Nikolay Marinov has returned to show us his latest project – a body pile material made in 3ds Max, Houdini, and Substance and inspired by the Doom series.
My name is Nikolay Marinov and I’m currently 23 years old. Since our last talk, I received my Bachelor's degree in Computer Engineering from the University of Ruse and took the next steps of my career by joining Frontier Developments as a mid-level Material Artist where I currently work on the upcoming “Jurassic World Evolution 2” and an official Formula 1 project.
Body Pile Material
My current goals are to delve deeper into technical art as a whole and bring my portfolio to a satisfying level of consistency. To do that, I have to take on challenges that make me go out of my comfort zone, in order to improve and broaden my skill set.
That is exactly what happened when I stumbled on the work of Austin Cline. I had no idea how he managed to achieve that result, so I took it upon myself to reverse-engineer the process in order to learn new techniques and pipelines. I think that when it comes to production – materials, good level design and lighting help bring everything together, so I think it can be a good thing to push the limits of those mediums.
The project definitely had me scratching my head for a while, as it took a lot of trial and error to get the desired results. I met difficulties and frustration at every step of the way and I even had to start from scratch. I started by breaking everything into different stages.
The first stage was the base mesh. I downloaded a mesh from the web that had the characteristics I was after. I then used a character base mesh provided by my friend Kaloyan Kalamov to wrap some good topology onto the initial mesh inside of ZBrush
The second stage was rigging. As I have no rigging experience, I used Mixamo to generate a rig on my first attempt, which led to many issues down the line. On my second attempt, I learned about a tool called KineFX inside of Houdini. It’s not very well documented but it provided me with the tools I needed to produce a decent rig with weight painting.
The third stage was posing the rig. My lead at Frontier advised me that it's always better to not leave everything to the physics engine when doing any kind of simulation. That’s why I brought the rig inside of 3ds Max and posed it the way I wanted it to be simulated instead of starting the simulation with a T-posed rig. I then repeated the process until I had seven differently posed rigs.
The fourth stage was setting the Agent Clips inside of Houdini and configuring the joints for each of the rigs. Then use the Copy to Points node on a few to get the initial starting location of the simulation. I then generated some geometry to serve as collision constraints and to break up to falling pattern inside of the RDB dopnet. I added my Agent Clip to the crowdsolver inside of the dopnet. Everything after that was tweaking the settings to my liking and running the simulation a few times with the differently posed rigs until I had generated a couple of piles. I then used the Alembic ROP node to bring the results into 3ds Max.
The fifth stage involved adding bones and skulls to some of the bodies. I downloaded a skull asset from the web and some bones from Quixel Megascans. I then brought two of the rigs into Blender and used proportional editing to manipulate the geometry, so that the bones could stick out properly out of the flesh. I then brought them back into 3Ds Max.
The sixth stage was making everything tile. I made a 4x4 meter plane to serve as the low poly mesh and started assembling the different piles from stage four as well as the bodies from stage five. I then deleted every object that was sticking out of the bottom and right side of the low-poly plane. Then made a copy of the ones from the top and left side and translated it exactly 4 meters to make the whole thing tile. The last step here was making the ID Map – I put 3 different colors on different bodies for color variation and a separate one for the bones.
The seventh stage was baking and texturing. I discovered that the Displacement Bake in Substance Designer didn’t read Vertex Smoothing which resulted in faceted results similar to ZBrush. The solution was to have a heavily tessellated mesh before the bake. As for the texturing part, the baking inside of Substance Painter was relatively straightforward.
The eighth stage was bringing everything together inside of Substance Designer. The most important step there was height modulation. I came up with a way to reduce displacement artifacts to a minimum with a combination of height blends, masks, and blurs. The next part was the ground surface and the blood puddles which are essentially the Water Level node but modified to look more like blood. The way I made the height modulation allowed me to use height blends to bury part of the bodies into the ground, based on some exposed parameters.
The final stage was rendering and presentation. I brought everything into Marmoset Toolbag 3. The first thing I did was to go to a high focal length and make use of the Lens Distortion it provides.
Someone told me that the final results resembled a sculpture by Auguste Rodin named “The Gates of Hell”. That led me thinking how I can convey that feeling even more. The solution was probably the easiest part of the project. I just downloaded a picture frame from Quixcel Megascans and it immediately gave it this renascence effect that I was going after.
Tweaking the Results
The texturing part was probably the most pleasant part of the project, most likely because it was the only part I felt comfortable with. For the color palette, I wanted to capture the aesthetic of how they would look if I would’ve to put them in Doom, without them looking too decayed. Because I was basically texturing a flat surface, the Curvature Map produced from the bake was my best friend here. I set up the base color and “fake” SubSurface Scattering and proceeded to derive all of my detail from the Curvature Map, including “faking” muscle tendons and flaky skin.
The most intriguing part was making the wounds. I generated a mask from the Curvature and Ambient Occlusion of where the wounds would be located and then used a familiar technique from Substance Designer – I got a Gradient Dynamic with a high contrast directional grunge texture. I then slope blurred it, duplicated the layer, changed the settings, and divided it with the one underneath. For the custom variation, I hand-painted some wounds, dripping blood, and skin imperfections. I repeated the same steps for the bones and hand-painted some cracks onto them. I then exported the results and the ID Map into Substance Designer.
I believe figuring out the whole pipeline was the best part of this study, next to the final result. One of the key aspects in creating materials that stand out is trying new things and experimenting. I find that most of the aspiring artists on ArtStation copy the work of others and you end up seeing the same material in countless portfolios. Although my study is based on existing material, it is by far not a common one.
I think the best way for beginners to stand out from the crowd is to stop focusing so much on what others publish and focus on finding their own unique, distinguishable style. I am guilty of the same exact thing but I believe that I’ve now developed a unique gritty style that focuses on color palettes that guide the eye towards the detail. This is quite a slow process, so try your best to not get discouraged!
You may find these articles interesting