Balancing 2D & 3D Techniques in Concept Art

Introduction

Hi there! My name is Artem Kurenkov, I specialize in 3D concept art and work as a Concept Artist. I graduated from CG-School a little more than a year ago, and since then I worked on several mobile projects for Wargaming Nexus. Recently, I decided to switch to freelance. Before becoming a Concept Artist I worked as a 3D Modeler. It's worth mentioning that my first steps on this path involved model making for Morrowind, and then I worked in various fields from gamedev to archviz and VR. 

I can’t say exactly when and why I changed my mind and realized that I wanted to become a Concept Artist. Perhaps it was because I was always more interested in the ideation process. Or maybe it’s because I’m too lazy when things come to retopology, baking and so on.

Approach

When I’m trying to deconstruct my approach I call it “3D-2D bunnyhopping“. During the concept creation process, I tend to switch between 2D and 3D media. I can’t say that I follow some strict rules but the rule of thumb is when I feel that I am swamped with one medium, I switch to another. And as I do this my perspective changes. I’m not entirely sure but I believe that the brain uses different zones for drawing and modeling. Hence different angles of view.

The perspective is not the only reason I do that, however. Every approach has its own advantages and limitations when we talk about ideation so it’s better to use them in proper time. As I said, I have no strict rule for “dimension shift” (sounds cool, huh?), but I guess one of my old works can describe the general Idea and, eh, philosophy well enough.

Note that I’m not going to tell you how to model things - I guess the readers of 80 Level can find more valuable materials on this topic on the website.

Process

2D is very good for the first stage, just to break “new canvas paralysis”. You simply start to sketch something and “go with the flow”. 

Every sketch in the first iteration can be imperfect (or even ugly) - its purpose just to crystalize the idea that will inspire you. It’s better not to dwell on any of the specific details. In the second stage, you can realize that this detail works only in 2D.

The second stage in my pipeline is usually a major/medium forms research. And 3D is perfect for this. 

Usually, I start with a sphere or a bunch of spheres. I just crumple them until I get the main shape of my initial sketch. I remember that when my colleague (she is a 3D Artist) saw the blockout for this particular project, she almost had a stroke because of the topology. 

But this clumsy blockout allows me to search for the best form and composition without risk of overdetalization. And since I’m doing it in 3D I can be pretty sure about how it’s going to look from every angle.

A hint of a century: you need to model fast. So an optimized working environment is crucial for this stage and all upcoming steps in 3D. If you need to go to the menu to use extrude it is time to stop the modeling process and apply some hotkeys, create some popup/pie menus and practice with it. You need to learn how to save seconds for the sake of saving hours.

The second hint is to use a Polyshift tool to move vertices - Grab tool in Maya, Sculpt Mode's Grab in Blender, Move tool in ZBrush. It prevents you from digging into details.

When I have all forms balanced I can switch to 2D again to search for details with lineart. The trick is that lineart is a very fast tool in terms of searching for small details. Your hand just does what it wants without thinking about “how I should model this tiny thing”.

It’s also a good time to use some references. 

Then I go to Max again to model the small details to the level where I can overpaint them.

The other option for this stage is ZBrush, I used it in one of my last works. As a bonus, you can throw the model into Keyshot (or even 3ds Max or Maya, if you know some tricks).

Materials

Material research is a vital part of any design. And this is where the 3D approach starts to cast some forbidden magic.

3ds Max, Maya, Blender - all of them have pretty powerful tools for material creation. And all of them have awesome renderers. I’m using 3ds Max with Corona renderer, so I’ll tell you a few tricks for it. 

First of all, in a classic 3D application you have full control over your geometry. You can apply any material to any part you want.

Secondly, you have control over UVs. You can control pattern direction, its scale, etc.

But these two things are mundane and can be achieved in 2D. The most important thing is:

You. Have. Nodes.

The only thing you cannot do in 2D in terms of materials is that you can’t change them in seconds. You can’t mix properties, patterns, colors, and roughness. In 3D, you can do it quickly and observe the result in Corona Interactive on your model. 

Only one application can give you more control, and it's Substance Designer. However,  it’s not very useful for searching for the right materials for a specific model.

Let's take a shoulder pad material for a new character as an example. The material itself is pretty simple. 

To specify what it does: it applies bump and displacement textures on the model. Displacement is masked by vertex color. I applied it through the Blue channel on the places where it was too wrinkled. 

Also, it adds the displacement height map to bump pattern for reflection glossiness.

You can see how easily the material can be modified. Substance Designer users might giggle at this moment, but it is enough for concept art purposes.

From here, I start to look for the right version of the material. I made 6 options and spent 2 minutes on that.

A will spend another minute to check out how it works with other materials.

The process is that simple. Note that it will get more complicated with more complex or render-heavy materials.

And yes, it involves modeling and applying propper UVs to create this material. It doesn’t sound fast which is important for concept art but you will get the time you lose on modeling back in the rendering stage. And your team will save another working day since the modeler will already have your model as a blockout. Oh, yes. This is an important thing about 3D concept art. Let it be the stand-alone chapter.

Why Use 3D at All?

Because of the overall profit. A 3D model - even with ugly topology and artifacts - saves you time and nerves.

There are two reasons to use it:

The first one depends on your skill level, but sometimes, a modeler can use your mesh as a high-poly. It needs to be tweaked a bit, of course. With the blockout stage, it can save up to 2-3 days in the overall pipeline at a studio. 

Secondly, a 3D concept can be rotated and observed from any angle. In some cases, you can export it into your engine to see how it looks in the environment. As a result, you and your team will have a general idea of how it gonna look in the game. It saves lives sometimes. 

The profits I mentioned above double for environmental design. An environment is a 3D scene per se. It has more than 3 projections and you need to feel it as a space. Any 2D concept here can only be described as an “atmospheric sketch”, in my opinion. Only when developing the concept in 3D, you can have a complete understanding of what you are doing. 

But does 3D replace 2D techniques? Hell no. Complements and strengthen, yes. But it is important to develop your sketching/rendering techniques, too. You need two legs to stand firmly on the ground.

Overpaint

This is the final stage of my process. For some works, it involves just fixing artifacts and adding fine details. For others, it makes up 30% of all the work (like in this project, for example).

I did the overpaint mostly in ZBrush in an attempt to shorten my pipeline. It was quite fun, but I had trouble with exporting the polypaint to Max. It's a shame, but 3ds Max cannot import ZBrush Polypaint through “.fbx’’ for some reason. However, Maya can. One of my friends successfully re-exported my model into “.fbx” so that 3ds Max was able to read the vertex paint information.

The problem is, roughly applied vertex paint looks awful on the model.

So, it took some time to adjust everything in the material editor to get the result that looked good enough for overpainting.

It’s too long to describe it here, but the main principle is to get the masks for glossiness and reflection from the surface color. Then I am just… painting. Below are two GIFs that describe the process:

When I have proper colors, reflections, and shadows it is pretty easy to render the details. Here are a few pieces of advice:

Firstly, it’s a good idea to put some rough brush strokes on your render. Novice overpainters give too much credit to 3D renders as they are physically correct. But it’s always better to add some chaos to them and then put it in order.

Secondly, use Lab color space and Lab sliders for color picking. In my opinion, it’s better for mixing colors. When you work in RGB for too long your picture starts to lose saturation. And Lab sliders give you the correct color tone gradient with its Lightness slider. 

Conclusion

3D modeling can be used to improve the concept art pipeline but you need to be fast in modeling. You need to understand when it's time to stop modeling and start drawing. But the benefits are obvious.

Take time to improve your drawing skills though. A lot of people tend to rely heavily only on one technique, and you need to be fluent in both.

Artem Kurenkov, Concept Artist

Interview conducted by Kirill Tokarev