Shooting Knife

Prop Breakdown

Dmitry Bely


Dmitry Bely

Senior Weapons Artist


Hi, my name is Dmitry Bely. I'm a Senior Weapons Artist at Starbreeze. Previously, I worked at Sperasoft on the Battlefield 2042 project, as well as contributed to another major AAA title.

Currently, I'm pushing the boundaries of my creativity by exploring Concept Art and Art Direction. I enjoy experimenting with different pipelines, as it allows me to diversify my approach to graphics. After working in the industry for years, it's important to keep things interesting. Additionally, it provides me with valuable experience in organizing production processes.


In this article, I will tell you how many details were removed during the creation of the NRS-1 project, and I will also try to share some thoughts about art and the industry.

My first thoughts about this project appeared in the spring.

I had a very productive year before that. I was active on the project in the studio, engaged in mentoring, built up my portfolio, experimented with the art direction, and took steps in concept design.


In general, everything was going well. However, such an intensive journey exhausted me a little, and I decided to take on a project “for recreation.” Subsequently, this task failed to achieve its intended purpose, but it did lead to a partial reboot.

It is worth noting that attempting to combine two work options (having fun and enjoying the process, while polishing everything for a portfolio) is not the best idea. Now it seems to me that it makes sense to work on a series of small projects.

To be honest, it is likely that I won’t succeed in stopping there; I will definitely want to do a little more, add a couple of details, and polish everything to perfection. Consequently, the project will inevitably grow.

However, in theory, this is the most effective way to grow and try new things. You won’t get tired or burn out on a small project. Therefore, it is better to move on to new ideas early, acquire new skills, and so on.


Based on this, I have a couple of thoughts about my attitude toward personal projects.

If your goal is learning, refer back to what was written earlier.

If you want to enhance the media component of your portfolio, be more active online. Write blogs and share work-in-progress (WIP) updates. The task is to increase your visibility and be interesting in the present moment.

It’s worth considering that if you do something solely to train in something completely new, don’t rush to get very upset if you don’t become instantly popular. Until you reach a certain threshold, your work may not attract much attention from the audience.

Yes, I’ve seen examples of this too. But don’t be deceived. Likes from friends won’t make you an expert. Focus on developing your skills.

Perhaps your goal is to join a studio and earn money with your art. Studios don’t look at the number of likes. Professionals evaluate the quality of your work. To determine this benchmark, the easiest way is to reach out to people on LinkedIn who work at your dream studio. Ask them for feedback on your portfolio and what areas you need to improve.

References & Inspiration

No references, no project. Yes, I understand that you can find images of everything you want on the Internet, but not every image will serve as a good reference. Therefore, I immediately rejected a couple of ideas before even starting the blender.

It’s like making a movie. If the original script is bad, no matter how hard you try during production, the audience will recognize its poor quality. So make sure your references effectively showcase all the aspects you will focus on throughout the entire production. Pay close attention to references for modeling and texturing, especially texturing.

Why did I choose this particular knife? Firstly, as I mentioned earlier, the goal was to create a small project that would take a couple of evenings. It didn’t work out that way, but I didn’t know it at that moment.

Some time ago, I was practicing modeling in Fusion 360, and one of my exercises involved different types of knives. That’s when I compiled a list of visually interesting knives.


Then I narrowed down the list even further, with the main criterion being to find an unusual object that can be described in a couple of words and still sound cool. Let’s face it, “shooting knife” sounds much cooler than a folding knife.

The choice was made, and now came the final test. Yes or no. And the answer was yes. Some genius on the Internet had taken great photos, and at that point, it became clear that textures would be involved.



I watched disassembly videos to understand how it all works. Initially, I had a lot of fun, and I wanted to model every single bolt in my project. Sometimes I see such works on ArtStation, and I feel envious of the authors.


But I suppose I’m not that meticulous. I want to focus more on the artistic aspect rather than the accuracy and detail of the modeling. However, traces of my original idea can still be seen in the final renderings.

I have screenshots of the disassembled state as well. I also decided against an extensive study of the internals; it was just a matter of a couple of bolts. Modeling them wouldn’t have been difficult, but the main problem would have been the countless checks and reconciliations with the references. It’s not worth it.


I started with the bullet cartridge as the first object; it served as a teaser for everything else.
Beginning with a cartridge is a classic practice in weapon art because it allows for certain size references, as the dimensions of the bullet itself are well-known.


I decided to omit the black rubber part as it became a bit boring. I didn’t include it in the project.

At this point, it became clear that another player, the one that falls under “optimization,” was being eliminated.
I’m starting to feel like a rushed publisher. Hahaha, evil laughter.


High Poly

I used ZBrush for this. I took everything from the mid-poly stage and smoothed the model to achieve chamfers and damage for the knife.


I also sculpted small details. I modeled part of the damage in advance on the mid poly geometry and only then polished the shape in ZBrush. Here, the question is a bit controversial as to whether it was worth doing it.

One advantage is that it allows for a more accurate coincidence between the sculpture and low poly. I was fortunate to have good references; without them, achieving a cool result would have been extremely difficult. In my case, I knew exactly what to do and what to ignore, which is equally important.


One of the micro-challenges for me was creating the diamonds on the grip. I wanted to make them on the high poly model so that I could also show them on the low poly version and bake them while preserving the silhouette.

The idea of using 3D noise in ZBrush and assigning it to a mesh immediately came to mind.

However, it turned out to be not so easy to do everything accurately through this pipeline. Additionally, in order for the small diamonds to look accurate enough, I had to use a giant polycount, which greatly slowed down the geometry editing process. In my opinion, this approach is counterproductive.


There is one point here that I am most proud of generating low poly rhombuses.

This involved using a very small decimate value, which resulted in a result very close to low poly. The rest was just coming and cleaning up. It required a lot of manual work.

Next time, I will be more careful with mesh preparation. In the meantime, I spent a couple of evenings cleaning it up.

The work was not difficult; I listened to music, reflected a bit on life, and eased the intensity of work. I don’t want to do this anymore; I need to clean up major issues in advance before using the booleans. Otherwise, it gets messy. I’m used to doing this for most mid-poly models, but somehow my brain turned off for this one.


UVing & Baking

I use Rizom in conjunction with Blender.

Rizom + Blender is a convenient combination, especially if you install a bridge. I’m using this one. Here is a link to Gumroad.


When working on weapons, I always start with the hard edges. My task is to achieve correct shading on the low poly model, which is no less important than shading on the high poly.

Sometimes it’s easy to overlook this step, especially when using 4k textures and adjusting the weapon to a high texel density.

This approach often works for portfolios, but in production, it can cause issues during optimization. All these mistakes become apparent during LoDing and mipmapping.

There is nothing outstanding in the baking settings. I use all the standard settings and practices, as in any production pipeline. Baking by name.


I also test other baking programs, and sometimes they have significant advantages and produce better quality. However, the simplicity, speed, and convenience of working with Marmoset cannot be overstated. So far, this program remains a solid favorite for production.

I used baking without dithering and with large files initially. Everything was fine at first, with 16 or 32-bit files yielding good results.

However, the files became too heavy later on, making it impossible to work smoothly in the painter due to lags. This was something I didn’t consider at that moment.


Materials & Texturing

For example, the vector mask was useful. In general, it’s a valuable skill to be able to create your own vector masks for logos and marks.

Their quality is higher compared to converting from photos. Additionally, it is often challenging to find high-resolution photos from certain angles, and that’s where Illustrator helped me. Moreover, understanding how to construct forms, even in 2D, enhances your predictive analytical abilities.

This understanding improved when I studied CAD systems, which advanced my comprehension and accuracy in classical polygonal modeling.


We utilize materials from Megascans, but it’s not as straightforward as it seems. Frequently, what appears good on the website might not work well for your project. For instance, I downloaded a set of materials for my project but ended up using different ones.

Furthermore, don’t forget that you can always make slight adjustments to brightness or color within the painter software. Therefore, it’s important to consider these factors when searching for and selecting materials on the website.


Regarding details, this is an extremely crucial stage that can be time-consuming and relatively slow. Often, the most challenging aspect is the lack of good grunge and alphas, especially alphas.

Therefore, if you come across a successful pack while browsing the Internet, save it immediately. Believe me, when you urgently need something specific, it can be difficult to find it. Murphy’s Law tends to come into play.

Don’t blame yourself if you can’t immediately create a million interesting details. If you encounter this problem, I usually recommend the following trick: set a goal to create 3 to 4 small unique parts per day.

It’s incredibly easy and fast, and it can really kickstart your progress. Within a week, you’ll have a substantial collection of interesting assets with dozens of unique elements. Make good use of it.


I have a rule: I rarely use a layer where only one channel is included. I always strive to incorporate all the details, such as roughness and base color, in the same layer.

Roughness and base colors are important.

It’s crucial to work with layers in a way that opening only one channel allows you to determine the object and material in front of you accurately. If you see simple fills or a mundane tiled texture without unique details, it’s a signal that you need to further work on the texture because it’s not yet ready for the final texturing stage.


Now, a few words about metallic and its characteristics. I believe most people are familiar with it, but I’ll clarify it once again. You can use metallic values other than 0 and 1. For instance, anodized metals or metals with impurities.

Additionally, the upper layers influence this channel. For example, dust, depending on its density, should subtract values from the metallic channel. You can also utilize gray metal.


personally created some alphas myself, specifically working on important elements like the logo. I developed them from scratch in vector format to ensure high image quality.


I would like to highlight the use of additional sources. I used a Megascans asset for the background, as well as for the base of some basic properties such as dust and basic grunge for plastic.

Using scanned elements or photographed details adds realism to the asset. However, don’t expect that you can simply apply downloaded textures to the model and achieve a good result. If you follow this approach, you’ll end up repeating the classic mistake of beginners who rely solely on smart materials without manual fine-tuning.

Experiments with scenes and 3D objects can be used to create an interesting background. It’s easy to experiment with them, and if one doesn’t fit, you can download another and conduct a new test.

There’s no need to spend weeks creating assets.


An important point to consider is that attention should not be drawn away from the main object.

Your portfolio should showcase your skills rather than how well-scanned objects from the Quixel team are.

Shaders & Technical

Regarding transferring the view from the painter to the marmoset, I also experimented with log (lut correction) in this project. Overall, I did it, but it made the job a little more difficult. A lot of time is spent on fitting since it’s specific to a particular render engine.

Additionally, you need to adjust the light and camera parameters uniquely. That’s why I stopped doing it for now. Instead, I adhere to the rule of making textures look good in the default painter scene to avoid rendering issues in other software.

Polish & Final pass

I like to do a little bit of polishing in Camera Raw for post-processing. It’s a relatively easy way to add a touch of photorealism to the render. I can even assign quick presets and make adjustments as needed.

However, I may have been a bit too enthusiastic with the flower picker as it appears oversaturated in some areas. I think the eye is also blurred. Now, as I compare the rendered and processed images, I find that I prefer the simplicity of the RAW version even more.

The new Marmoset renderer has more settings for the ranger, such as separate control of light and shadows.
I found this feature very convenient and easy to achieve good results. In my latest project, I didn’t use external post-processing at all, yet the result is more than satisfactory for me.



One drawback of setting up the light in Marmoset is the limitations when working with HDRI. It’s not possible to rotate the HDRI on the second axis, which is sometimes necessary.

To avoid making numerous changes to the scene, I put everything in a folder and rotate it along the horizontal axis. Visually, the scene remains the same, but the HDR changes its position relative to the entire scene.


And of course, I always turn on Ray Tracing since it significantly affects the visual effect of the scene and the behavior of light. Therefore, I adjust the light considering this technology from the beginning.



In general, the rendering process can be lengthy and complex. This is mainly due to the need for numerous experiments and the search for successful solutions, such as finding the right camera position.


I also want to mention that I worked on two separate large render projects. One of them involved cartridges in a jar, primarily to relieve me during the main project’s development. At that moment, I took a unique bullet for this knife and created a teaser.

It’s funny because it’s nearly impossible to guess which weapon will be in the final project. Shooting knives didn’t become an ordinary thing, after all.


Frame format

For this project, I decided to use a 1 to 1 frame format. I enjoyed working with this aspect ratio. The only problem is that when I maximize the views, each frame needs to be scrolled independently to view it completely, regardless of the monitor size.

As a result, the project isn’t very quick to read. I received this feedback from my colleagues.



I worked with the background, generating quick folds in Marvelous and using fabric material from Quicksilver Megascans.

Another thing I want to mention is a detail that I didn’t work out extensively, and on the renderers, it appears scattered in the background.


I really like horizontally elongated renders. You can come across them in every project in one way or another. I appreciate them because they don’t conform to standard ratios and provide a fresh look.

However, embedding them beautifully on websites can be challenging, so I usually create only one frame per project.



Initially, my goal was to work on a quick and small project to unwind and relax. But as is often the case, the project started off well and immediately inspired me. I felt the urge to experiment and create something super cool. I dove back into it.

At some point, I realized that I needed to reconsider the project’s importance (lowered its significance in my eyes) and removed some of the content. I didn’t focus too much on refining the forms and decided against adding a few extra elements in the texturing process.

This project couldn’t be brought to a satisfactory state, and I simply didn’t present it. All you can see from it now is a blurred image in the background. Nonetheless, I still love it.

It’s better to finish a good project than to have an unfinished one, even if it’s not ideal.


I wanted to showcase more mistakes because we’re all human. Art shouldn’t be confined to Instagram. It helps us open up and prevents us from isolating ourselves in our comfortable little worlds.

It’s crucial not to let our egos get in the way. Supporting and leading the team at work and in the company is of utmost importance to me. I also engage in mentoring because I believe it’s a critical factor in motivating young professionals. After all, we will work together hand in hand on future projects.


I hope you found something interesting in this article. Feel free to subscribe to my ArtStation for updates on new projects.