Our Character Art Team Lead at Plarium Kharkiv, spoke to us about the epic design process for his 3D model in Soldiers Inc: Mobile Warfare.

We were tasked with creating a unit for a near-future military setting, which meant cultivating a hi-tech theme without sliding into futuristic sci-fi. The idea was as follows: a soldier clad in exoskeleton armor equipped with mechanical elements that help lift heavy weights. This soldier should also possess a large automatic weapon, active armor (basically a shield), and a communications system. First of all, we gathered suitable reference items.

Once the reference list was ready, our Concept Art Team Lead drew several sketches. From these, we then chose version D.

Then we needed a more detailed image.

The sketch expressed the main idea, but it was still unclear what the character was wearing under their armor or how the elements were attached. Some of these elements I designed on my own, while our Concept Art Team Lead and I devised others collaboratively.

The model was to be used in .jpg format for static purposes in the game (such as the unit information window) and as promo material. Therefore, there were no polygon or texture related restrictions.

Fortunately I didn’t have to start modelling from scratch – I had previously made a human dummy especially for soldiers. The model was a sturdy, muscular guy, which I used to sketch general outfit shapes in ZBrush. Such an early 3D concept allowed me to see the character’s mass and proportions, as well as the way the armor fits.

I prefer modelling outfits layer by layer, in the order a person usually gets dressed. This approach results in the model looking realistic, with each element of the outfit in its correct place.

That is why I started by considering what he might wear under his armor. The first version I came up with turned out to be too futuristic. However, the second one was more what the setting required – it looked more like modern military apparel, but improved.

Then the long process of form selection started. Our Concept Art Team Lead and I walked this bumpy road side by side. We discussed our ideas, improved details, and tried different approaches. Corrections would often been sent as pictures or gifs.

As soon as we had approved an element, I reduced the number of polygons using Decimation Master, and exported the model to 3ds Max. After this, I modelled a clean version of the element on top of it. The only things I imported back to ZBrush were soft materials such as fabric, because I would model the folds by hand.

Working on separate parts of the model allowed me to avoid waiting for final approval on the design. In fact, we worked on the exoskeleton concept right up until the time for the final render came.

The upper pieces of armor caused no trouble. We prepared several versions overall, but we chose the final one reasonably quickly. As usual, I modelled some of the smaller details using geometry, such as zippers, tiny pins, and even some of the stitches and seams on the clothes. I find that this approach makes the rendered model more appealing.

The arms of the exoskeleton were probably the most complex elements of the armor. I designed both the inside and outside of these elements as I tried to figure out what the construction of the mechanism might be like, and how the different parts were supposed to act while working the wearer’s elbows, shoulders and wrists.

Mostly to save time, we finally attached the exoskeleton elements using modern orthotic methods.

The Unit’s gloves, as well as the main elements of armor, were modelled layer by layer: fabric, metal elements, hydraulics, then protective kevlar material.

The helmet took the longest to make. We split the tasks between us – our Concept Art Team Lead made the design-related changes, while I looked after the technical ones. The helmet’s shape is nothing new, but the devil was definitely in the detail. First, we needed to make sure that the helmet could be easily unlatched, and that it didn’t scrape the character’s face while. We also needed to consider navigation, communication, and ventilation systems in the design, which is why the front of the helmet ended up being more wedge-shaped.

Once the general form was approved we added optics, cameras, and an antenna to the helmet. These had actually been present in the initial concept, but we redesigned them slightly.

After this, we added a backpack to the model and ended up with the following result:

I used a ZBrush plugin named UV master for the UV-mapping. I didn’t actually upload the entire model at once, instead uploading it piece by piece. For instance, I mapped the leg and uploaded it, then did the same with the arm, and repeated the process until the entire model was in 3ds Max. The reason for this was to make further operations in Substance Painter much easier.

Appointing each object its own material makes Substance Painter create a separate Texture Set. It also helps you to hide the elements you don’t need and texture them separately. For example, I created some of the textures in the PBR Spec Gloss pipeline for Marmoset.

The model was created with the help of TurboSmooth. I uploaded a 3ds Max model into Substance, but used the details from the high-polygon model on ZBrush to bake it. All the details were then projected onto the low-polygon model and added via the Normal Map texture. These manipulations resulted in a mid-poly model with high-poly details.

Our Concept Art Team Lead had also developed a series of labels exclusively for this character. I added these to the armor in Substance as well.

The model was then rendered in Marmoset. I always open the model in Marmoset and appoint textures in parallel with Substance. I do things this way because if I draw some details in Substance and refresh the textures, Marmoset picks up the updates right away and you can see what it will look like as an end product. I used regular settings for this: HDRI map and Environment.

I always make sure that the character is able to move in their armor, so for this project, I made a rig in 3ds Max using the Biped module, configured the Skin modifier, posed the character and sent it to Marmoset as an .obj file.

After all of our work was complete, this is what the final version looked like:

In total, it took us a month and a half to develop the character. We spent so much time devising the concept that at one point, I felt we had completely lost our way. Yet fortunately, we made it. When you’re working on a project as exhausting and time-consuming as this one, you need to brace yourself and just keep working. As soon as some good results arrive, your motivation will return.