I’m a Senior Concept Designer in Entertainment, which means I mostly make art for games and Netflix shows. That means I’m one of the most vulnerable people to be replaced by AI, or so the AI industry says.

Whether it is true or not, it's beyond the scope of this blog post today.

What I saw were these flamethrower concept designs, made by AI and that inspired me to make my own:

Why these inspired my Weapon Concept Designs

And no disrespect to technological advancements and people who use AI in production, I think these were done using Krea, an AI image generator that has more settings to fiddle with than most, but seriously? This is where we're at now? Settling for mediocrity.

You can see the annotations I made, which will show you exactly what is wrong with the designs: weapons that don’t have a way to reload, scopes you can’t see through… rifle magazines in a flamethrower?

Usually, an artist like myself would research and develop weapon designs upon request for game studios, so we, artists, know a little about a lot, and we know enough to be able to recognize bullshit from a distance.

Our eye is trained for this, unlike most non-artists.

After stumbling upon these flamethrower concept designs, I was inspired to make my own. Sometimes you just need a trigger, something that tickles something inside your artist brain and motivates you to do something. These AI designs were it; below I’ll add some notes on why I thought they were insufficient, and what I did to make my own instead to avoid the mistakes AI made.

Disclaimer: This is not a post to dismiss AI capabilities, but rather to ground them, discuss them, and lay the groundwork to analyze them and break them apart; this is, after all, a case study. As my father, a successful salesman, used to say: If your product is good, you don’t need to trash your competition; in fact, it’ll be worse if you do. People will not remember the comparison you made about how your product was more valuable; they’ll remember a possible insecurity you had about your competitors — and just like that, this post is to inform, rather than compete.

If I sound overly critical about AI designs, I assure you, I am, and that’s because I was raised to “do my best, and nothing but my best.” so when I see flamethrower designs with no clear liquid fuel, no clear way to reload, and no clear way to… operate, to put at least, my designer brain is triggered.

The correct Process for creating Weapon Designs

Before using AI, take a step back, think about what you’re doing.

Don’t jump into the pool cannonball if you don’t know what’s in the pool. That is to say, don’t start prompting frantically if you don’t understand your subject matter; you cannot bypass this. Attempting to do so will result in work with design flaws.

First, I went to Wikipedia read about the history of flamethrowers; (ugh, reading, am I right? I don’t wanna read…) and it was crucial to understand their purpose, why they were invented and what I can remove, replace, adapt and create on them. First understand, and only then try to build.

I found out plenty about flamethrowers, and the more I learned, the more iterations were flashing through my head. I highlighted and extracted quotes from Wikipedia and other sources on what stood out to me the most, this is an important distinction, this is my flamethrower design, and it’s no one elses, not an algorithm’s, not Midjourneys, mine, mine alone, my interpretation. Artists are hired for a vision.

Essential Reference Gathering for Weapon Design

Maybe the coolest part of the research work is how the understanding of the subject matter shapes as you go, and your assumptions about what a flamethrower design should look like, change as you work on it, you may think you got it in the Wikipedia stage, but you don’t, and you learn about it as you go, it’s like getting to know a person, you may think you know them, but the chances are, they’ll surprise you, and you know the tip of the iceberg about them only.

That’s insane.

I stumbled upon this reference while using the SOOT SPIRAL search image engine, and I thought: “Does that hunk of vegetables kind of look like a flamethrower? This is crazy,” and then it hit me: it’s not the cylinder or the red color; it’s everything - the silhouette and how bulky it is - that defines our perception of the weapon. Flamethrowers are usually bulky, just like… whatever this is!

Because of this, I decided to set out and find out what the best silhouette for a flamethrower is. When I uploaded these to Instagram and made a poll, most people voted for numbers 7 and 2. I was honestly very keen on number 3; there’s a brutalist feel to it I adore, but it wouldn’t make sense to poll and then reject the feedback, so I went with 7.

These were mostly to figure out how the silhouette could be interpreted.

Each stage of development has a purpose; the black-and-white silhouettes don’t all look functional, but the purpose here was to explore shapes that would be recognizable by most people as a flamethrower, at first glance.

Style vs. function? Why not both?

When I was studying concept art, Adam Adamowicz was kind of my hero; he was the concept artist I looked up to because he was so, so original. He’d use unconventional objects, like a cheese grater, as armor for a super mutant in Fallout 3. That set off a light bulb in my mind: all I have to do is re-imagine the same objects for a different purpose, and I’ll be just as original. That probably influenced the way I choose reference; for instance, when it came to colors, I tried to pick those that remind me of something warm, like fire, but also that would give me the same aura that a flamethrower would have (something dangerous).

The secret to being original is to look at the mundane and reinvent it.

Of course, I’d be a hypocrite if I didn’t look at a looot of flamethrower references before blocking out the work.

Building the Blocks — The Weapon Design Foundation of Any Good Design

Use a tool that works for you, not one you work for

To block out the work, I start with 3D, this is useful because I don’t have to think in perspective, otherwise I’d have to draw everything in tight perspective fundamentals if I was using drawing as a method to achieve this design, with 3D, I can focus only on how the forms fit into the 3D space, and how they connect together, this is very useful for the actual design part. A lot of concept artists look down on 3D and see it as cheating, but 3D isn’t trying to do what 2D is doing, and vice-versa.

I made a few changes as I worked on this, because while I’m making this, I’m learning, I’m learning about how the tubes connect and why they’re there, and as soon as I click them together, the reason for them to be there clicks in my head, so when a director asks “Why not to remove those tubes?” I know what to say, I’ll walk them through the cons and pros of doing that. This doesn’t mean I know better than the director; it means I was tasked with this, and in this specific case, we’re a team, and we all pull our own weight.

The Process Should Drive the Outcome

Let’s take a quick detour and analyze this painting for a second:

This painting by Wouter Tulp, for instance, perfectly describes the process of development: an idea doesn’t begin and end in a preconceived assumption; instead, it is the fruit of research; this can be an unfinished painting, or it can be a mysterious portrait; the process, a lot of times, drives the outcome. When you focus on the process, you may not know where it’ll take you, but you know you’ll want to be there; and if you focus on an outcome, you will already be where you’ve always been. You may argue this is “good enough”, but good is a known quantity; it’s what we all agree on. As long as you stick to good, you’ll never allow space for your project to grow.

Theme references

AI Designs don’t consider every angle — like the time period, theme, and function; those are things you need to think about, study, and explore to solve any design problem.

Assuming a two-line summary from ChatGPT can pack the whole spectrum of human experience, existence, and a week-long study is silly. Studying is important, and being smart is cool, I promise you.

The theme is important; a flamethrower for Metal Gear Solid will be different than one for Apex Legends or Fortnite. This is also part of the design: the manufacturing limitations of the time period, the faction that built it, and even their philosophy and outlook on life shape the way they build something, because, as said before, people bring their unique vision to the process of creating something, and you can see how these sort of fall into the same style, since they were made during the same time period, and people had different cultures, ideas, and technological limitations than they do today. In previous blog posts, I also shared some tips on how the environment shapes how people look

I like to explore as many possibilities as possible, and then work backwards from the chaos by removing the ideas that wouldn't work and limiting what's possible to take forward.

The result of the human-led design:

All designs by Miguel Nogueira, sketched in Photoshop, modeled in ZBrush, and rendered in KeyShot.

Some aspects could still have been pushed, but for an exercise like this, I was relatively satisfied with the result.

Final notes on AI vs. Human Design

Balancing fiction with function is a challenge a concept artist often faces in production.
Copying your real-world reference directly risks creating boring product design. Worse, if you don't design for your virtual world instead, you miss out on creating an entertaining gameplay experience for your players.

Radars like these were used on airplanes or ships; it wouldn't make sense to use them on a weapon, but when gameplay enters the scene, it's up to the concept artist to reinvent them.

The benefit of studying your subject matter, collecting references, and overthinking is that by now, we know which parts we can keep and which we can replace for this sake.

In this design, you're invited to consider the entertaining shape language of a retro-cool camera sight as a seismic scanner for underground or nighttime missions. It provides the needed cover for a flamethrower wielder, while allowing the enemy NPCs to detect the player gradually as they fire. This ever-present vanishing edge forces them to calculate their next move.

This is a story of being detected, vanishing into the dark, and starting over. Essentially, we are reimagining hide-and-seek mechanics.
And as such, the game begins!

Do you feel your hair greying out in real time because of all the concepts you have to solve in your JIRA backlog? Say: hello@menogcreative.com

These AI Weapon Designs inspired me to make my own