What a great and helpful guide you've written. Respect.
Just updated it!
Why didn't you share the URL to sign up in the article
Weapon artist Ben Leary has most recently finished his work for XCOM 2. He’s been creating various futuristic weapons you see in this amazing tactical strategy. In this interview, he shared some of his thoughts on weapon creation, talked about the technical side of things and discussed the importance of functional design. Check it out.
Hello, my name is Ben Leary and I’m a weapon artist from Central Florida USA. I’ve contributed work to games such as Firaxis‘ XCOM 2, Strike Games’ Final Rush, and some indie titles as a contract artist. Right now I am working at n-Space in Orlando on a firearm configuration application called GUNSTRUCTION where I model licensed gun components.
I became interested in weapon art in the 90s as I was obsessed with early FPS games like DOOM, Quake, Goldeneye 007, PO’ed, and Turok. I used to design my own weapons on graph paper in school when I should have been paying attention to class. Passion for it grew as the Shooter genre expanded. To my surprise, “weapon artist” eventually blossomed into an individual role instead of a subset of character artist duties. As my hard surface modeling skills grew along with my knowledge of firearms, I began getting offers for weapons work in the video game industry. I have love for all phases of gameart, but weapons are what have always interested me the most.
Modeling of Weapons
Creating weapons is different from other types of gameart by how they tend to be very prominent in a game, usually taking up a good deal of screen space and usage by characters. Regardless of the game type, things like shape language and silhouette are paramount. The designs are often approved long before a weapon artist is tasked but I always enjoy having design input. As a specialist, there are often opportunities to use my expertise to influence design. Even if only within the gaps not specifically outlined in concept art.
I approach each new weapon project by first nailing down the shape language with a thorough blockout. In this earliest stage, I am mindful of how the high poly, the low poly, and its’ projected cage will need to work in harmony. That step is vital to saving time during the baking process that’s otherwise spent adjusting the hipoly/lowpoly models or tweaking the cage. If done correctly, you’ll bake a normal map with a default cage projection resulting in no errors. Which also means if anyone on your art team needs to edit and rebake your weapon, they can do it themselves without any hassle, further proving your value.
3d Modular Structure
There are many schools of thought on how to properly create a weapon. Personally, I feel it’s better to model weapons how they’re constructed in real life. As separate components which are machined to fit together snug. This modeling approach makes more logical sense to me workflow-wise and end results tend to look more authentic that way.
Every weapon has its’ own anatomy and it’s important to figure that out at the beginning. How was it manufactured? What kind of ammo does it use? How is it reloaded? How is it aimed? How does the character holster it? What features does it have to dissipate heat, recoil, and noise? All important factors to understand from the beginning in order to make a weapon feel authentic.
I attack each component the same way. I’ll draw a side or top profile with polygons that form the major shapes. Next I’ll block out the other dimensions of those shapes and start to fit them together appropriately. I’ll then model macro details with special attention to landmarks and ergonomics, then continue to work my way down the chain into the micro details. Eventually, I incorporate support edges to round everything out.
In the end, the game resolution model will be merged into one piece aside from any components that animate. This is mainly for the benefits of optimization, cage projection, rigging, and light leaks.
Is there functional logic to sci-fi modeling? There damn well should be. Too often you see details in sci-fi art that is just slapped together to look cool. Pointless greeble added without specific intentions. I’ve heard this described as “shotgunning” detail onto models. I like that term. Sci-fi still needs to be grounded in reality, and the reality is every single detail on real-world weapons (or anything) is there for a reason. There are aesthetic details of course but they tend to be limited, angular or cosmetic in practice. Aesthetics are where the shape language comes in. Every fine detail should always have a purpose. The “rule of cool” has limits in my opinion.
Screws fasten panels. Bolts hold together frames. Wires supply power and transfer data between components that need to communicate. Holes and vents dissipate heat. Recessions reduce weight. Rounded surfaces serve the ergonomics. Rough areas give the user more grip. Extrusions provide added support strength. Paneling outlines the manufacturing/assembly process. Lights display information to the shooter. And so on. These are just examples, but the key is to give value to everything. Real world weapons don’t have details that serve no purpose, it just wastes money in the manufacturing process and will likely reduce efficiency in one way or another.
A good way to understand this is to go to your local gun store (while they aren’t busy) and ask the salespeople what different details on any given weapon do. Nine times out of ten, they will not say “to look cool”. This is actually a good way to study firearm design too. They love to talk about their wares, trust me.
At the risk of sounding pretentious, this is actually the best thing about being a weapon artist; Telling the story of your weapon. You bring it to life by grounding it in its’ own anatomy. Treat it as a character with individuality. Use science fact and your artistic intuition to create pleasing science fiction. It simply doesn’t get more fun than that.
Creating textures, I’m a little more old school in that I’ve been using Quixel’s dDo to automate creation of base materials, then hand-authoring those textures in Photoshop. More of a classic method but the industry seems to be moving away from it.
With the rise of software like Substance Designer and Quixel Suite, the texturing phase has been turned on its head. It’s become much easier to create realistic materials quickly. Things like surface wear can be automated with clever use of nodes and masks all in a non-destructive workflow. Iterative adjustments are only a few clicks away. With Substance Painter, you can have the freedom of painting directly on your assets with the power of the procedurally-generated materials you create in Substance Designer. I’m currently in the process of adopting this workflow and am already falling in love with it.
The differences with creating weapons for games instead of just virtual models is mostly in how the weapon is used in-game and at which angles it will be seen. Besides having certain components that have to animate, the usage of the UV space can differ. For example, in an FPS, there is no sense in having the UVs on the front sight (farthest area away from the player) take up the same amount of relative space on the UV map as the rear sight (area closest to the player). Interior sections and surfaces along the bottom that will hardly be seen can have their texture usage cut by half. Also, if the right side of the gun will never be seen, there’s no use in having any unique UVs there. Examples like these improve the overall quality of the entire model. A conscious effort to always make the most of your budget makes you a valuable asset to an art team.
As far as a polycount goes, weapons aren’t that much different from anything else. Only use what you need. You just have to be mindful of the viewing angle’s silhouettes, animatable geometry, and the number of divisions on rounded segments. Pretty standard stuff.
I would be happy to share advice on hard surface modeling. Besides my argument against shotgunning detail, my first tip is the general understanding that your hipoly model is likely the most important phase in weapon production. Poor hard surface hipolys make for poor normal map bakes and poor textures. It pays off to take your time and get the high poly solid, the best it can possibly be. The software you use doesn’t matter, your execution does.
On that note, floating geometry should only be used sparingly. For those who don’t know, floaters are separate geometry that hovers just above a surface giving the impression its’ connected. It’s meant to save time and be less complicated. The problem is you’ll often run into weird bake errors that you’ll have to fix down the line. I already made a point about why you should keep the baking process straightforward. Most importantly, it’s easy to start leaning on that technique as a crutch in your modeling workflow which can be detrimental to the growth of your hard surface modeling skill set.
Also, ngons are your friend. Use them. Only thing that matters is the look of the end result and hard surface modeling allows plenty of opportunities for ngons. Terminate edge loops into them, use them where geo pinches, put an ngon where two complicated shapes join. Try to use ngons appropriately as you model your hipoly. If it subdivides fine and looks good, it’s good.
Lastly, zoom out. Remember to look at the model periodically from realistic in-game viewing angles. Don’t get obsesed modeling details that no one will see. Focus on how the major shapes are reading. Don’t get caught up on fixing troublesome errors that don’t show when you zoom out. And don’t put support edges so tight that the edges end up looking super sharp, even when you want them sharp.
Thank you to 80.lv for honoring me with this interview, I enjoyed it. Another thank you goes to the readers, hope you learned something new.