Winged Hussars Arrived in 3D: Modeling & Texturing a Medieval Polish Warrior

Introduction

I’ve been drawing and painting since I can remember but my interest in games started with such games as Diablo and Heroes of Might and Magic when I was a kid. I got introduced to Maya during one of my internships as a programmer in a small game studio and was immediately hooked. It was the first time that I felt like I was “drawing” in 3D, being able to work on a character and see it from different angles to make it look more alive and real. That’s how my journey of becoming a character artist began. I worked on some Steam games, VR games, and mobile games. I currently work as a Senior Staff Character Artist at Sucker Punch Productions. The most recent game I worked on is Ghost of Tsushima.

The first creature model I made using Box Modeling in Maya vs. another dragon model I made in late 2017:

Project Idea

Initially, my goal with this project was to experiment in Unreal Engine 5 with various look dev materials and lighting scenarios. To this end, I looked for a character that would have a bunch of different materials on it and stumbled upon the Polish hussars. They have metal, fabric, leather, feathers, fur, and jewels on them which perfectly fit my goal of experimenting with materials. They also hold a distinct visual aesthetic that looks like a blend of Eastern and Western armor designs giving them an original and appealing look. What’s more, they ride horses which I saw as another opportunity to expand my process and play around with anisotropy in UE5, which proved to be a bit tricky at first but was an enriching experience.

Once I decided what kind of character I was going to make, the next step was creating the concepts. For most of my personal projects, I try to create original designs or alter the designs and references I find since I enjoy the creative process it allows me to go through. I didn’t want to model a completely historically accurate hussar, but rather fantasy versions of it with a consistent and unique design twist to make it my own.

I gathered plenty of reference images, and separated them based on their potential use like “armor”, “horse”, “weapons”, “textures and materials”, “lighting and rendering”, and “pose”. The more reference images I looked at, the more my project expanded and took shape in my mind. I decided to make “dye variations” for my hussar as you would have in some games as cosmetic options. For each dye variant, I then designed modular pieces like unique helmets to give each dye variant a sort of theme and make them look distinct without too much extra work. 

Creating Multiple Versions of the Hussar

When I tackled the concepts, I knew I wanted to create a dark fantasy look, so I tried to keep the colors relatively muted and dark even though there were plenty of colors across the different dye variants.

Designing the Base Hussar was the most straightforward since I didn't bring any major changes to the historical hussar armor design. All I did was introduce some asymmetry, add some decorative elements, change some proportions to avoid uniform and boxy shapes that the historical armors have, and add a bit more color. It was still a crucial step in the design process of the overall project since the base hussar design set the design language and the mood of the entire project.

I then decided on different colors of the dye variants I wanted to make. They needed to look as distinct and unique as possible. The base colors are primarily silver and red. I also chose black, white, gold, and bronze for the other variants. 

For the Black Hussar, I wanted an undead or a classic black knight theme, so I gave him the skull mask and dark colors, as well as a brutal axe with a long spike on one side. I liked the Spanish conquistador type of helmet for this particular design because it looked interesting with the mask. It is not a common design element for a fantasy black night, so it felt unique. This element also allowed for a cool cast shadow to cover his eyes and gave him a mysterious look that fit the theme.

For the White Hussar, my goal was to create a dye variant that would look like a white knight in shining armor, an antithesis to the Black Hussar. His modular pieces of hanging chains and decorations as well as finely embroidered fabrics and long flowy feathered, jeweled helmet make him look regal. I gave him a large mace that’s supposed to resemble a king’s scepter, in line with his regal design language. 

The Gold Hussar has more of an oriental look, with rich gold-colored armor, dark purple fabrics (which were rare in medieval times), an exotic and expensive tiger pelt, and intricate helmet design. I thought this was a good way of distinguishing him from the rest of the lineup, giving the Gold Hussar his unique aesthetic. An elegant-looking spear seemed the right choice to go with his elegant design language.

For the Bronze Hussar, I went with a more primal, barbarian-like design since bronze is the most primitive metal used on armor out of all the dye variants. Bull horns, an aggressive-looking mohawk, a rough wolf pelt, and heavy chainmail covering the eyes give the face an animalistic aesthetic which makes it hard to recognize as a person to be reasoned with but more of a beast-man look. His armor is also the least well-looked after compared to the others, having the most amount of patina, again to support the overall aesthetic. A war hammer, a brutal blunt impact weapon, seemed appropriate for this character.

Modeling

I started the modeling process by creating the horse first. I had already modeled a horse skull previously so I utilized that as an anatomical blueprint when making the head. I used a lot of different references for the head and the body to make sure that the proportions and the musculature were right. I wanted this particular horse to be a war horse and make it subtly stylized to fit the style of the hussars and the fantasy element of the project while also looking grounded. To this end, I adjusted the anatomy in the low poly by making the neck and the legs a bit thicker and giving the horse a more broad-shouldered and powerful look, fitting a war horse able to carry an armored knight at full speed. 

I started from a low subdivision level, pulling and pushing shapes using the Move brush, and then worked my way up to secondary and tertiary shapes with a lot of Clay Tubes and Smooth brushes, finishing with the surface details like the veins and wrinkles. 

I then moved on to modeling the hussars. I usually start with a very rough blockout of simple shapes to be able to establish the overall silhouette, weight distribution, and shape hierarchy. This is where I tend to modify the concept for the first time if necessary, given that the proportions of the base model I use may not always match the proportions of the 2D concept which was the case for this one. 

Another area I deviated from – or rather added to – the concept was with the tertiary shapes and details. A lack of fine details and rough lines may work on a 2D concept but a 3D model, especially if it’s going to be seen close-up, needs to have fine details to increase the depth of the model and reward the close-up shots without creating unnecessary noise.

Once I established the overall proportions and silhouette, I jumped into Marvelous Designer to create the clothing elements. I like to use Marvelous Designer sparingly, enough to obtain a decent shape for the piece of clothing I am making. Further, I prefer sculpting it in ZBrush. That said, ensuring I have interesting folds and shapes in Marvelous Designer is key since it is the base on which I build up all my sculpted cloth details in ZBrush.

For the armor, I used primarily ZModeler. After I was happy with how the blockout turned out, I used the Standard and Dam Standard brushes combined with the Lazy Mouse option to create the intricate floral details on the helmet and crest.

I left the high poly model simple in terms of surface detail, damage, and wear and tear since I didn’t want these characters to look too beaten up. Most of the real-life references of armor that I found were “worn” but not necessarily damaged, so I decided to create that effect with textures and materials rather than sculpting them in.

Also, making unique scratches and dents on the armor high poly model would mean that each dye variant would have the same damage which is undesired. Making the surface wear with textures meant that I could adjust it for each dye variant, feeding into the uniqueness of each one. 

I created the chainmail in ZBrush using Array Mesh to set up the correct repeating pattern and then used NanoMesh to create the chainmail chunks I wanted. 

For the feathers, I tried a couple of different methods but FiberMesh gave me the most realistic and flexible choice so I decided to go with that. I created a bunch of variations for the feathers to avoid repetition on the model as much as I could.  

Retopology & Unwrapping

I tend to limit myself as little as possible when I do personal projects. I was liberal with the low poly and the character wasn’t going to be rigged and animated. In that sense, personal projects are relaxing and freeing, meaning that I can push the boundaries and don’t need to worry about some of the technical limitations that I need to work with daily at my regular job. I made sure the high deformation areas had enough edge loops and there wasn’t a lot of faceting for the close-up shots and that was about it when it came to the low poly. The low poly models that I ended up with for both the horse and the hussar have a reasonable density for current-gen games, easily feasible for next-gen games. I used Maya to do the retopology as well as unwrapping the UVs. I used UDIMs for both the horse and the character to maximize the texture resolution.  

I tried utilizing the second set of UVs (UV2) for the armor pieces to create the etched floral details. This way, I would be able to change the pattern per dye variant if I wanted to. It would also allow me to have a much better resolution compared to having it baked on from the high poly model or even adding it in Substance 3D Painter while texturing.

I created a seamless tiling floral etching pattern for the armor and applied it to the armor material in Maya for testing after setting up my UV2s. I got decent results but thought that it would look too sharp and high-resolution compared to the rest of the textures, looking like a decal and not etched into the armor. In the end, that was the case, so I opted to incorporate the etching during the texturing phase. It was still a fun and beneficial experiment and a learning opportunity.

I used UV1 and UV2 for the armor plates and UV2 for the floral etching decorations.

For the fur pelt on the shoulder, I used FiberMesh in ZBrush by adjusting the settings to make each fiber into hair cards instead of hand-placing them.

Texturing

I did the textures using mostly Substance 3D Painter. I used other programs such as Substance 3D Sampler, Photoshop, and Mari. I created the materials from scratch based on the visual aesthetic and style I set up in the concept stage. I textured the characters, adding enough contrast at the focal points and consistent material definition. Having weathered armor and fine details for the close-up shots meant that each material required multiple layers. I achieved that by initially creating the base material with the correct PBR values and then adding wear and tear on top, using the Curvature and AO Maps. 
 
Once I established my desired look for each material on the Base Hussar, I turned each material into a Smart Material. Further, I utilized them as starting points for the dye variants. Then, I modified and added their unique materials, for example, slightly iridescent white plaster on the White Hussar’s armor or the Bronze Hussar variant with the teal-colored patina on certain parts of the armor.

For the high-definition embroideries, I used Substance 3D Sampler. First, I created the pattern in Photoshop in grayscale and imported it into Sampler, where the magic happened, and, with a few clicks, it output a threaded embroidery version of my pattern. I exported Normal, AO, Alpha, and Roughness Maps of the embroidery from Sampler. Then, I used them in Painter, employing separate layers to lay them on top of the fabric to create the embroidery effect.

For the face, I re-used another personal project I had made where I utilized Albedo and Displacement Maps of face scans from Texturing XYZ and projected the detail to the head using Mari. I baked a Cavity Map using XNormal, gave it a reddish hue, and then applied it as a subtle layer on top of the existing Albedo in Substance 3D Painter. A lot of the skin quality comes from the rendering side using the Subsurface Scattering function in Unreal Engine 5.

When texturing the horses, I made the hair for the horse body in XGen and then exported a Height Map. Then, I used it in Substance 3D Painter as an overlay to add height and normal information as well as roughness and AO to add depth to the body texture. 

Once the base version of the horse looked well, I started creating the rest of the 12 color variations, which was fun to do. I also made 3 mane and tail variations: short, long, and braided.

To decide which color horse would match with which hussar, I tried to use the horse colors as secondary to the armor colors so that the main focus would still be on the hussars. For the saddle and gear color, I mirrored what I did for each hussar variation to provide consistency. 

Rendering

I did the LookDev and took all renders in Unreal Engine 5, which was one of the major reasons why I started this project in the first place. I was pleasantly surprised by how good the Unreal Engine renders turned out in terms of image quality, color, and contrast. I didn’t make any additional adjustments to the renders, such as level adjustments or color correction. The only renders that I did some post-work on are the “mood” shots, both cool and warm versions. I wanted to give them a dreamy, movie-like quality, and using Photoshop to do that gave me more options and was faster.

One of the biggest challenges I faced during look-dev was technical issues that appeared when setting up my materials in UE5. I specifically had a hard time setting up the horse anisotropy since I kept getting egregious UV seams.

After some long nights and with the help of a few friends (shout out to Omar Aweidah), I managed to fix the seams and move on from the horse LookDev. The issue turned out to be a faulty Flow Map I made in Mari. Even the tiniest flow direction mismatch between seams created atrociously visible seam lines in Unreal because of the use of anisotropy.

Once I got a hold of the material setup, node graphs, and lighting setup in UE5, the look-dev of the hussars was straightforward. I had to do a lot of back and forth between UE5 and Painter since the colors, contrast, and definition of materials I viewed during the texturing phase in Substance 3D Painter ended up dramatically different than what I was viewing in UE5. I did plenty of tweaking to get the right material definitions to show up in UE5.
 
The fur pelt was an interesting challenge. I had to make a leopard pelt with the sophisticated spotty pattern that the actual animal has. Using regular cards and textures to create every spot was out of the question, so my options were to use splines, shell stacking, or using alpha cards with the base color being grabbed from the UV2. I went with the third option since I wanted to make these next-gen game characters and thought it would look better than shell stacking at close-ups.

For the warm and cool mood renders, I took the raw renders into Photoshop, and mainly, what I did was blur out the hot spots, add some motion blur, and a vignette to get a movie-like look for those images. My goal was to show these characters in a pseudo-environment and dramatic lighting to simulate a game or cinematic.

To light up my UE scenes, I used many different lights and settings based on the character I was rendering. I used an HDRI sky in low intensity as a base to get a realistic reflection on metal armor. Then, I utilized a combination of directional, spot, and point lights to light the character, as well as the ramp in the background to get the final vignetted look.

For some of the shots, I used special lights to illuminate the head area or just the weapon blade to get a nice highlight in those areas.

Having two rim lights, often one slightly cool and one slightly warm worked for me on these characters. I also had a backlight that worked especially well with the horse renders since they made the thin whiskers on the horse’s head nice and visible.

Project Overview

It’s hard to put an actual time stamp on this project because I didn’t work on it as one continuous chunk. I made the concepts back in 2021 and didn’t start modeling until a few months later. Once I finished the horse model, I stopped for a few months, continued for another year, gave another several months break, did some other projects, etc. 

If I were to work on one of the hussars full time 40 hours a week from the start of modeling to the end of look-dev, it would probably take me about 7-8 weeks to create a detailed character like that. Posing would probably take another few days. 

Challenges

I can summarize the biggest challenges in two categories: the technical challenges of learning new software and the mental challenge of undertaking such a big project. 

In terms of the technical challenges I faced, the performance of my PC was one of them at the early stages of texturing. It even led me to upgrade my PC to be able to continue working on this project. Working on a character having multiple 4K texture sets requires decent hardware. Unreal Engine, although it is artist-friendly, posed some challenges as well, mainly concerning the rendering of shadows and the horse anisotropy. 

I initially started this project to make only one character, the regular Winged Hussar, and taught myself UE5 in the process. However, when I embarked on the creative journey of brainstorming and making concepts, I got a bit carried away and ended up making the horse, horse color variants, hussar modular pieces, and dye variations. Since the project got way larger than I initially anticipated, it was also a challenge to keep myself interested, avoid burnout, and keep the quality bar up as time progressed. With long projects, there’s usually a tendency to rush, especially by the end, and drop the overall quality, losing sight of what was visioned at the beginning. I took my time and made sure I spent ample time on renders and posing by the end of the project. When I felt burned out now and then, I jumped on other projects or worked on other aspects of the same project. Weapons were all pretty much done that way in between various tasks.

Conclusion

Part of my initial plan for the final renders was to create a few scenes where all of the hussars would be charging on their horses together as a unit, as in a battlefield. I didn’t get to do that because other things came up, but I may be able to get back to it in the future.

If I ever do it, I expect I will have some environment and effects in there as well, like the hussars coming out of the fog, steam coming out of the horses’ nostrils, mud and pieces of grass flying away as the horse hooves beat the ground, etc. Here are a few inspirations I gathered as references for those specific shots. 

My advice for any artist beginning their journey would be to focus on their own work and not get too distracted by what other artists are doing. It’s good to share knowledge and get inspired by others but getting too attached to other people’s styles or interests would make you lose your uniqueness and cause frustration. Originality is getting harder to come by due to the immense interaction we have with each other over the internet.

Getting attention on social media is also shaping the future of art because many young artists are doing what they think others will like instead of what their actual passion is. I think it is a mistake. That said, with the amount of resources we have available now, I see a lot of excellent emerging artists at the beginning of their careers. It is promising for future progression. I wish luck to whoever is passionate about character art and just starting their journey.

Emre Ekmekçi, Senior Staff Character Artist

Interview conducted by Gloria Levine