Mastering Wrinkly Skin Texture by Making an Old Medieval Dwarf in 3D

Introduction

Hi, I’m Nikita. I got into 3D about 11 years ago. I had this strong need to create something, and at first I actually wanted to start my own animation studio because I’ve loved animation and movies for as long as I can remember.

But I quickly realized that without any art skills, it was extremely difficult to make something good, even if you managed to find people to collaborate with. So I decided to learn 3D myself. Thankfully, by that time, there was already a lot of information available online. I gathered as much as I could, watched tutorials, and basically started recreating everything I saw in those videos step by step.

After about eight months of learning 3D, I started building my first portfolio, mostly with vehicles and weapons. Four months later – about a year after I first opened a 3D program – I got my first job in the industry. 

I worked on many different types of assets before eventually moving fully into character art. Over time, characters simply became the part of 3D that interested me the most – there is so much you can explore through sci-fi, fantasy, and stylization.

Over the years, I’ve contributed to more than 13 projects, including Call of Duty, Diablo IV, Warhammer 40000, HELLDIVERS, Star Citizen, The Elder Scrolls Online, and Escape from Tarkov, among others.

Dwarf

Like most of my personal projects, Dwarf started as a study. Usually, I choose something I want to improve or something new I want to try. Or explore something different from my studio work: for example, when I got the idea to work on Dwarf, I was working on Star Citizen, so I wanted to diversify my daily tasks with fantasy personal projects.

One of the main reasons I chose this design was that I wanted to work on an older character. I find old people especially challenging to sculpt convincingly. In my personal difficulty ranking, a young man is probably the easiest to sculpt, then comes a young woman, then various creatures, and an old man is somewhere near the final boss level of difficulty.

I gathered references mostly in 3 categories: old men, medieval and Viking clothing, and other artists' work. But basically, I have a big dump where all the references are stored. I put every interesting thing I found in there, plus there's another file with anatomy references.

For the most part, I stayed very close to Even Amundsen’s original design. Most of my changes were in the areas that weren’t clearly visible in the concept, so I had to figure out what could logically be there. The shirt under the cloak is one example. I looked at medieval clothing references and tried to build something that would naturally connect to the parts already visible in the drawing. I changed the boots a little as well. I wanted them to have a bit more complexity and, honestly, it was also a good excuse to practice making shoes.

At first, I wanted to create a customizable character with different variations of clothing, but later I decided to lower my ambitions and simplify it.

The helmet needed a bit more thought because from the beginning I wanted it to work both as something he could carry and something he could actually wear. I ended up looking at Roman helmets, since the shapes you could see in the concept reminded me of them and gave me a good starting point for the parts we couldn’t see.

And then there was the sword. I made a few changes there too, mostly because I wanted it to feel like a functional weapon instead of reproducing the silhouette from the concept.

Modeling

I usually start either in 3ds Max or ZBrush, using simple primitives and focusing first on matching the proportions and overall shapes from the concept. I always finish the high poly in ZBrush, and after that, I move on to the low poly, which I usually create using TopoGun and 3ds Max. 

Whenever possible, I try to use clean topology for the final high-poly assembly as well. In the 3ds Max image, you can see that the mesh already has fairly clean and readable topology. At this stage, I can also unwrap it, which gives me more control when adding fine details, patterns, or surface noise later in ZBrush. I can then reuse the same geometry as a starting point for the low poly in some cases. 

My usual approach is to bring this mesh into ZBrush, create Auto Groups, add thickness with ZModeler, and then project the details from the simulated or previously sculpted geometry onto it. This gives me a clean final mesh with proper topology and UVs, so I can add smaller details and UV-based patterns or noise without unnecessary distortion.

Once the high poly is ready, I retopologize it in TopoGun and 3ds Max, then I make the UVs using the good old UVLayout and bake everything in Marmoset Toolbag. From there, I move into Substance 3D Painter for texturing, Blender for rigging, and finally bring everything into Unreal Engine 5. 

I sculpt the head from the base sphere we have in ZBrush, with decimation and symmetry on first. Then I retopologize it in TopoGun or use the existing base head topology I already have and modify it. Then I create asymmetry by working with big and medium-sized shapes, trying to achieve the right look, and finally add small details.

What I’m trying to achieve is a stylized look that's still realistic, so basically I’m trying to find my own style.

For the clothing, I simulated the cloak in Marvelous Designer, while most of the other clothing pieces were modeled and sculpted directly in ZBrush.

Also, while sculpting his head and adding details, I tried to check the result with SSS in Marmoset Toolbag, so I decimated and loaded it in to check everything out.

The raven was made in probably the simplest way I could think of at the time. I didn’t want it to look too realistic, but I also didn’t want to push it too far into stylization, so I created a simple body and built the feather cards on top.

For the feathers, I used the photos I found online as a base, painted over them, and reused the same material setup I had already created for the dwarf’s hair and fur. That also saved me some time and helped keep the character consistent.

The only part where I tried to save time was the raven's rigging. Since the raven was basically a supporting prop for the dwarf rather than the main focus of the project, I kept things simple and built him around the final pose instead of overcomplicating the setup.

Maybe a good way to save time is how I create quad topology for an HP base before moving on to finalizing it so later on I could use it for my low poly.

I found in one of 80 Level's articles that artists use Paragon hair materials, so I gave it a shot and used it for hair and feathers. It also saves me a lot of time creating my own hair material. Even though Paragon hair is an old material, it has a lot of features you can tweak.

Retopology

As I mentioned earlier, I do all of my retopology in TopoGun and 3ds Max and then unwrap everything using headus UVLayout, which is the old-school way, I know.

For this character, I split everything into five UV sets: one for the head, one for the body, two for the clothing, and one shared set for the weapons and helmet. At first, I wanted one more UV set, but decided to reduce their number and simplify my project a bit.

And I also deleted backfaces on most of my clothes because you don’t even see the difference and it makes life a bit easier. 

I textured the character in Substance 3D Painter, with occasional quick checks in Marmoset Toolbag. For me, Marmoset is much faster when I just want to see how the materials react to light without loading everything into Unreal Engine and setting up the whole scene again.

In Unreal Engine, I used a master material that I had originally created for my previous project, Mogg Tallow, and then adjusted it for Dwarf. A lot of what I know about materials comes from watching and rewatching good tutorials online, including videos by artists like Jared Chavez.

These days, I also use Google's AI tools or ChatGPT for quick technical questions. For example, if I need to understand which parameters are worth trying for skin SSS in Unreal Engine, or why a material is behaving in a certain way, it can give me a quick direction for investigation. A few years ago, you could easily spend hours going through forum threads trying to find someone with exactly the same problem, so this part of the workflow has become much faster.

And also, you can ask ChatGPT to write a script for Maya, for example. I already tried it, and it worked perfectly.

One of the biggest improvements to the look actually came from something very simple – switching to the Path Tracer in Unreal Engine. The way light interacts with the skin surface feels much more natural compared to the standard lighting mode with ray tracing, and it makes the face feel much more believable in the final render.

For me, getting the right look is not only about making good textures. You can have perfectly decent textures in Substance 3D Painter and still destroy the result with bad lighting. I’m definitely not a technical lighting expert when it comes to optimization or everything happening under the hood of an engine, but I can usually tell when something looks better or worse and keep adjusting it until it feels right.

And probably the biggest cheat code for me as an artist is simply taking a real break from the project. I don’t mean a few hours or one night – sometimes I put a project away for several days or even a week and work on something else.

When you look at the same character every day, at some point, you just stop seeing some of the problems. When you come back with fresh eyes, those things often become immediately obvious. I used this approach both during the high-poly stage and while texturing the dwarf, and almost every time I came back, I found something I wanted to improve.

Lighting & Rendering

One useful piece of advice I remembered from my former colleague Anatoly was to start with a Post Process Volume and properly set up the exposure, especially the minimum and maximum exposure values. That gave me a much more stable base to work from. It's the first thing I add to the scene every time.

Everything else was mostly experimentation. I really enjoy figuring things out by myself. Sometimes it probably takes me longer than following an exact setup from someone else, but for me, a big part of a personal project is the process itself. There is always this moment when you finally see your character working in the engine and think, “Okay, now it actually looks alive.” I really enjoy that part.

I also wanted to add that I prefer to use 75-85 mm lenses in my scenes, as those give me the right look I'm aiming for. It's not too flat, like an orthographic projection, and not too distorted like a 35 mm lens. And it doesn’t create any extra space, which is good because we are focusing on the character and not on anything else. And what's funny, I don't have any extra video memory for the environment to bring into this scene anyway.

For the lighting, I used one main fill light in front of the character, a top light around the head to create some shadows, several supporting side lights to bring out the volume, and a few backlights to separate the silhouette from the background. I also added smaller lights for specific props or whenever I wanted to create a nice highlight on a particular material or shape. And I don’t try to optimize those – I always try to achieve the right look, so I don’t mind adding an extra source of light, especially since his fur cloak covers all of the light from the side and from behind.

The final images are rendered with the Path Tracer in Unreal Engine. There is no additional compositing in Nuke and no external color grading. What you see in the final renders is essentially the raw in-engine result, with the look achieved mostly through lighting and material setup.

For the videos, I only assembled the rendered sequences afterward, without changing the actual look of the character.

Final Steps

I actually started this project a long time ago, but for most of that time, Dwarf was just sitting in my closet, waiting for his moment to shine. I was working full-time at a studio and also part-time at a 3D school, so by the end of the day, I usually didn’t have much energy left for personal work.

A few years ago, after finishing the high poly, I put the project on hold for almost a year. I returned to it properly only this year, after I was laid off and suddenly had more time for personal work. That’s when I decided I finally wanted to finish him.

The part I probably enjoyed the most was texturing. You start looking through photos of older people, trying to understand what actually makes them look old: the skin variation, discoloration, roughness, small spots, and all those little imperfections. I really enjoy searching for those details and figuring out how to translate them into textures.

And, of course, I also love seeing everything come together in the engine. I really like that moment when you finally load everything into Unreal Engine, and it starts to work as a complete character. 

The biggest challenge for me was definitely the hair and fur. In my daily studio work, I mostly create cosmetics, armor, clothing, and other character assets, but not the hair. Usually, there is always one or two people on the team who are much better at grooming, so those tasks naturally go to them. With this project, I wanted to force myself to learn it instead of avoiding it.

I think I’m getting better at it, although I’m still not at the level I’d like to be. Maybe next time I’ll just collaborate with a really good grooming artist, and we’ll make something amazing together. But at least now I have a much better understanding of how hair and fur should be built, and I also feel like I improved my skin texturing quite a bit during this project.

And I learned that Unreal Engine has a lot more buttons worth pushing if you want to make things look better.

As for tutorials, I don’t really have one single tutorial that taught me everything. I usually collect information from many different artists and videos and combine whatever is useful for the problem I’m trying to solve. Jared Chavez’s videos, which I mentioned earlier, were definitely helpful for materials and skin.

Conclusion

I don’t think there is one cool tutorial that will suddenly make you better. For me, it’s mostly about practicing and experimenting. With every new project, I try to do at least something I haven’t done before. Sometimes it’s a small detail, sometimes I go deeper into one part of the workflow, and sometimes I just try a completely different approach.

I’m always fascinated by other artists’ work. When I compare what I see now with what people were doing even ten years ago, the difference is crazy. Every day you open ArtStation and see another amazing character or environment, and you think, “Okay, how did they even do that?” The level just keeps going up.

Sometimes it feels like we are one huge creative monster that keeps getting stronger every year. Of course, it also makes it harder for new artists to get on board, because the overall level is so high now. But the tools are getting better, too, and there is much more information out there to help you learn. 

So yeah, just go and make something. 

Nikita Popp, Character Artist

Interview conducted by Amber Rutherford