Art Spotlight: Feudal Japan - Oiran

Back to overview

About Me

Hello! My name is Wiktoria Kubien, I’m 22 and I’m a Character Artist. I was born in Poland but I’m currently living in Belgium where I recently graduated from Game Graphics Production major in Digital Arts & Entertainment.

When I was a kid I used to play lots of video games with my older brother. Then one day he brought a game that would start it all – Final Fantasy IX. Though at first I yelled at my brother for putting it on, because I wanted to play The Little Mermaid, eventually it became one of my biggest inspirations and goals – I knew that when I grew up I wanted to be a part of video game creation.

Since then I’ve been on my journey to become a Character Artist. I realize I still have a long way to go but my goal is to find out everything there is about creating characters for games: design, human anatomy, animal anatomy, creatures, clothes, hair, weapons. Literally, gotta learn it all!

After finishing my internship and summer job at Larian Studios this summer, I wanted to move on by creating a portfolio piece that would reflect my current skillset and gain me some exposure in my search for a new job. I got linked to the ArtStation Challenge Feudal Japan: The Shogunate and found the perfect subject for my portfolio piece.


For this project I used ZBrush for sculpting, which was the biggest part of this assignment. I also used 3ds Max for hard surface objects, Marvelous Designer for cloth, 3D-Coat for retopology and unwrapping, Substance Painter for texturing, Marmoset Toolbag for rendering and Sketchfab for real-time viewing. Sometimes I also use Photoshop or Topogun.

Choice of the Concept

It was a 7 week challenge, so I started my search for a concept doable in this timeframe. This meant that I was looking for the kind of design that wouldn’t need much interpretation, that wouldn’t require too many extra references or conceptual work in 3D to find out what’s going on. After all, this was going to be my first full real-time character.

But then I found the concept created by Yi that caused me to throw any logic out of the window. The silhouette and shapes of this design inspired me so much, that even though it was the complete opposite of what I was looking for, I knew that this was what I wanted to work on for the next couple of weeks. “After all It’s supposed to be a challenge, right? riiiight?”

The awesome concept by Yi which I ended up picking.

References & Modelling

Since I was in love with the concept, I wanted my 3D version to be as similar to it as possible. My first step in trying to figure out the functionality of all the different pieces of the design was to search for as many references as possible. I ended up gathering them throughout the entire process of creation and not just at the start, because sometimes I would realize I interpreted the concept a bit differently than what was drawn.

Some of the references I gathered.

Once I had the references as a starting point, I moved on to creating a basic blockout in ZBrush. First, I tried to make the general shapes feel similar to the concept. I kept layering parts of cloth on top of each other, so it would look like the character is wearing several different pieces of clothing. I started with the skirt and the “under kimono” and worked my way up by adding the kimono on top.

Basic blockout with the help of Marvelous Designer.

One of the things that surprised me was how long the legs (and arms) on the concept actually were, whilst at first glance they felt quite normal. It was disguised a bit by the amount of clothing she has. Once I started making the blockout, I would have to continuously make her legs longer so that the 3D model would resemble the posture of the concept better.

Where I started vs how long the legs actually got.

When I was satisfied with the general blockout, I decided to pose the character using Transpose Master in ZBrush, to match the sleeves and folds closer to the concept.

As you can see, the arm and thigh area are “wobbly” in the picture above. When posing her, I didn’t pay too much attention to how her limbs looked, as long as the positioning was similar to the concept. I did this because they would be covered by heavy clothing and you wouldn’t see them anyway. I later completely removed the arm and thigh area in low-poly to save the triangle count.

After the posing, I began to work in more detail. I focused on modelling the cats at the bottom of the sleeves. I decided to use a bit of artistic freedom and not follow the design exactly. I tried to model it so that the connection between the cats and the sleeves wouldn’t feel awkward when converted to 3D. This was done by implementing cloud shapes inspired by traditional Japanese cloud paintings.

My version of the sleeve-cats (cat-sleeves?).

Once this was done I tried to figure out what was going on on top of the character’s head. After quite a lot of research into Geisha hairstyles I figured out that the design was instead inspired by the Oiran hairstyles and clothing, so this is what I mainly used as reference. What was missing from Oiran references I borrowed from Geisha references.

At first, I focused on the hairstyle. It is a bit of an exaggerated version of Date Hyogo and it took me several tutorials to try and figure out how it is combed. I decided to experiment and see if I could make it with the use of haircards, with a simple base mesh under it. At first, the result was not that impressive.

Unimpressive first test

But after revising the shader and fixing the positioning of some of the planes it started getting the shape and effect I wanted. The settings of Subsurface Scatter and Anisotropy played a big role in this.

Shader revision and optimising plane positioning for better results

When that was done I used 3ds Max to create a base for the hair ornaments. I spent a while concepting them out in 3D. I tried to find a composition that didn’t make it feel like all the items were clashing with each other, while at the same time felt “just busy enough”.

First tests of hair ornament balance

One of the main challenges I ran into whilst working on the head were the tassels (those hanging thread things that are sometimes used to tie curtains). I wanted to find a quick and easy way to create them myself, since I could not find any tutorial on the internet that explains this well .

So first, I created a cylindrical plane in ZBrush and applied Fibermesh options to it. When I achieved the result I wanted, I reduced the number of splines and exported the curves. Then I imported those into 3ds Max, gave them thickness, lowered the interpolation, then added the editable poly modifier on top and adjusted the vertices so that the curves would have the shape I wanted.

This way I got that “lots of small threads” tassel feeling that I wanted to achieve.

Finished head composition

After I was done with the head, I continued to work with the rest of the ornaments, stitches and details all around the model. I also wanted to make sure that the back of the design ( which was not shown on the concept), would not feel too empty when compared to the front. This required some concepting in 3D, to get a result that looked cool. Without the necessary kimono references and tutorials on how to tie a drum-shaped obi, this would have been, like, 10 times harder and less accurate.

The back of the model

Retopology & Unwrapping

I retopologized and unwrapped the model in 3D-Coat and assembled everything afterwards in 3ds Max. I had to make sure that the polycount was as low as possible, since there was a 100k triangle count limit for the challenge and the hairstyle on its own was already taking up a solid 45k. In the end I managed to get everything just under 99k triangles. This was thanks to the Maya Mesh > Reduce option that, when used on the hairstyle, lowered its triangle count to 34k.

Putting all the UVs together in 3ds Max


I baked the character in Substance Painter with a cage. Most of the bake didn’t cause any issues. Some artifacts showed, but I manually painted them out since baking 4K textures on my laptop takes ages and I just didn’t have that much time left for the challenge.


This was the most fun part of this project – it’s where everything started to come together. For the ornaments I used alpha brushes that I created myself, but the main pattern on the kimono was mostly painted manually. Parts of what I painted, I copy pasted in a mask so that it would be mirrored on the other side and then fixed a little bit by hand. It sped up the process and made the pattern more repeatable and cleaner. My main reference for this was the concept itself and tiger patterns.

I tried to achieve a shiny, “expensive” look of cloth material, something between satin and silk. I did it by working with Anisotropy and Microfiber in Marmoset Toolbag.

Substance Painter viewport

All the textures

Rendering & Lighting

Before this challenge I never paid that much attention to my lighting settings. That’s why it took me a lot of trial and error to get them right and good looking.

A tip that you have probably already heard a lot (but is there for a reason): Never ignore your light setup and learn it well as fast as you can. You will save yourself a headache.

In the end I had one main light coming from the right, then a small fill light on the left and back (so that the umbrella gets some lighting as well) and two rim lights at the back of the model so that the character would be distinguishable from the background and wouldn’t blend too hard.

The skylight I chose was the Desert Road with Low brightness and quite high Child-Light Brightness.

Skylight settings

Final render result:

Sketchfab Setup

At first, I used a Marmoset Viewer for my model. However, ArtStation only allows up to 15 mb Marmoset Viewer files and no matter what changes I made, my model with 4K textures was taking up 87 mb. So I decided to upload my real-time model to Sketchfab instead. Here I could upload my 4K textures without any trouble and, if necessary, the website will optimize the textures on its own, while leaving the HD version as an option!

By the time I decided to upload my model, I was lucky enough to find out that Sketchfab just added Anisotropy in the PBR settings! ( Go Sketchfab Team ! (ノ´ヮ´)ノ*:・゚✧ )

It took me a while to find a background that would fit the character. In the end I found that a slightly foggy, custom background worked best. For the environment I used Urban Exploring Interior and for the lighting, three-point lighting with some changes.

Final thoughts

This project was an amazing challenge that I will always remember fondly and I’m very proud of what I achieved. The time limit, amount of detail necessary for the model, the feedback I received from friends and fellow artists, and the fact that I got a very strong cold for over a week and still refused to give up on the deadline all give me a feeling of accomplishment. I knew the concept would be an overkill for the time I had and I still finished it.

There are things that can still be improved (e.g., the bake of the skirt since the low poly of it was too low poly, lighting setup, background choice, final touches) but this project proved to me that the most important thing is to just be stubborn enough to achieve the result you want.

Oh, and it also showed me how much I love creating characters. But that’s something an 8-year old me already knew and everyday strives for.

I would love to thank the Sketchfab team for giving me the opportunity to share my 3D work and process 😊

If you would like to see more of my work, check out my ArtStation!


About the author

Wiktoria Kubien

Character Artist

Leave a Reply

Your email address will not be published. Required fields are marked *

Related articles