Theodor Gerhamn talked about his amazing work and shared tips on recreating realistic looks.
Introduction
My name is Theodor Gerhamn, and I’m a 3D artist working at EA DICE here, in Stockholm, Sweden.
A couple of years ago, back when I was a student at Futuregames, I wrote an article called “Secrets of Realistic Look in Real-Time Scene” about an Unreal Engine 4 scene I made and my process of creating it. Since then, I got an internship at EA DICE and later employment, where my first game I got to work on was Battlefield V.
This time, I will write about the process of how I created my new personal piece: the Rolls-Royce Pelican.
Gathering the Reference
I like to do a lot of my own projects at home after work, where I can practice new software och techniques. This project started off when I wanted to practice modeling complex shapes and interpreting references by modeling a Rolls-Royce Merlin engine, which was used in the Spitfire planes during the second world war.
When I was almost done with the high poly of the engine I started to feel like I wanted to take it a step further and use it for something.
Whenever I start a new project I try to have a personal development goal in mind. Do I want to learn new software for this project? A new tool or script in Maya? Or make something new that I haven’t done to push myself creatively.
Most of my non-professional portfolio is from when I was studying and trying to get a job. Back then I almost only did models, environments and textures of things that already exist and that I could find good references of. But I had never really designed anything myself.
During my work on the engine, I found these beautiful racing cars with Merlin engines in them. This gave me an idea of seeing if I could take this engine with some inspiration from those cars, cafe racers, and Spitfire airplanes and make something new.
Designing
I wanted to use the engine that I made previously for the bike as a centerpiece. Initially, not so much in terms of composition or focus for the viewer, but to build around. Like the poor mechanics that made the bike would have been given the engine and, then, tasked to design something around it.
Since I had never really done any designing before, I tried to study my different references and break them down to what made me find those cars, bikes, and planes look appealing and interesting. Looking at the older racing cars that also used the Merlin engine, I noticed that they don’t really look slim and sleek like modern sports cars. Frankly, they look kind of heavy, clunky and hand made.
With this in mind, I started to plot down some keywords that would describe the bike that I could refer to when I was modeling and designing it. I knew I would experiment a lot with the design during the process and I was aware that it was possible that I would just get a cool idea to put on the bike, but it didn’t really fit the aesthetic that I was originally planning. So the keywords would keep me on course during the process.
Some of the keywords I chose were: Heavy, Powerful, Simplistic and Ridiculous.
Ridiculous was one of the later keywords I added to the list, and I think it might have been the most important.
When I finally started to block out the bike, I originally placed the engine low to the ground, as the center of gravity would have needed to keep low if you wanted to ride this heavy thing. But I didn’t like how it looked, but I couldn’t really motivate why I thought so. I looked more at the references of the Spitfire airplanes and the cars, and I realized that some of the most iconic points of view of the Merlin engine are from the cockpit (or a similar angle) of a Spitfire plane. Where the pilot can see the exhausts from the nose of the plane spitting flames. What would be the point of having this iconic engine on a motorcycle if I can’t create a connection and reference to what made it famous? Having it hidden from the rider would feel wrong. I needed to put it up high near the rider. This is where “Ridiculous” started to become the main keyword. I started to let go of the worries I had of creating something “realistic” that actually was functioning and set my creativity free to try and create something that would be more fun to me. Granted, having a sound understanding of real-life mechanics and design will help anyone when designing anything. Whether it be a realistic prop that is supposed to be functioning with our laws of physics or a fantasy gun where aesthetics are the top priority. This is something I try to become better at, and I think it will make me a better artist in the future.
With the engine placement determined and my desire to create a connection to WW2-era fighter planes, I started to focus on top body cover. This is what the rider would see the most when riding and if I could get this part to a shape that I liked, then I could form the rest of the bike to fit around it. I try to imagine that there is a person who ordered the build of this and that I’m building it for him/her and not for the average Artstation viewer.
The white top cover of the bodywork.
I had done the blackout of the front half of the bodywork and main parts of the frame first. Taking a step back and looking at the progress, I realized that if I continued with the size and shapes of the front for the back of the bike, I would create the thickest bike ever conceived. I tried to create some good shapes that would fit with the front. But I saw that the focus of the ridiculously large engine started to drown amongst the large shapes of the bodywork. This is where the decision to keep the back of the bike minimal came in. The harsh contrast of the large front and that tiny seat could really emphasize the large engine and the ridiculousness of the whole build.
The small seat and negative space create a contrast of size with the large front.
Modeling in Maya
As for my workflows, I originally had plans to learn some CAD software like Fusion360. But having to wrap my head around design, quickly blocking out and executing on my ideas, I felt like Maya would work better for this project as I knew the UI and didn’t have to spend time to learn a new one.
Using standard SubD modeling workflows and Creases is something I’m very comfortable with and is what I used to create all the shapes. For the tubes of the engine and wires along with the bike, I used the QuickPipe script for Maya, created by the awesome Adnan Chaumette. Another of Adnan’s scripts I love to use on almost any personal project is Mesh Blend.
Mesh Blend allows the user to save various kitbashing meshes into a library which you can then use to place as floaters along with your asset. It can also embed them into the surface of your target mesh. I used this script extensively for all the screws, nuts, and bolts on the bike.
Texturing
The first question I had to answer when it came to the texturing was the amount of dirt and wear the bike should have. I think it is a common thing for us artists to add a lot of grime, dirt and wear to our assets because that creates details, which will hopefully look good and interesting. A metal surface with a bunch of rust spots and dents usually looks more interesting than a fairly clean metal surface. Usually. And this was my first plan of attack. But I love texturing so much in comparison to any other part of the 3D workflow. I had just finished designing my first model and felt fairly happy about it and saw it as an achievement. I could now challenge myself with the texturing as well.
Similar as with the modeling workflow and the keywords, I decided on a sentence to give me some guidance for my texturing: The owner has taken the bike to a photoshoot and gave it a quick cleaning before arriving.
This sentence might mean something completely different to you than it does to me. But reading that sentence instantly plays the scenario of the owner cleaning his bike in my head. He did his best, but honestly, he missed a few spots and he was sort of sloppy with certain areas due to stress.
I used Substance Painter for all the textures on the bike. When it came to the engine, I imagined it being fairly grimy and dirty since the sides of it are exposed to the air and dirt when riding. The owner gave it a quick rundown but focused more on the more visually distinct and clearer parts of the bike. So I didn’t cover it with dirt, but I made sure to gather dust, grime, and oil in most crevices of the engine. Some rust leakage from the rusty screws as well.
This engine is fairly clean, but if you look closely at it, you can see some residue of grime that is still there.
The hardest challenge was probably the painted metal body that takes up a lot of screen space. How do I make a smooth and fairly uniform surface interesting without adding a bunch of dirt and wear? Thinking back to my guiding sentence about the owner and his shoddy cleaning I imagined that he hosed down the bike a bit and then took a rag and started to wipe it down. Doing this poorly would leave some wiping marks on the body from the soap, water, and remaining dirt. It would probably also leave dirt and grime in between screws and tight corners where the rag couldn’t reach. As well as small dried stains from running water.
Substance Painter has some really nice grunge maps already. Several of them look similar to having been wiped with a rag or something similar. I added one to the bike in both the Glossiness and Base Color, but just adding it onto your mesh will look uniform and unrealistic. Even if the owner is really bad at cleaning his bike, he wouldn’t leave the same pattern of wipe marks across the bike. To fix this I started to add more and more grunge with different types of wiping patterns to them, as well as some grunge to break up the surfaces in different ways. Some grunge effect added more roughness to the painted metal, whilst others made it more glossy. Some were also used to mask out the effect. And some were used to mask out the masks! I tried to layer lots of grunge and smart masks to get a variation in the gloss map and the base color, but also to make the grunge and patterns look unique wherever you looked.
If you look at a lot of Substance Designer materials posted online, you can often see a repeating pattern sooner or later in the textures. This is what I wanted to avoid. Trying to make it look like every stroke of the rag on the bike had been affected by the shape of the body, the previous grime that had been moved around and multiple elements that could affect the outcome.
The key here, I think, is being subtle. It is easy to always drag the opacity slider up until you definitely see that the added grunge you applied is showing up and affecting your material. Instead, I tried to keep the strengths of the grunge effect really low and let them build up together when everything is in place. But that might not be as simple as it seems!
This subtlety and build-up of dirt and grunge is something I tried to maintain all over the bike. I only added more of the dirt to areas, which would have been really tough to clean and were constantly being exposed to it.
A quick tip I would give to newer 3D artist and students is to blend shapes. This is a 2D drawing principle, but you’ll find that it works really well when texturing 3D assets as well! By this, I don’t mean that we should add some weird filter to our assets that blur edges and whatnot. But if you have a sharp concave corner or intersecting meshes, like a nail against a wood surface, then try to add some dirt, darkened color or something that slightly hides that intersection and merges the shapes a bit.
Credit to Marco Bucci for an awesome breakdown on this.
When you zoom out you will notice that your asset’s different parts don’t seem to hover on top of each other anymore and that they are more grounded and connected.
Rendering
I wanted to render this piece similar to what you would be seeing for car renders and photoshoots, but try to keep it from not looking to clean and perfect. Modern car renders and advertisements are very clean with harsh highlights and shadows. Even though, I’m a big fan of those cars and product renders, I wanted to maintain this human element where it is actually a bike that someone owns and takes is to get it photographed.
As I knew, I still wanted some type of studio environment I chose to try Keyshot. At first, I tried a fully black background with some artificial lights in a 3 point light setup. But I wasn’t too big of a fan of the flat reflections it created. I later tried to just add one of the standard HDRIs of a large warehouse that comes with Keyshot. Right away I started to get some more interesting reflections! With that, I rotated the HDRI into a position I was happy with. I was surprised to see how simple the solution to my lighting was when I was prepared to battle with it for a long time. For some shots, I stitched together two different rotations of the HDRI if I felt like the lighting wasn’t enough.
The aim of the lighting was not to make every visible surface interesting with a bunch of reflections and highlights. I chose an area/subject of each camera angle and tried to pull focus there. I’m a big fan of landscape photography (and all types of photography), where it often helps your photos if you have a clear subject to draw the viewers eyes towards. Having a focus often makes an image more interesting and clear rather than a bunch of cool things smacked together into one single image. It might seem obvious, but commonly in 3D renders we try to show off as much as possible in all images we produce. But this can often muddy the focus of the image. If you mainly want to show off our amazing texture work on a prop, then highlighting everything might work. But there is also value in leaving things out to simplify the image and maybe add another camera angle to show off the things you left out. But that is also easier said than done when you’re super excited about your awesome project and you want to show as much as possible!
So I was always adjusting the rotation and strength of the base HDRI to get a good base lighting. Where the silhouette was emphasized and my subject area got some highlights onto it.
Afterword
This project was so much fun to do, and I think it really helped me to grow as an artist on all fronts. Daring to try to create something yourself and not rely purely on something that already exists is hard, but it is very rewarding once you do it. It was also fun to try and find a nice balance in the textures that helped to sell my ideas and rendering them in a complementary way. Mainly, going away from this for me has meant daring to push myself outside of my comfort zone and take it slow and steady and plan/revise my workflows during the process. Even if nothing comes of it, you have still practiced and gained experience and that is the most important part.
Lastly, I want to thank Jonas Holmedal, Ludwig Lindstål, Jake Oliver, Andreas Ezelius and Mikael Hellberg for their feedback during the process.
If you want to see more of the bike, then you can head over to my Artstation! I am always trying to improve as an artist so if you have any questions or thoughts then I’d love to talk about it!
I hope you enjoyed this article and might have some use for it!
Theodor Gerhamn, 3D Artist