LogInKristina Papakina talked about the Millers Falls Hand Drill No. 980 project, explaining how she textured the prop to replicate real-life materials and how she worked with wear and tear.
Introduction
Hello, dear 80 Level community! My name is Kristina Papakina, and I'm a Game Graphics Production (GGP) student at Howest Digital Arts & Entertainment (DAE). I have just started my second year there and am very excited to write this interview!
My 3D journey started last September, when I enrolled in the university. However, I've been in the art field a bit longer, as my previous degree was in digital design and multimedia. I also worked as a Graphic Designer for a couple of years, which helped me develop good observational skills and decent time management.
As a long-time gamer, I've always looked at the industry from a consumer perspective, but I really wanted to get the backstage of it and see how it's done. So when I saw the GGP major, there was no point in waiting any longer.
Millers Falls Hand Drill No. 980
This project was done as a part of the Game Asset Pipeline class at the university. The goal was to texture a Millers Falls drill, utilizing Physically-Based Rendering (PBR) techniques in Substance 3D Painter.
Since we were given a low-poly and a high-poly model, our main focus was to texture it as realistically as possible. And while it was the main task, we still had to find the mistakes in the low-poly, fix and document them, unwrap and bake the model beforehand.
I had two goals in mind starting this project: understanding proper unwrapping methods to get a better bake and to push my texturing skills as much as I could. Thanks to my friends and teachers' feedback, I was able to achieve both of them.
After I figured out the exact model of the drill, I started gathering references online and analyzing them. The main goal there was to get a good visual library of differently worn-out drills to get as much information as possible.
Unwrapping & Baking
Moving in parallel with gathering references, I started to unwrap my model. Since it was my first time baking an asset, I wanted to just start with something and see the result in Substance 3D Painter. A good bake means good unwraps, and good unwraps mean close to no stretching, equal texel density, and proper padding between the UV islands.
Initially, I started to unwrap the model in Blender, but then switched to RizomUV as it gave me more control over each island and faster straightening of curved UVs. I tried to follow a rule that whenever there's an angle close to 90 degrees, there should be a seam.
There were, however, several places where I had to add extra seams to lessen the stretching or break the rule of applying a sharp edge on parts that had close to a 90-degree angle. For example, it was necessary to cut off these curved parts (selected in blue) from the main shape to straighten the islands properly and optimize texture space usage.
I didn't apply a sharp edge on a selected seam (see image below), because it was creating artifacts there. But not placing it there meant I'd have more stretching at the back part.
Another big question was which parts to overlap. Since I could only have one 2k texture set, I had to overlap some parts to save texture space. But I couldn't overlap a lot since this was considered a hero prop, and it would be very noticeable. Here are the parts that were overlapped:
The logic I had in mind was: which parts would be least affected by AO and least visible to a user? In the case of the gear at the front, I could fully overlap it since it was hidden from both sides and almost invisible to a player. Unwrapping took some trial and error to turn out well, but here's how the final seams look in RizomUV:
Which resulted in properly baked normals:
Texturing
Since I was all set with the bake, it was time to start texturing. My personal preference here was to pick a main reference photo and try to replicate materials from it as closely as possible. It helped me to keep the wear of the prop consistent and logical, without thinking about hows and whys more than I needed. Here are my main reference images:
Here's how it turned out:
And here's how it was made. Since I already researched the types of base materials that the drill had, it was time to set them up. I needed 2 metal materials and 1 wooden one made from scratch. But let's talk about metals first.
They gave me the hardest time not only because I was aiming for photo-realism, but also because I didn't understand how to make metal look "real". Steel was the first material I started with, as it was used on the front parts of the drill. After I put the base on, added some anisotropic noise, and threw in some scratches, my material looked like this:
It was too glossy, flat-looking, had way too intense height information, and the pattern was unreadable from a distance. I got all of the pieces of the puzzle, but it was just not coming together. So I went back to research and looked for feedback.
Thanks to my friend Bohdan Ishchuk (who is also a great Weapon Artist working at Ulysses Graphics), I learned a lot about texturing metals and how they interact with different materials, such as paint, wood, rust, dust, and damage in general.
The first fix was adding color variation to make the metal seem more realistic. You can see in the video below how much it enhanced the look. All of the color values were very low, sometimes had a subtle roughness variation or height information, which, when stacked all together, gave the final look.
The next part was fixing how the pattern looked and read on a metallic surface. Here, even if it meant to deviate from the reference, it was better to increase the size and decrease the height values. All indents had increased roughness, since they gathered more dust and were less exposed to external damage.
The final part was the wear and damage layer of detail. There, I added some anisotropic noise, roughness variation, surface scratches, and a bit more color variation to make all of the layers pop.
The next step was malleable iron material. This type of metal was also tricky to replicate as it was naturally noisy and had a lot of subtle details. The layer structure here was less complicated, though.
I threw in a grainy material as a base, tweaked it so that rough spots were larger, then added colour variation. Analysing the reference, I spotted minor bumps as well as an extra layer of rough spots and dirt, so it was all combined. You can see my main reference image and the final result below.
The only metal parts remaining were a drill bit and a handle. Those materials were made as a variation of the main steel one with more roughness on the handle and more anisotropic noise on the drill bit. They were both different in tone, so the overall base metal material started to look like this:
And the last part was texturing the wooden base. After a couple of unsuccessful tries to use a base wood material from Substance 3D Painter, I decided to build my own. During this step, I followed a tutorial from Jason Ord that explains in great detail how to build the base materials for your models.
It helped me create a material from an image and extract height information from it, as well as add subtle details that made it look realistic enough for the next layer of texturing.
Looking at my main reference, I started to build up a layer of varnish. Not every detail was clear, so I looked more at other references here, trying to find how the material was worn out and figure out a middle point between all of them.
The general logic was that roughness increased greatly at the spots that were touched the most and decreased towards the corners. It was also reflected on a base color, along with minor scratches and bumps.
I was trying to replicate layers of wax added over time by adding them the same way, layer by layer. This allowed all of the grunges to mix nicely and transition well at the right spots.
The last step of texturing was adding paint and the overall history of the prop. At first, I tried to replicate peeling paint with grunge maps, but they just did not give me the right result. It felt too generic and boring. So the only way out was to peel the paint by hand.
The same was done for the black paint, but I ended up using some extra grunges to break the pattern even more, as it was a larger detail.
Continuing to work with paint layers on the prop, I moved on to the back handle and added some there too. The most important thing here was to spread paint splatters all over the handle, but still keep it clean and looking good.
Analysing the main reference, I split all the paint into several layers and tried to replicate the way that they were placed on a real handle. In order to make the paint feel thick and with multiple layers, the main trick here was to add a very small height difference to each layer so it plays nicely in the light.
I had to remove some of the dirt spots as they were making the handle look too dirty and were drawing attention from the drill itself. The final touch was to add some fresh damage to the wood with a light color, increased roughness, and height difference. I repeated the same process for the other two handles.
Another thing that added realism and extra detail to the prop was a Millers Falls logo and model stamp on a handle. I created alphas in Adobe Illustrator because I couldn't find the same logo and letters on the internet. Feel free to use them if needed.
Rendering
This project was rendered in Substance 3D Painter using Studio Tomoco as an environment map. I set the environment exposure to 0.03, aligned the environment to the camera, and used a clear color as a dome. I also used the sRGB color profile. Camera settings and post effects are displayed in the image below.
To check if my renders were good, I turned them into grayscale to see the values properly. It helped me to spot some over-exposed areas and see if the drill looked separate from the background. I didn't use any colour correction and tried to avoid using too many filters to showcase how the materials look in real-time more accurately.
Conclusion
This project was a huge roller-coaster, and I enjoyed every bit of it. I learned so many things about texturing, but that was just scratching the surface. The main challenges I faced were connected to texturing metals and making them look readable. It was fascinating to learn how setting very low-opacity layers with minor details can combine into a realistic wear on an object and tell a story.
If I could give a piece of advice to myself during the project, it would be to research the materials better and look at other artists' similar projects to see how they build up their reference boards, how they approach texturing, and how they set up their presentation. It is all about learning and observing. It's a skill that needs to be trained as much as any other, and I believe it is very important to every artist.
Looking back at it now, I can see so many things that could've been done better. However, I am very happy with how it turned out and am incredibly thankful to all the people who gave me feedback, and especially to Emma Collins and 80 Level for allowing me to share this breakdown!