Using Weather Shifts to Set Up an Atmosphere For a Medieval House Scene

Introduction

Hey! I'm Finn, a 20-year-old and highly motivated Technical Environment Artist, currently finishing my studies at Howest Digital Arts & Entertainment in Belgium. As I approach graduation, I'm looking for an end-of-study internship to further develop my skills and gain real production experience.

I'm very passionate about games and art, especially the magic of creating worlds so visually striking that anyone can appreciate them. Although I'm still new to the industry, with around three years of 3D experience, I've completely fallen in love with environment art. What started as a study quickly became a true passion, something I enjoyed all day, every day, something that I don't just do for grades or an extra line on my CV, because I genuinely enjoy improving and pushing my craft forward.

This project was created as part of a mentorship program at Sierra Division Academy, where I took the Crafting Environments: Career Kickstarter course led by Alex Beddows.

Medieval House

As mentioned before, this project was part of a mentorship course at Sierra Division Academy. I am very grateful to be part of Alex Beddows' first class here, and I have learnt so much that I am extremely happy with. So, I knew I wanted a killer piece on my portfolio, a piece that could set me apart.

Since I have already experimented with some different styles, this time I wanted to go for something realistic and grounded, so I went to work. Before I looked for a concept, I analyzed what I wanted to make, this way, I could have an easier time looking for a concept. Mainly looking into pipelines and styles done in different companies, and also just to be aware of what is really out there!

The main inspiration for this environment came from Junhyuk Yoon's amazing concept. The concept honestly blew me away. It's not much, just a house, but there was something about it. The execution was just so beautiful, I knew I had to make this right away.

Of course, in any project, the first step should be to find references, so that's what I did. Another tip for this stage is to really break down the reference and work part by part. I usually use drawovers to already "divide" my concept into different meshes and materials. This way, I'd always have some reference on hand.

Now that I've figured out why I wanted to make this, I can get to work. My first steps in any project are usually to match my camera settings in fSpy. This application is very useful and easy to get into. After setting up my camera, I just made a quick blockout in Blender and imported it into Unreal Engine to visualize it.

The key here is not to think too much about the details, and really only try to look at the bigger picture. I got the shapes down quite quickly, and it was looking very nice already in my opinion, good start!

Sculpting

I sculpted the asset in ZBrush. Since I had little to no sculpting experience before this project, it was definitely a challenge, but one I really enjoyed! I mainly used the Trim Adaptive brush to create the main shapes and forms, and then refined the surfaces using some Flat brushes for cleanup. I didn't focus too much on micro-detailing, as I planned to use tiling textures with Normal Maps later on to achieve the finer surface detail.

I then decimated them and got them into Unreal Engine. The focus here was to really get in the macro read, without having to spend too much time yet. Only if I've validated that it looks good, I would go on to retopo, UVs, and bakes.

As you can see, many of my assets are built from a set of small rocks I created. I wanted a workflow that allowed me to assemble the full structure directly in Unreal Engine, while being able to reimport and update just the four rock meshes as needed. Simply grouping or merging assets didn't give me the level of control I was looking for.

That's when I discovered Packed Level Actors (PLAs). These essentially act as prefabs or blueprints that store all the meshes and instance any duplicates, great for both flexibility and performance.

Another major advantage of using PLAs is that they can be exported as a single mesh. This turned out to be extremely useful: since my rocks were arranged directly in Unreal, I didn't have a sculpting reference for the mortar. By exporting the PLA, I was able to bring the entire structure into ZBrush, sculpt the mortar around it, and then export only the mortar mesh back into Unreal to integrate seamlessly with my PLA setup.

I used this technique for the chimney, the little hut, as well as for the wood planks used for the door, the side roof, and the cart.

Discovering this allowed me to create many complex assets in Unreal using my modular kit, with lots of modularity. This was good because it kept my production time good within the given timeframe.

I decided to use auto-generated UVs in Blender for this project. I baked my high-poly mesh from ZBrush onto a decimated, auto-UV'd low-poly version. I'm aware that this isn't the ideal approach for achieving perfectly crisp close-ups, but it was a time-saving choice, and the results still turned out well.

Decisions like this are exactly why I focused on making a clear, consistent pipeline from the start. These weren't choices made on the spot, but part of a structured workflow I followed throughout the entire process. 

Texturing

This whole scene was textured with Substance 3D Designer. I really wanted to improve my Substance 3D Designer skills, and this was a perfect project to do so. Here's a preview of some of the materials. Of course, there's way more I wanted to refrain from using any megascans for this scene.

Making the materials fit together was a big focus point. I did this by using similar grunges and patterns. I always work from height first, then roughness, and then base color. Making the patterns is usually a lot of experimenting for me till I get it right, but I feel like I did a pretty good job here.

Of course, I needed more individual control, which I managed to do with RGBA masks and decals, which gave me the control I needed to use tiling materials and still have a beautiful result.

Also, for the individual props, these masks came in handy. I managed to make the cart with just my kitbashed stone planks, here's a preview of that:

Here's a small preview of how much the decals impact the look of my scene:

A recurring thing I usually like to implement in my projects is some sort of mood/weather shift. My previous article at 80 Level was also one where I did this!

I personally think it’s very cool to visualize a scene in a different scenario, also, it gives some freedom to move away from the concept art!:

Just after setting up a quick z-up shader and implementing the use of Ultra Dynamic Sky (a UE5 plugin that is extremely useful for snow effects), I managed to achieve this look. I instantly knew this had the potential to be extremely cool.

So I went to work, I refined my snow material and shader, added a few subtle storytelling elements ( smaller trees, no hay, shovels, etc), and I touched up the lighting. I really, really loved the vibe of this scene. It felt very tied together, and it just worked.

After getting in all my materials, the scene was still very far from finished. It started to go into assembling an actual scene now. I started with the landscape. This scene doesn't have that much terrain, but for the little part you could see, it was very important that it read well and looked good. I decided to just use the landscape painting modes to make my path.

For the foliage, I decided to use Dekogon assets. These were a good option for me as I didn't really have an interest in making foliage myself, and they would've taken too long for the time I had, so asset packs were the best option here for me. However, out of the box, they really looked out of place. I needed to make a lot of adjustments to get them right.

Just changing the colors already did a lot! But we are still very far off.

At this point, my scene was still looking very bad, I couldn't point at it right away myself, so I went to ask for feedback. Luckily, I had my mentor, Alex Beddows, and my friends to help me out with this.

The key to an art block like I was having here is asking for feedback. There were a few feedback points I was constantly receiving: 

• Where's the AO?
• Lighting is flat (will leave this for the last pointer).
• Composition?
• Material Response.
• Ground texture is too blurry (We changed this in the previous point).

This part of the process did prove to be quite difficult. At this point, I lost some faith in the project. I wasn't too sure if I was heading in the correct direction, but trusting the process really is the way to victory, so I kept at it and went to work on these points.

I decided to go for distance field AO, since the normal Lumen AO wasn't giving me any results I liked. I set it up quickly in my master material, which applied it to all my materials, and the difference was extreme.

Just this AO fix by itself did a lot, here's a preview: 

This was a big step forward in my scene. I continued to apply some of the feedback, and this was where I managed to reach. Now it's time for the final piece of the puzzle, lighting.

After polishing up the house, I realized something was still missing. The building itself was done, but the scene just didn't feel right, something about it lacked the atmosphere I wanted, and I honestly didn't know how to fix it. So once again, I went out and asked for feedback.

I want to give a huge thanks to Alex Beddows, my mentor, for all the valuable guidance, and to my friends Miroslav Baev and Chris Pereira for their extra tips. I can't stress enough how important it is to have people who can give you honest feedback and help you grow.

From the feedback, we noticed a few things: my foliage felt off, and the lighting and overall mood didn't quite make sense. I struggled with this part a lot. Every lighting setup I tried just felt wrong. I even experimented with a warmer light direction, but it didn't fit the tone or vibe I was going for.

That's fine, though. I kept studying references and searching for what felt right. After a while, I realized the trees might've been throwing me off, so I decided to delete all my lights and trees and start a completely new lighting pass with a fresh composition. And wow, it finally clicked. It took me a while to get here, but this was exactly what I wanted to achieve. I couldn't stop looking at it myself. I absolutely loved the result.

I used Ultra Dynamic Sky, even though I probably didn't need it since I only used it as a base. I prefer it for the easily adjustable weather systems. You will see later in the post how I use this to my advantage, but for the base scene, I didn't use it too much.

Here's the value shot:

That's it from me, definitely not the most complex piece technically, but man, it was fun to work on. Here's the final Transition video:

Conclusion

This whole project took me around eight weeks to finish, planning included. As long as you're planning well, breaking things down into manageable pieces, it's actually way more doable than it seems from the outside. I advise everyone to go through the boring process first, collecting references, breaking the concept down, and looking into workflows and pipelines before diving in. It will save you a ton of time!

Thanks so much to 80 Level and Emma Collins for featuring me! It means a lot! I'm always looking to improve, and your platform has been an amazing help in my journey. This project was amazing fun to work on, and the whole experience at Sierra Division Academy was great. I learnt way more than I expected to, and most importantly, I gained confidence in my skills and managed to overcome difficulties along the way.

Thanks for reading through my article. Hopefully, there was at least some useful information in there that you guys could learn from! 

Finn Bogaert, Technical Environment Artist

Interview conducted by Emma Collins