as guidance, directing the player through the space;
• Help balance out composition values, making sure spots are not too dark.

Tertiary function. These are small, short-reach lights placed to catch your eye with some highlights and high-value contrast. Used sparingly, these are a great way to add interest points in areas of your scene.

Quick sketch to hopefully illustrate light hierarchy, direction, and purpose:

I prefer keeping a somewhat neutral tone in my UE scene and handling most of the post-processing in DaVinci Resolve. I find it can be quite distracting and misleading to work with a heavy post-process volume inside UE. Below is a comparison of the touch-ups I did. It's only at this point in the process that I start treating the scene like a 2D picture and fixate on small details and locally adjusting values, etc. Despite this, the final result still stays somewhat true to the in-engine version.

Sample of the very basic still I used for most of my shots:

Conclusion

The project’s idea came to me in December of 2024, and I sat with it for a couple of months, until I ended up visiting Stephen's Dome in Vienna in April 2025. That's when more of the pieces of what I wanted the scene to be started falling into place, and I felt I had enough to start a project with. In production terms, I would say preproduction was from December to April (4-5 months) and production time was from May to November (6-7 months).

I think overall I did a great job of not overscoping my scene with too much stuff I'm not comfortable enough in. The main challenge I saw coming from the beginning was having to figure out how to set up my asset pipeline and how to produce my assets according to it.

The planning ahead this required made me not work on the project for days on end. The other challenge I accounted for was my small props atlas. I had set out to only use one texture set for all my props, so creating an atlas/trimsheet without knowing exactly how many props I want and what they would look like was very intimidating initially. I learnt a lot about making meaningful prop blockouts from this.

The somewhat unaccounted-for challenge lay in the composition. I set out to create a cathedral that blended Gothic and Baroque architecture with Elden Ring motifs. This gave me a very straightforward framework to build out my scene. A lot of that went out the window once I removed the room's backside, and I had to think on my feet about how to justify the newly formed balcony, as well as how it impacts the rest of the scene.

I sometimes look back at some of my very early projects as a beginner, and I think to myself, "Wow, good effort, but I did not have any clue about composition." This made a lot of my efforts lack focus, direction, and vision, spending needless time trying to do everything at once. It's still important to explore all the neighboring disciplines you will have to work with in production (level design, tech art, materials, props, lighting, concept, etc.), but definitely don't try to tackle everything at the same time. Scoping down your projects to only cover a couple of topics you want to know more of, while mostly dealing with what you're already comfortable with, helped me grow the most.

That about sums up the journey I've been on with this project. Congratulations on making it to the end, and thank you for your time!

Horia Mocanu, Senior Environment Artist

Interview conducted by Gloria Levine