Creating a Stylized Fantasy House in Blender and 3DCoat

Introduction

Hiya! I’m Brian Nguyen, a stylized 3D Artist, a person who loves fantasy and cute, weird, wild, and wonderful things. I got into 3D art after trudging through various different art paths like animating, compositing, and filmmaking in animation which I still like to do as a hobby to practice writing and make projects that aren’t still images. 

The stylized part was inspired by Ashleigh Warner after I took her CGMA Course "Creating Stylized Game Assets". She answered a part of the question I've always been fascinated by: "How can you take 3D and make it feel 2D?"

During my film studies, I saw a teaser trailer for Arcane six months after its release and I wanted to make something that looked like it and understand how they made it. I started a small project group called "D.R.E.W." which let me practice painting, lighting, and compositing, where I got a brief understanding of how Fortiche achieved Arcane. From there on, I was inspired to continue this pipeline and improve the quality of the work that followed it. 

The Big Fantasy House Project

I was working at SMG Studios on Moving Out 2 which was art directed by a talented person called Jack Kirby Crosby. This talented person made an incredibly detailed concept art which they didn’t really have a concrete name for and I thought, "Why not take up the challenge?"

I wasn’t going to do the whole thing. My goal was to learn more about color, shape, and structural intent/design. I was going to focus on "The Big Fantasy House".

Fun fact: When I told my colleagues I was going to hand paint something made of 900 different parts, they immediately called me a masochist, and I couldn't argue with that remark. The workload was kind of crazy, and I'm not talking about modeling – that was the fun part. It was the UVs that made me want to quit and start a life on a farm.

Planning the Initial Composition

The first step to planning was to understand the biggest shapes that make up the building and the focal length for the camera. In this case, I strategically stared at the concept and drew out the major shapes I could see. These would serve as an important reference to get the camera setup working. Getting the correct camera to line up with the concept is so important and an extremely mundane task. It must be done and that’s why you should be thankful for your layout artists. 

In the end, the camera couldn’t align better with the concept so I had to change the back of the roof to a unique shape causing all the models in that area to become bespoke and slanted. It later became a happy accident as it looked more interesting in 3D.  

Modeling

My workflow began in Maya but as I learned more about Blender, its selection tools, and its wonderful hotkeys I migrated the scene over. 

In Blender, it was the standard approach. I started with the big shapes and worked our way down to the smaller ones. When it came to the details, it was really hard to see them. When you import an extremely detailed concept into Blender, you can see a lot but you might miss a detail or two if you’re not careful. It’s why planning goes a long way. It helps you understand what you’re making.

To be honest, the only trick I have is setting my y-axis shortcut to "C". If you model enough in a lifetime, you become very fast and intuitive at it. My mind defaults to figuring out how to hit a shape in the quickest and best-looking way possible. If I've made a similar shape before, then I can recall the knowledge gained from that experience and try to improve on it. That's why practicing and making different things can help you improve your modeling skills.

As an example of something that isn't purely art, this post showcases 20 different ways to create moving Lego contraptions.

For things like grass, roof tiles, and vines, to speed up the process I used Geometry Nodes and modifiers like "Array" to scatter objects as a base which I spend time refining in Blender.

Texturing

My texturing workflow begins by creating a rough trim sheet, which I bake in Blender, that I use to help tag what kind of object and material I’m painting on. I paint over the trim sheeted textures in 3DCoat to get the first base of colors.

The next step is very important and I found it hard to be confident in. You start painting in the colors that contrast the most with each other in hue and temperature with a basic round brush. This is where we plan the texture details and temperature of our objects. After this stage, I refine the shapes and start adding in the smaller details then finalize it by blending targeted brush strokes together. The emissive map is painted after.

Due to the object having so many textures, I had to compromise on detail for each texture and settle on adding details to focal points in the composition. This means I rarely used Photoshop as an external editor.

Assembling the Final Scene

Assembling the scene is the easy part and extremely useful to do in the middle of the project to catch anything you can improve, especially in my pipeline. In my pipeline, colors change drastically due to shadows and can only be changed either in compositing or directly through colors.

Final Lighting Setup

The lighting setup was very simple. It’s sunlight and two lights. The lights influence the shadow pass and only their position impacts the result.

The kind of complicated part was the compositing pipeline. Instead of Nuke which I usually use, I worked directly in Blender’s compositor which made the process very streamlined. The AOVs I used were: Color, Ambient Occlusion, Shadow, Emission, and Cryptomattes <Objects, Material>.

For compositors, it’s like playing with your AOVs to get new results. The way my compositing pipe works is by adding on different passes on top of the color with various opacities. For example, Shadow was set to 75% and Ambient Occlusion was set to 20% opacity. This gave a graphic look to the result that I aimed for.

Between the basic blending modes I influenced the shadow and ambient occlusion passes with cryptomattes and color corrections to add color and mood to them.

Animations were driven by armature rigs in Blender. For foliage, I used the "Wave" modifier and for the water, I created my own shader influenced by dynamic paint. You can find the tutorial on how to do this here. Getting the right settings and colors mattered heavily.

Main Challenges

This entire project was a challenge! It is the most complex model in terms of the quantity of geometry I’ve modeled to date. UVing 900s objects was a nightmare, always seeing there were details I missed and going back to add them during the texturing process taught me to maintain my "flow" but it always felt incredibly annoying.

I achieved my goal of learning how to paint more confidently, and to recognize and create shapes with depth and structural intent/design. I’ve also learned that just like drawing human anatomy, you need to find your reference points to help guide you to hit the concept and get your proportions feeling right. 

For those struggling with complex art problems, my advice is to allow yourself to process chunks of information, understand it and then play it by ear. If you have a fundamental understanding of something, you can slowly figure out difficult problems if you give yourself the time and mind space for it.

If you lack the motivation to do your own art, find the fundamental truth about why you do art. Mine is to "inspire joy and community" and it's why I work in games! I find that most of the time, we love the recognition of our art but when art becomes an invisible language that benefits others, that feeling of elation becomes immeasurable.

Brian Nguyen, 3D Artist

Interview conducted by Theodore McKenzie