Making a Gun From Mass Effect

Hi, my name is Thomas Datoussaid and I’m a 3D artist on The Сrew 2 since June 2017. I dived into the 3d world with Morrowind, as I spent hours to create mods, and wanted to go further and make my own models. That completely failed, but I knew this was exactly what I wanted to do later. Therefore I entered in HEAJ (Haute Ecole Albert Jacquard, a CG school) in Belgium, and graduated in 2016. Our last graduation project was MHL, a ship racing game, and our team had the chance to present the project at Gamescom 2016, and earlier in Scotland with Abertay university at the “Dare ProtoPlay – Dare to be Digital “.

Since I graduated, I work on my free time as a self-taught artist to practice mainly hard surface, learn new techniques and software.

 

At the very beginning, I wanted to enhance my realistic texturing and rendering skills, as they are by far my weakest points. So I merely choose something that I really like, a concept from the Mass Effect games. My goal in the industry is to become a weapon/hard surface artist, so I also need to learn reading et understanding a 2d concept.

 

This concept was perfect; Superb shapes but quite an easy modeling, so I could spend all my time on the textures, different materials, and flat surfaces to make interesting.

 

 

Well, as I had a concept, with proportions, colors, etc, all I did was creating the basic shapes of the model.

This step is mostly to answer some questions about the thickness of the gun, and how mechanical parts work with the others.

 

So I fitted the original concept as close as I could and prepared the different pieces (separated by a line) to help me later during the high poly stage. Even if the concept of Mass effect was planned for a 3rd person game, I knew I wanted to add some elements of an FPS game model, but later in the modeling stage: a safety catch, a firemode button, an iron sight, ammo & magazine.

 

 

I worked with 3ds MAX, and used the traditional sub-d modeling & quad chamfer, depending on the piece. Working with the quad chamfer technique is quite fast, non-destructive, and allows to quickly have a base model to create the low poly mesh.

 

Other than that, nothing fancy, really. As I prepared the carving stuff/lines as separate objects during the previous stage, the only thing I had to do was adding some geometry for the fine details. I also used floating geometry for a few small elements ( caps on the flat area).

 

It should probably be smarter and faster to do all the surface normal detail within Substance Painter though, but I really like modeling and the shapes remain simple to do.

 

The entire texturing is done with Substance Painter. The key here was finding good references to materials in reality. See what works, what doesn’t, what could I put in the albedo, in the roughness, etc. I took some pictures on the internet, and as a friend suggested later, I also looked around me: my mouse and keyboard, the desk, etc. The simplest object becomes a source of interest when it comes to the texture study.

 

The process consisted of splitting the texture by similar materials.

 

Plastics are all different, with unique damages, various imperfections, and colors, but they share similar base layers to construct them. That allows me to borrow layers or groups that work well to another, with a simple tweak in roughness/color/height value.

 

I wanted the grip and the main body with a different feeling of touch, a quite rough with apparent grease here and there, and a glossier and smoother one with another type of imperfections. Something that worked well on the plastic was to add some noise patterns in the rough only. That adds this nice rough-looking grain on it.

 

Same goes for the metal. As the instance, the model has only a few very different metals. But the details on each piece have been worked separately to gain variety in the model. The challenge was to have enough variations between each piece and materials though. And once the polishing with fingerprints, texts, grease, etc was done, the difference was even more pronounced. But having strong base layers is mandatory to define the primary aspect of what you want to obtain.

 

Another thing to keep in mind that’s important, is using the generator smartly. A generator is merely that: it will generate a mask according to your AO, cuvature, etc …and that’s it! But in reality, that’s not the way it works. Wears and tears, scratched, rough and color variations, each detail is due to something specifically, not randomly placed. That’s why I used a lot custom brushes and masks on top of existing smart masks.

 

And finally, a really important thing to improve the texturing is feedback (And I think overall skill in general). I asked friends in the industry that give a fresh eye on the model & textures. That helped me a lot improving my work and my art style.

 

I had the chance to add lights with the only goal to present the model within Marmoset. It’s not like a scenery with gameplay where the lights have an important role which is leading the player through the level.

 

So everything here is to reveal and showcasing a part of my work. I’m not the best at this, but I often see a lot of images with no interest, the model is lit with a simple directional light, no citation, no context, nothing except a lit model.

 

That’s why I use a few lights on top of the sky system of Marmoset. I always add a directional light and a backlight first. For this specific piece, I also used quite a lot spotlights to add visual interest on parts that wouldn’t be seen otherwise. I had to pay attention to catch the eye in a specific part, leading to the rest of the image.

 

And I don’t add too many lights or nothing will pop out and everything will be lost in an ocean of information. This was probably a more difficult and time-consuming task that I thought, but in the end, the result is pretty good.

 

 

The same rules for the post process effects. Whatever I do, once again, is to present my model in a final image.  So I threw my renders in Photoshop. I added some contrast, eye-catching details like the glow/dusty effect. The emissive in the 3d render was kinda weak. A soft brush and overlay mode did the trick.

 

 

I never had the chance to work on assets like this in a game production. The way I create an asset for a folio purpose is not the same for a game with specific constraints. On one hand, the 3d model is thought to be used in video games. Only a bit less than 10k tri, mag included. (The bullet wasn’t planned at first, so I went a bit crazy for it). On the other hand, the texture size is 4k, and I would probably use more symmetry in the UV to gain some space.

 

In an FPS game, you usually see only one and always the same side of the model at a time.

 

Therefore, using decals for the text on an actual decal sheet seems smart,  to use them on other assets without losing too much resolution and avoid flipped texts, if that matters of course (considering it’s from Mass Effect with a wide range of weapons from the same manufacturer).

 

I could go further an think about how modularity with the several parts of the model could be done : Grip, stock, etc. Last but not least, I would export masks to change the colors in the game engine rather than a color scheme for one albedo at once, and pack all my textures according to the game engine.

 

Thanks for reading, and I hope it was helpful or at the very least, interesting.