Learn How Rock Texture for Skull and Bones Was Made in Substance 3D

Introduction

Hello everyone, I'm Chun Wei Ong, originally from Singapore. Currently, I'm a Senior Material Artist at Krafton's PUBG. Prior to this, I was in Ubisoft Massive, where I worked on Avatar: Frontiers of Pandora, and Ubisoft Singapore, where I worked on Skull and Bones.

My first ever experience with 3D was making CAD models for my secondary school projects at age 15. The software was called Pro Desktop. It's discontinued now, but I was decent at it. I started making money by helping my classmates make their models.

Afterward, I attended an animation course at Nanyang Polytechnic in Singapore to learn art, even though I knew I wanted to work on games. Very quickly, I knew I liked texturing, even though it was done in Photoshop back in the day.

When Unreal Engine 4 first became free, they released the kite demo which I was so impressed by. Everything was in real-time, and it looked better than anything I have ever made with Maya's Mental Ray. So, I started learning Unreal Engine 4 and Substance 3D Painter by myself, even though I didn't even know what Normals were, and I had no concept of texel density. I remember texturing a whole house in Substance 3D Painter (don't do that).

Afterward, I worked on a number of other personal projects. I kept learning more about game art when more and more information was becoming available online. FlippedNormals, 80 Level, and Polycount threads were among some of them.

Getting Started on the Project

After I had finished my military conscription in 2018, I saw an opening for a 3D Artist internship in Ubisoft Singapore, which I quickly applied for. However, a week later, I saw another opening for a Texture Artist (Assistant/Intern). I was semi-panicking because I wanted this more than the 3D Artist internship, so I applied for it anyway. 

Surprisingly, I got called up for an interview for the Texture Artist role and they told me the internship position was filled. I took an art test and another two interviews and I was offered the role of Texture Artist Assistant, which was a junior contract position.

I spent about four years working on Skull and Bones, initially working a lot on the world and environment. I learned a lot from my lead, Nathan Tang. He taught me a lot about making trimsheets and terrain textures, which are invaluable skills for a Texture Artist. 

We used a lot of procedural rules for terrain, which was iteration-friendly, as you can define which material should be spawning at which height and slope. For example, you always want beach sand from 0m to 2m, but they should only spawn at 0 degrees to 20 degrees. Otherwise, it should be mud or rocks.

These things helped a lot with Level Artists resculpting the terrain and not having to repaint terrain materials.

In the latter half of my career, I worked more on ships and a bit of characters. I worked closely with and learned a lot from Xu Xiao, a Senior Technical Artist, and assisted him in defining the ship pipeline and how customizable it should be.

The Choice for Substance 3D Designer

Substance 3D Designer is a program of endless possibilities, isn't it? There are insane projects like remaking Lion King scenes by Kelly Recco.

The procedural nature of 3D Designer is its selling point for working on games. Not too long ago, I was given feedback on reducing the number of leaves from muddy ground material. If the material were made in Photoshop or Substance 3D Painter, I would have to spend hours Clone-Stamping out leaves. In 3D Designer, it is just a slider to reduce the amount of leaf scatter, and it can all be done in a minute. It is also easy to make a different version of a material by changing some parameters, the color, and the random seed.

Texturing

Often, as a Material Artist, when you receive a task, it is accompanied by a reference. And it will frequently be the smallest, blurriest picture in existence. 

The first step is to find out what the thing is called or where the place is. This is super important. Ask the requester or do a reverse Google search to find higher-resolution images and more angles.

Now that I know this is the sandstone at the Isimila Stone Age Site, I can start looking at what I need to work on.

I know that this is how the asset artist plans to split the material. The darker rock on top will be material 1. And the pillar will be material 2.

Always work from big to medium and then to small whenever possible. This will often be the big shapes of a silhouette on an object or character. In material art, it is hard to discern silhouettes. Think of it as, if you were to displace the material, it should give you the biggest shape change. In this case, it's the big carving-like features.

And then proceed to the medium features. In this case, it's these horizontal layers and some of the smaller vertical erosion.

And finally the small, high-frequency details. Beginner artists tend to overdo the small details, making the image very noisy and taking away attention from the other parts of the material.

Finally, do a balancing pass to make sure you have some rest areas for your eyes so it doesn't become too noisy or overbearing. This often works differently from material to material, and it would take a lot of practice and iteration to get right.

Let's look at this in a Height Map creation process for this material in 3D Designer:

And also another Height Map creation process for another rock material:

Creating Appealing Textures

Everyone would agree that good Height/Normals are the undisputed champions of good material, but I think a good Albedo is just as important.

Think about this: before we had PBR materials in games, the textures in games were often just photographs of a nice material in real life. Game textures used to work without Roughness Maps or even normals. So, I think the key is to make Albedo interesting and detailed. I like to use the Gradient Map node instead of Uniform Color whenever possible.

Let's take a look at this trim texture:

I try to avoid having flat colors whenever possible because if you look at any scan data, you cannot find any flat colors. To break down some of the things I love doing in the Albedo:

• Use Gradient Map Node
• Put some saturated areas (needs to be subtle, few and far between)
• Use some complementary colors for wear and tear
• Splash some analogous colors to break the flatness
• PBR Accuracy.

And one more example of this:

Timeframe & Main Challenges

Time to finish one material ranges from half a day to a whole week, depending on the complexity of the material. A basic wood tileable material would take half a day to a day. A complex trimsheet with intricate details can take up to 3-4 days or even a week.

Typically in production for materials, you would never want to put 100% polish on every material right away. There will always be a very high demand for a large volume of materials at the beginning when you are trying to build a library or fill up the world. You need to churn out materials by the buckets at the beginning. I would say always get the material to look good but not great first, about 70-80% quality.

Later in production, as the library matures and Level Artists are happy, you will have some pockets of time here and there to reopen one of those materials to look at it with fresh eyes and add polish, not to mention you'd always be older and wiser.

The last 20% quality of art usually takes as long to achieve as the first 80%, so leave that till later when production slows down.

Conclusion

There is an abundance of resources online for artists to get into game development these days: 80 Level, ArtStation Learning, and lots of YouTube tutorials. But understandably, it can be very taunting, and you never know if what you are learning is relevant. Thankfully, if you are starting out, you shouldn't worry too much about that.

The best place to learn is actually during production. You will learn more about game development in a 6-month internship than a few years in school. Of course, you would need the fundamentals before you can get an internship.

For an artist, I would recommend starting out making something you find cool, but avoid making it too large in scale. Don't try to build your own cyberpunk city, you will never finish the project. Start with a cyberpunk prop, or a little diorama. Train your eyes to observe and find cool details. If you want to make a prop, don't make your own concept, build one from a concept artist you like.

As cheesy as it sounds, the most important part is to have fun. We are in a fortunate position to work in the game industry, and if you have fun with what you do, you won't be burning out anytime soon.

Chun Wei Ong, Senior Material Artist

Interview conducted by Gloria Levine