Bedouin Elf: Working on Realistic Facial Details and Skin

Tomas Sackmann talked in detail about his ultra-realistic cinematic portrait made primarily in ZBrush, Substance Painter, Mari, and V-Ray.

Introduction

Hi, my name is Tomas, I am a 3D artist and Industrial Designer from Argentina. Currently, I am working mainly in Dubai, UAE, as a Senior 3D designer/Creative Director in the Experiential Marketing, Interiors, and Product Design industries and doing some 3D Character Art as freelance work for other countries.

Studying 3D and Character Art

I started working in 3D at university while studying industrial design but mainly using CAD software. After that, I started exploring 3ds Max and V-Ray for interior architecture and events as my main design tool.

After some years of working with it, I started to feel that I needed to become more efficient with the tool to be a faster and better designer. This is why I decided to go back to school.

2 years ago I started working at a company called Lightblue where I worked with Dimitar Dimitrov who was studying at Think Thank Training Center. After a few months of seeing his work, dedication, and how much he improved his skills in a really short time (without sleeping much I have to say), I decided that it was time to join Think Tank.

As mentioned previously, before joining I was planning to enroll just to improve my design skill for the industries I was working in at the moment but Think Tank, I realized that the world of 3D was much wider and fascinating than what I knew until then.

Since I am an Industrial Designer, during all these years almost every project I designed in 3D was hard-surface, products, buildings, spaces, and I didn't know anything about sculpting. This is why I decided that I was going to focus my studies more on Character Art to learn those new skills. And again, as it happened when I joined Think Tank, I realized that I loved making characters.

1 of 3

Bedouin Elf: Reference and Tools

The first thing that captured me in the reference was the expression of that man, the smile, and the look in his eyes. It made me feel happy and connected with him. The composition, lighting, and the depth of field also make you focus directly on his eyes and smile. The clothing tells us a bit about the story of this person and location. Basically, it is an amazing photograph (by Robin Moore).

I took a long time observing some of the key details to get them right like the wrinkles on the face, the hair structure, the dryness of the skin, the reflection in the eye, and the materials and imperfections of the clothing.

The main tools I used were ZBrush, Maya, V-Ray, XGen, Substance Painter, Mari, Photoshop, and Nuke. Each of those tools is strong at different stages of the process. I decided to use them because they all are widely used in the industry.

Sculpting the Face

All the sculpting was done in ZBrush, starting from a sphere and slowly blocking out the head and body in the A pose. During this phase it was really important to look at a lot of references, reference is key. I created a PureRef file with references of skulls, people from African-Asian countries, wrinkles, and smiles.

At first, as I didn't have a reference of the character with a neutral face I decided to start the sculpt based on a picture of someone similar to him but younger. You can see in the image below how the process started.

After getting the proportions more or less right (honestly, they were not great at that moment), I moved the sculpt to Maya to do the retopology manually.

When I had the new topology in place it was time to pose the character and start working at the expression. This was the most difficult and long part of the project and for a really long time, it was looking very bad.

Regarding my knowledge of anatomy, honestly, I didn't really have any at that moment. This project taught me some basics of facial anatomy and also made me realize how much more I need to learn and how fascinating it is.

In the gif below you can see how the sculpt changed week by week:

Regarding the brushes, I used only a few of them, nothing complex. Mainly Clay Build Up, a lot of it, then Standard, Move Topological, and Dam Standard.

For the skin texture itself, first I tried using a Texturing XYZ map provided by the school, but it really didn't match the look I was looking for so I ended sculpting the skin pores by hand using a mix of some skin alphas from the internet and some individual pores alphas I sculpted myself. For the hands, I used a hand displacement from Texturing XYZ and kept sculpting on top of it to create some breakup. After having the pores in position, I gave pore directionality with the Standard brush, in color spray and with the alpha 60. The main wrinkles were all done by hand mixing Dam Standard and Standard brushes.

At the end, I used the Elastic brush to break the evenness of the skin. The Elastic brush allows you to add volume without removing the superficial details.

Regarding the geometry, I tried to keep the topology clean and even but there were some points of weakness I should have addressed in a different way.

As you can see in the image, the topology needs a lot of improvement to be fully ready for animation. Since the plan was only to make still images and I ran out of time at the end of the project, I couldn't update the topology. But if I were to use this character in film production, I would need to adjust the topology of the face to make it follow the wrinkles. This was one of the key things that I learned in this project.

There was also a problem with Displacement maps. In the image below, you can see the space between the main wrinkles next to the eyes and on the forehead.

This problem was mainly because the topology was not following the wrinkles correctly.

Skin Texturing

My texturing process was a mix between Substance Painter and Mari.

The texturing started in SP by hand painting the face and focusing on defining the color zones and specific details of the character, like the freckles, moles, and birthmarks.

Also in SP, I baked Curvature, Ambient Occlusion, and Thickness maps and saved them externally.

With all this in place, I jumped into Mari and continued refining the Diffuse using photo projections of real humans to add an extra level of realism. References of bruises, spots, and old skin helped a lot. At this stage, it is important to use a high pass filter on the images to remove any lighting information. The Diffuse texture needs to look somewhat flat to behave correctly with the light in the render. As this project was made with a film pipeline in mind, I didn't bake any shadows information in the textures.

From that final Diffuse I started working on the rest of the maps like Displacement, Bump, Spec, Roughness, SSS.

The most prominent uneven spots were added in two ways: 1) either with photo projections or 2) by hand and with different layers of tileable textures and masks to generate the small surface imperfections.

I did the same with the bruises and scars as you can see in the image below:

Honestly, I finished the texturing process only when I reached the deadline. There is always something that can still be improved and changed in a character like this, and if I had two more weeks or a month, I could have given more realism to the piece.

Facial Details

I used a lot of references and studied facial anatomy for this step. Understanding where the bone landmarks, the muscles, and especially the fat pads are located and how they behave with aging is important. It helped me to get a more realistic face.

I took the eyes that I modeled for my previous project and modified them to make them more organic. In my intermediate project, the eye had a perfect sphere with only the iris extruding out. In reality, the eye shape is not so even; some deformations on it helped make it look more natural.

For the eyes shader, I used a workflow learned from Magnus Skagerlund, my supervisor in the intermediate term. It involved using VrayFresnel to adjust the level of reflection depending on the angle of view. All the maps for the eye were done in Substance with a mix of photo projections and some vein brushes from Substance.

For the lips, I used photo projections of a real lip as a base, but since I was not getting the result I needed, I painted the differences in tones and values by hand. In the shader itself, I set a lower level of roughness and a higher one for specular to make the lips look less dry than the skin. Honestly, I think the lips were one of the weakest points of this piece and they need some improvement.

The hair was done with XGen in Maya. I used splines to control the main shapes and modifiers to break down the curls and get a more organic look. I used the V-Ray Hair Next material.

Clothes

Making clothes was tedious. Both articles of clothing are just two rectangular pieces of fabric with sewn edges that I placed on the character manually, freezing some parts of the fabric with the Pin (box) tool and simulating the rest until I got it in the right position. I redid this process around five times because the results were not good.

Since the two pieces of clothing were completely different materials, in Marvelous I had to apply two different types of fabric and with different thickness to make them behave as naturally as possible.

For the shemagh, I used the Cotton_Sateen present and a thickness of 0.27 as it is a really thin and light fabric.

For the body cloth, I used Cotton_T with a thickness of 0.41 as it needed to be harder and thicker.

The orange dots spread around the fabrics are the points that held the fabric in the desired position. As there was some loose fabric, it was the best way I found to make the clothes stay where I needed them and prevent them from falling.

Once I had the fabrics more or less in place, I took them into Maya, retopologized them on the flat version, and transferred attributes using the UVs to the Marvelous mesh in position.

With the retopologized mesh, it was time to go to ZBrush and sculpt the final wrinkles and details.

As you can see, the mesh is quite dense because this character was planned for a film pipeline and I needed to capture as many details as possible from the cloth simulation onto the mesh itself.

Lighting and Shaders

The scene was lit in a pretty simple way, with an HDRI from HDRI Haven and a 3-point lighting setup, plus an extra light to highlight the reflection in the eye.

To adjust the lighting I prepared a scene with the model without any hair and Displacement maps applied, to speed up the render time. With this in place, using V-Ray IPR, I first started placing the individual lights one by one. First, the HDRI to get the main tone of the image and then the rest.

The 3 lights in the front were warm lights with 5100-5800K each and the rim light at the back was a cold light of 8000K. This difference in temperature is difficult to perceive in the final render but it helped me get better contrast and separate the character from the background.

Also, an important thing in the lightning process was the addition of the walls that you can see in the background in this image below. I put this to reduce the effect of the HDRI on the character.

This was a back and forth process because depending on how the shader skin was looking I had to adjust the tone and intensity of the lights accordingly. In the image below, you can see part of the progress of the model, materials, and lighting:

All the renders were done in V-Ray, and I used the V-Ray Al Surface shader for the skin, and the V-Ray 2 Sided material for the clothing.

For the skin shader, I got some great advice from my instructor, Raffael Frank. Basically, the core principle was to start simple and keep polishing and creating more complex material step by step.

I started doing some test renders with the V-Ray AI Surface material, with no maps applied except for the Displacement, which allowed me to see if the sculpt was behaving as expected after applying the Subsurface. It’s really important to understand that the SSS will eat a lot of the details of your sculpt, so it is key to start rendering early and go back and forth, modifying the sculpt and checking how it looks with the material on.

When I was more or less happy with the raw material, the next step was to apply, one by one, all the maps and check how each of them was affecting the material. I used Diffuse, Roughness, Specular, SSS, etc. But by doing it this way, I realized that I didn’t need so many maps, so I ended with a much cleaner and efficient material and a better result.

For the clothing, as mentioned before, I used the V-Ray 2 Sided material to give a different tone to each side like in the reference and also add a subtle see-through effect to the fabrics.

The materials for the fabrics were done directly with the maps from Substance with the V-Ray Metallic/Roughness UDIM Material exporter. I used some color corrections and remap value nodes to adjust the materials during the look dev process.

You can see my texturing process in Substance in more detail here.

Challenges

By far, the biggest challenge was to capture the expression. When I started the project and was searching for references, I decided to look not only for real-life references but also for 3D characters with a similar smile. While doing my research, I found out that there were not many of them, and after working on the project for some time I understood why.

To make a character realistic on its own is already a really tough challenge, but when you work on a smiling character, it is completely a different story. There is a really thin line between a cute and a creepy or fake smile. Even in real life, if we try to smile when we are not really happy in general, the human eye spots this automatically.

Check, for example, this image showing how the sculpt looked at the beginning:

This is because it is not only the mouth that changes when we smile (the mouth was completely wrong at that moment anyway) but also almost all the facial muscles are set in motion. For example, when I couldn't get proper wrinkles around the eyes, the smile was looking completely fake and creepy. Also, I had to make the eyes a bit “watery” to help with the expression. It took me around half of the entire sculpting process to get the smile to work.

One of the main things I learned was how important it is to know anatomy and understand what is happening underneath the skin to get a realistic look on the outside.

Also, I understood how important it is to do good reference research and study the character or asset that you are creating. Observation is key, and you need to put a lot of hours into it.

Next Project

Currently, I am working on a new character with a really different style. In this project, I am doing a full character with some hard-surface props and focusing more on body anatomy and asset creation to expand my portfolio. You can take a look at some WIPs here.

Tomas Sackmann, 3D Artist

Interview conducted by Ellie Harisova

Keep reading

You may find this article interesting

Luna: Cloth Simulation and Hair Setup with XGen