Night Sight

Prop Breakdown

Alexander Petukhov


Alexander Petukhov

Lead Weapon Artist


Greetings! I'm Alexander Petukhov, Lead 3D Weapon/Vehicle Artist at Digital Forms.


Initially, I wanted (and still want!) to create an experimental wheeled self-propelled gun. However, I realized that starting with smaller components is crucial for larger projects.

I began by crafting individual parts that stand out even when detached from the main object. I chose various targeting devices as they are the most logically separable parts.

The night sight APN-6-40, like almost any military equipment, is simplistic in form yet possesses a certain charm. It wasn’t designed for aesthetics; rather, it must function effectively in harsh conditions with simple controls and be assembled robustly to withstand any damage.

This sparked a particular interest: how to make the sight visually appealing despite its form and the complexity of its components?


I utilize Blender (UV PackMaster, EasyRef), Marmoset Toolbag 4, Adobe Substance Painter and Adobe Photoshop.


When working with references, I use PureRef and always organize photographs based on the sides of the object – all images of the left side in one column, the right side in another, and so forth.

For detailing specific parts, I prefer duplicating general views and cropping them, retaining only what is necessary. However, the challenge with this sight lies in its scarcity online.

Despite investing significant time in research, I couldn’t find dimensions or blueprints, which somewhat complicated the creation process as I aimed for precision.



Since I couldn’t find any specific dimensions, I proceeded as follows:

Initially, I roughly sketched out the general details and refined the proportions by eye and then…

Camera Alignment

This is a meticulous task that demands time and patience to achieve accuracy in modeling. The theory sounds straightforward: if we have a photograph of an object, it was captured by a camera with a certain lens at a specific distance.

Most 3D modeling packages offer various lens settings and parameters, theoretically allowing us to replicate real-world camera settings. However, in practice, photographs are often heavily cropped or lack EXIF data, necessitating manual adjustment of virtual camera parameters.


I must admit, this is one of my least favorite tasks, especially when it’s work-related, as it consumes considerable time and doesn’t always guarantee 100% accuracy.

Yet, after years of practice, I’ve grown accustomed to it. Returning to the task at hand, what do we need? It’s possible to perform this task in any up-to-date 3D modeling package; the key lies in convenience.

I work in Blender, and I’ll describe the process for it. We require the EasyRef plugin, which greatly simplifies virtual camera manipulation (instructions are available on the manufacturer’s website).

Additionally, we need to switch the camera control to trackball mode, allowing us to rotate the camera freely along all three axes.

Now, let’s proceed – create multiple cameras with background images (uncropped) and begin positioning them around our object. To achieve reliable results, we must set up at least three cameras to compare details from different perspectives.

It’s advisable to select perspectives that are as diverse as possible. Knowing the object’s dimensions significantly streamlines and accelerates this process, as the initial mock-up already has the necessary boundaries, facilitating camera positioning.

In my case, lacking dimensions led to instances where I had to adjust the size of different parts of the sight back and forth.


The modeling approach follows the standard pipeline – starting with the high-poly model and then transitioning to the low-poly version.


For the high-poly model, I opted for a Subdiv approach, which I find meditative and akin to solving a spatial puzzle.
For instance, in this project, around 75% or more comprised a single integrated detail!

It made the process particularly engaging.


When modeling in this manner, the key is to assign the necessary number of edges to all rounded and cylindrical details upfront, allowing for a relatively swift transition to the final low-poly model upon high-poly completion.

I didn’t impose any restrictions on the polycount since this project was personal.

However, I wouldn’t say the model turned out excessively detailed. I believe newer projects demand far more detail for similar objects.

UV & Baking

I divided the object into two materials and, accordingly, two UV sets – one for all opaque parts and the other for lenses and protective glass.
I increased shell sizes where inscriptions were present and reduced them in inconspicuous areas, such as the inner part of the tube and eyepiece.

Due to the asymmetry of the model, significant overlaps weren’t notable, except for the internal parts of the tube and eyepiece.

Additionally, the chain consists of only one link. For baking AO, normal, and curvature maps, I utilized Marmoset Toolbag 4.

Texturing In texturing, I aimed to emphasize three aspects – inscriptions, paint irregularities, and dustiness.
Working with inscriptions was amusing; I didn’t want to search for specific fonts, create alphas in Photoshop, etc.

I found that one of Substance Painter’s default fonts closely resembled what I needed, albeit lacking support for Russian characters.

Fortunately, many letters are similar in both languages, so I ended up typing amusing phrases that matched in letter count and spacing.


Subsequently, armed with a round brush and mouse, I made adjustments where necessary.

I can’t say it was quicker than finding the right font, but it was certainly more enjoyable.


Paint irregularities aim to break the perfection of the sight’s form. References show interesting irregularities at corners and recesses.

I created several layers with strong height values and used a soft brush to draw these irregularities.
Additionally, various technical openings, screws, and the like fall into this category.


Dustiness has the most significant impact on the object’s appearance; by correctly adding dust to corners and generously sprinkling it on top, the model gains volume.

Regarding damage, I followed the following logic: this night artillery sight is expensive and complex to manufacture. It’s not just a system comprising large lenses; it’s also a sophisticated electronic device.

Considering its unpretentious exterior, internally, it’s precise and intricate. Such a sight receives the utmost care, making it challenging to envision huge scratches, dents, or other signs of rough wear and tear.

Thus, I limited the damage to areas where the paint was worn down to the primer and scattered scratches.


I painted using metal/roughness, allowing me to focus on just two maps for 99% of the drawing time.

The primary goal was to ensure that details were visible on the object from any angle and under any lighting conditions.

Important Note

From the beginning of the work, I decided to avoid close-up shots – I didn’t want to delve into micro-details and perfect material rendering, as it’s very time-consuming, and I have had enough of such tasks at work.

I was interested in the entire scope and wanted to make it appealing from a general perspective, as there are hardly any small interesting details on it.

So, here’s a useful tip – determine in advance for which angles and views you will be painting your model – this can help reduce working time without sacrificing the quality of the final appearance.

Rendering & Lighting

For rendering, I chose Marmoset 4, which allows for very fast and convenient work. Tone Mapping – ACES. To emphasize the size of the object and the realism of its materials, 3D scans of real objects are very useful.

On the one hand, during the texturing process, you can compare them and immediately see where your model looks duller, and on the other hand, they deceive the viewer well during viewing, making the model look more realistic.


I found a great table on the Megascans website, which was painted with red paint – the paint was peeling and cracked, and I placed the scope on it.

Since I didn’t want to think about the realistic environment for this “still life,” I decided to take photos from above.


The red paint contrasts beautifully with the green scope, but it turned out that the table attracted too much attention – the focus should be on my prop, not on the excellent scan.

Here, Photoshop came to our aid. First, I created a mask for the table and applied various levels, contrasts, and saturation to it until I got the desired result – now, at a glance at the picture, attention is drawn to the scope, not the table.


To make the model respond better to the HDRI map, it’s worth adjusting the Reflection Intensity parameter in the render settings.


For convenience in working with different angles, lighting, and the position of the model, you can do the following – for each new angle, create a new camera and model.

You can also add lighting sources according to your preference (some views need additional lighting).

In Marmoset, you can even have several Sky objects with different HDRI maps, so you can quickly change the environment.


For those angles where the scope is just against the background, I did the following – a circular gradient, added glow around the scope contour, slightly improved sharpness with the unsharp filter, and then applied noise to prevent gradients from deteriorating when compressed on ArtStation.


I assembled the video from individual frames with a transparent background, and then I added the same circular gradient with noise applied to it as the background.


I hope you found both my work and this article engaging.

Moreover, I trust that the insights I’ve shared will prove valuable to you in your future endeavors. Here’s to creativity, learning, and endless inspiration ahead!