Alex Engelmann did a breakdown of his UE4 environment Tug Boat Interior made during the recent Polycount Challenge and talked about his approach to translating a 2D concept into 3D.
In case you missed it
Study more stylized pieces
Hi, my name is Alex Engelmann. I’m currently a Lead at Torch Technologies focusing on R&D visualization work.
I got into 3D in high school when I discovered the computer lab had 3D Studio Max installed. It didn’t take long for me to get hooked and focus on that as a career. I went on to earn a Bachelor of Fine Arts with a focus in Multimedia Design at the University of Wisconsin-Stout. The program at UW-Stout covered a wide variety of aspects related to game production, on both coding and art. I attribute a lot of my success to the foundations formed during college.
After graduating, I applied to many game studios and was fortunate enough to land a contract at Turn 10 Studios, part of Microsoft, working on Forza 3 as an Environment Artist. I learned so much from the team there and that experience really cemented my love for Environment Art in games.
After my contract at Microsoft ended, I moved back to the midwest and took a job at CRASH+SUES, a post-production studio in Minneapolis I had interned at during college. There, I learned a ton about lighting, PBR materials, particle simulations, animation, and high-poly modeling. Back then, there was a larger difference in doing 3D for real-time applications and VFX work. Things like PBR materials were first becoming popular in offline renderers like Mental Ray. And things like ray-traced lighting and particle simulations were well outside the realm of real-time applications. In time, those skills I picked up all made their way to real-time applications, so I felt like I had a head start when they arrived.
After a number of years doing VFX and motion graphics work, I started my current job at Torch Technologies. They had just landed a contract to create a real-time training application for the military and they needed a team to build it. As one of the first members of the team, I was tasked with setting up the art pipeline and creating much of the original art for the training software. Below is a trailer that highlights the training software:
Over the past four years, our team has grown from 5 to close to 40 people and we are still going strong. In that time, I’ve transitioned from an Art Lead to a Lead in the R&D group. My group’s current work focuses on incorporating virtual reality into training software. We just released a trailer for a medical anesthesia trainer we are developing:
I would attribute most of my success to always trying to learn new things, even if they aren’t immediately applicable to my job. I am constantly researching new techniques and software through articles online and Youtube and practice these skills in my free time on small projects. Picking up new skills helps give me a better understanding of the game development process as a whole, and assists me in solving issues that arise during the job.
Tug Boat Interior
Tug Boat Interior: Origin
Overall, I have been very pleased with the end result of this Polycount Challenge project and the response it has received. For those unaware, Polycount hosts bi-monthly art challenges that ask participants to choose from a limited selection of concepts, props, and environments and recreate them in 3D. I enjoy looking through the concepts when a new challenge is posted but with my time restraints (between my job and kids I don't have a lot of free time) do not typically participate. When I saw the Tug Boat Interior concept by Nik Henderson though I was immediately drawn to it and knew I wanted to take the time to enter. I felt the concept was well defined, had a ton of personality, and would be enjoyable to work on in my free time.
I’m sure it comes as no surprise that blockouts are an essential part of any environment work. They help identify issues when translating concepts to 3D, establish scale and proportions, and can be used to estimate the workload for the project. For this project, the concept was really well done, so there wasn’t much issue translating to 3D, however, the blockout helped me size many props correctly before investing any time into detail work. I did my initial blockout modeling in 3ds Max. As you can see, many of the shapes are just glorified boxes, but I paid close attention to scale and proportions so I could rely on that later.
I gave myself a month of evenings to work on this project, so the blockout helped me set a pace to complete everything on time. My initial schedule paced me at over an object a night, which sounded like a lot at first, but many of the objects were simple and could reuse materials.
After blocking out the models, I imported them into Unreal, threw in some basic lighting and textures for the walls and finalized my blockout. From here on out I could iterate over each prop individually confident that it would fit in its place within the context of the environment.
Working on Stylized Forms
One of the reasons I gravitated towards this concept was how the shapes nicely balanced cartoony silhouettes with real-world volumes. I wanted to try and recreate that in 3D. In order to give them a more cartoony feel, I had to add extra geometry to many of the props to allow for curved surfaces and irregular silhouettes.
The curved silhouettes and tapered parallels broke up the silhouettes enough to get rid of the normal 3D rigidness. Undulations in the surface geometry also allowed lighting and reflections to appear irregular.
The final step I took in the modeling process to achieve the stylized look from the concept was to pass most assets through ZBrush for a damage pass to help break up the surfaces. In addition to adding normal wear and tear, I went over many of the sharp edges by hand with the trim dynamic brush to add a hand-painted feeling to the edges of the model.
During the retopology phase, I chose to keep many of the edge bevels in the final model. In my day job, we deal with very high-poly scenes and have found that Unreal does an amazing job with auto LOD creation, and handling large amounts of geometry.
In order to establish the overall style, I focused on the refrigerator first. I took the blockout mesh into ZBrush, built and sculpted the high-poly model, then decimated it and brought it into 3ds Max for retopo and UVs. When I was happy with that first asset, I copied the process for all the other assets.
As with the modeling phase, my goal was to balance realistic material properties with the stylized painterly feeling I saw in the concept. To achieve that goal, I started by blocking out base materials for all the components of a model. During this first stage, I’m focusing more on the material definition of all the components, trying to nail down base color and roughness values. At this point, I don’t worry too much about stylization.
Once I’m happy with the base materials, I do a pass adding in some painterly effects to the materials. I usually start by adding subtle gradients to all the surfaces to break up color uniformity and add what I think of as a large scale ambient occlusion pass. Then, I add soft edge highlights that lighten the color and lower the roughness on a lot of the edges. Once the overall material effects are added, I will do a grime pass where I paint in dirt and stains on the model.
In both the modeling and texturing phase, I’m trying to balance realistic detail with stylized abstraction. I try not to make any details too small or noisy and let the model and material be defined by larger shapes and gradients. I feel that hitting that balance is what makes a stylized asset work well.
Similar to the modeling and texturing, I wanted to try and translate the mood of the concept with the lighting. In particular, I really liked how you can almost feel the air in the concept. Once I assembled the blockout in UE4, I did a rough lighting pass with just a directional light and a skylight.
After adding details and textures to the models, I looked at the lighting again. I felt that the amount of sunlight coming through the windows didn’t match the concept, so I added portal lights near all the windows with the sun coming in. This helped simulate the scattering of light through dirty windows, and it added a glow around the windows that helped to give warmth to the scene.
As the project neared completion I ended up taking out the skylight as it was making the scene too blue. In its place, I added portal lights to all the windows. This gave me greater control over the color and intensity of the lighting all around.
To finish it off, I added a pretty heavy-handed exponential height fog and dust mote particles to give the air some thickness.
A few other things that contributed to the lighting were the reflection probes and the post-process volume. I can’t emphasize enough the importance of reflection probes in UE4. They contribute quite a bit to the final image, especially if you have reflective items in the scene. They are simple to set up, but sometimes they are easily forgotten. Lastly, I used the post process volume to add a little bloom and chromatic aberration to the scene.
Time was certainly a big challenge in this piece. However, good planning upfront and the feedback and encouragement on the Polycount Challenge thread made it easier than expected. In the end, I found that sticking to the concept while trying to inject my own flair into the piece was probably the biggest challenge. That is something that I am always striving for. I don't know if I overcame it, but it is something I always work toward and continue to improve upon.
Working on these personal projects, especially if it’s part of something like a challenge or contest with external feedback is a great way to improve as an artist - both in improving your own work, but also in learning how to translate other people’s concepts into 3D. The lessons that I learn in projects like this one are invaluable, which is why I enjoy participating in them.