Home 3D Art Showcase & Critiques

Abandoned Dacha Environment - Modular Kit with interiors

polycounter lvl 6
Offline / Send Message
Frozishe polycounter lvl 6

Hey everyone!

I’m a Senior/Lead Environment Artist with about 7 years of industry experience. For the last 3.5 years, I've been heavily focused on building modular environment systems for a VR project.

Anyone who's worked in VR knows the constant struggle between visual aesthetics and aggressive optimization (though to be fair, that's not strictly limited to VR!). So, for my portfolio, I decided to take these exact game-ready assets and render them in UE5 using Lumen just to see how they hold up without VR limitations.

A bit about the workflow and technical side:

Modeling: Everything was modeled in Blender.

Modularity: The system ended up being extremely flexible. From this single kit, you can build several distinct villages with completely unique houses (small shacks, massive 2-3 story buildings, varying roof sizes, etc.).

Texturing: Created from scratch, mostly in Substance Painter (I personally prefer it over Designer for my workflow). For some base textures, I used Megascans normals as a starting point, but heavily reworked them and added a ton of custom layers and grime on top.

Engine: Fully assembled and lit in Unreal Engine 5 using Lumen.

This is actually my first time posting my work publicly in 7 years. Emerging from "production hell" to finally show what I’ve been making is a pretty emotional moment for me!

If you want to see the full technical breakdown and more renders, I’ve uploaded the full project on my ArtStation: https://www.artstation.com/artwork/8BAOPE

I'm totally open to answering any questions about the VR pipeline, texturing workflow, modular systems, or just my experience in general!

Replies

  • Frozishe
  • sacboi
    Offline / Send Message
    sacboi grand marshal polycounter
    I've no crits - your material observation is very on par and pretty much faultless :+1:    
  • Frozishe
    Offline / Send Message
    Frozishe polycounter lvl 6
    sacboi said:
    I've no crits - your material observation is very on par and pretty much faultless :+1:    
    Thanks ) 
  • Vastra
    Offline / Send Message
    Vastra polygon

    Really impressive work!  =) Especially the modularity and how well the assets hold up outside of VR constraints. 


    I’d be very interested to hear more about the VR pipeline you mentioned. How much did it change your approach compared to a more traditional environment pipeline, particularly in terms of geometry, materials, and modular design?

  • AlexandrL
    Offline / Send Message
    AlexandrL polycounter lvl 8
    Hi, Frozishe
    I heard on a different platform what you was using UE Materials Layering System.
    Can you please elaborate a bit more about that expirience? Did you or other artists in the team expirienced any issues?
    There was so many times I've personally heard that Layering System is undercooked, buggy and some teams was refuze to use it at all, so I believe it would be really intheresting to hear about positive expirience and an example where Layering System was succesfully used in shipped product. 
  • Frozishe
    Offline / Send Message
    Frozishe polycounter lvl 6
    Thanks everyone for the kind words and the great questions! Let’s dive in.

    Working in VR is truly like walking on a razor's edge. It’s a technical hell where every decision must be calculated ten steps ahead. Performance requirements (high resolution across two displays + high frame rates) leave no room for error, and you have to layer all of that on top of specific gameplay constraints. For me, this was a serious but fascinating challenge.

    Metrics and Scaling
    The first major task was fine-tuning all the scales to match the VR character metrics. In a headset, volumes feel completely different, and standard "flat" game metrics often fail. I had to ensure rooms didn't feel cramped, doorways were wide enough to navigate with a weapon drawn, and ceilings didn't feel oppressive while avoiding that "medieval castle" look.
    At that stage, the player dimensions were still being tweaked. We actually had a situation where the game designers decided to increase the standard character height, and suddenly, the player couldn't fit into the attic anymore. The capsule hit the sloped roof. You had to crouch just to move forward. Luckily, we caught this during the blockout stage and fixed it quickly. The main lesson here: test everything in the headset constantly. The demand for "environmental truth" is far more critical in VR than in traditional games.

    Hybrid Mid-Poly Pipeline
    I implemented a flexible process by blending the mid-poly approach I used back at Call of Duty with the harsh realities of VR. In virtual reality, players can lean in and inspect every corner, so the priority was a polished silhouette.
    We completely ditched "heavy" features like Parallax Occlusion Mapping (POM), displacement, or complex shaders. I had to find that sweet spot where the geometry was dense enough to look smooth, but the overall polycount remained as lean as possible. You can't rely solely on LODs in VR either. The headset is extremely sensitive to LOD popping which is a total immersion killer.

    Exterior Walls (Shells) and Interiors
    In the final version, once the building layouts were approved and finalized, individual modular wall pieces were merged into solid meshes (Shells). This drastically reduced the actor count in the scene (essential for our high-density maps) and eliminated lightmap seam issues. I ended up with 7 main shell configurations for all building types.

    Each house required a different number of interiors with varying layouts (you can see how they swap inside the same shell in my ArtStation video https://www.artstation.com/artwork/8BAOPE ). The interiors were made as separate "Boxes" single-sided geometry with inverted normals. These interior walls have a minimal offset from the shell walls to prevent z-fighting and ensure bullet decals project correctly.


    Visualizing the 'Shell & Box' approach. On the left: the merged exterior shell. On the right: the single-sided interior room box

    Lightmaps
    The separation between shells and interior boxes was primarily driven by lightmaps. Using a single mesh for both would require massive atlases to get the necessary density. After the split, we pack the exterior shell faces into the available space, while the interior faces of that same shell are treated with minimal resolution (acting only as plugs to prevent light leaks). The interior boxes get their own high-density lightmaps because they contain more props and local light sources.


    Lightmap density optimization. Blue indicates lower resolution for the exterior shell, while green shows higher density for the interior surfaces where players spend most of their time


    Material Layering
    You’re absolutely right about Layered Materials. They are both a blessing and a curse. From a workflow perspective, they’re great for creating rapid variations. You can stack layers, swap blend modes, and get results quickly. But under the hood, there are plenty of pitfalls, and our team definitely ran into them.
    It’s crucial not to use layered materials where they aren't needed. If a building is "clean" gameplay-wise, don't use a 3-layer material. Even if those layers aren't painted in, the shader cost is still calculated in full. It’s much more efficient to swap the instance for a basic master material. The same rule applies to props.
    In VR, we encountered bugs where layered materials would cause certain meshes to turn completely white. A quick "band-aid" fix was to override texture maps inside the material instance (checking all the boxes for every map). It was strange, but it worked. Eventually, we brought in the programming team to resolve the underlying technical issues, while we also transitioned the assets the assets back to standard material instances wherever layered materials weren't strictly necessary.
    For comparison: a simple master material might have 247 base pass instructions, while a layered material (even with just a background layer) jumps to 273. It might seem like a small difference, but on a massive VR map, every single instruction saved helps the budget.

    Environment art in VR is less about painting and more about engineering. You have to find that balance where the visuals look like a modern PC title but consistently stay within a tight 11ms frame budget
Sign In or Register to comment.