May I?
It’s practically impossible to go on about our daily life without bumping into news about the latest AI advancements. We are getting more and more used to using it every single day, and we use it for everything: from writing an email to deciding whether we should visit a doctor about a symptom that we first share with our favourite AI agent. Some debate how far it can go, but one thing is for certain: it’s here to stay in one form or another, and life will never be the same as before.
The game industry will be a big beneficiary or a big loser of the AI era, depending on how you look at it. It will be a double-sided transformation for sure with a lot of creative, ethical, and practical questions and hurdles along the way.
With procedural content generation covering everything: models and textures, levels, animation, music and sound effects, dialogues, etc., large-scale game development that once was only the privilege of AAA studios will be democratised in a very short period of time. The level of these kind of “big buck” games will be in realistic reach of indie game development studios. Once again, independent teams will be able to explore and experiment with interesting game mechanics without the manual and labour-heavy content creation for their games.
But there is a catch. Now that the barrier to entry is practically gone, the floodgates are wide open. We are about to see a tsunami of mediocre, AI-generated “slop” to hit the market. If anyone can create amazing RPGs with a few prompts, then everyone will. The stores will be buried under mountains of competent but soulless clones. Standing out from the crowd is going to be harder than ever before. Visibility will be the new currency, and for a developer to survive in that noise, “good enough” won’t cut it anymore. You will have to be extraordinary.
Triple A companies won’t have this luxury or agility. Their profit-oriented business strategy will demand the profit to keep on increasing, which will shackle them to the ground and force them to stick with the good old recipe of homogenous, boring games that they’ve been producing. They are trapped in a golden cage of their own making. They simply can’t afford to take risks because the cost of failure is too high. So, they will likely use AI not to innovate, but to churn out the same “safe” content faster. We’ll reach a point where better graphics will stop making games more fun or better altogether. Now, AI is about to make “perfect graphics” a commodity. If a three-person team (or even a solo hobbyist) can generate a photorealistic open world using AI tools, then visual fidelity stops being a unique selling point for the billion-dollar studios. The “shiniest toy” won’t be special anymore because everyone will have it.
So, what is left of the art of creating games when making them is as easy as baking a cake, and the market is flooded with thoughtless clones?
The physicality.
When the visual spectacle becomes a commodity, the only way to rise above mediocrity is by bringing an exceptional kinetic interaction between the player and the instrument of entertainment. The effect of how the player actually feels the game and its mechanics…This direct visceral connection has been traditionally made through peripherals such as joysticks, steering wheels, but not with handheld gamepads…until now.
This brings us right back to why we are here, why I am determined to make Zento a reality. In an AI-generated world of infinite visual complexity, the physical connection: the input, the resistance, the feedback, these tactile sensations will become the only true grounding excitement for the player. So, AAA studios can keep their shiny pixels, the future belongs to those who can make you feel alive playing their game, not just consuming it.
Theory and practice
If everything worked out as intended in theory, life would be a fairy tale. I was so naive, thinking that everything I designed on the computer would work flawlessly on the first attempt. In practice, reality kicked me in the face with a steel boot. Even though I was lucky 90% of the time with component fits and clearances, the moving parts—especially as they got smaller in dimension—really made me question everything I thought I knew and had achieved. I found myself lying at the bottom of the Dunning-Kruger effect’s curve.
The small gears, which are the main parts of the power transmission between the motors and the thumbstick, turned out to be a hard nut to crack. Originally, I wanted them to be 3D printed out of plastic, but the motor shaft would just slip inside them. So, I had to resort to copper gears. Ideally, they could be CNC machined, but because of their small size, all the CNC machining services refused to make them—probably for a good reason. The second best thing I could opt for was metal 3D printing. I had no prior experience with metal printing, but the parts came back looking surprisingly good. But “good” wasn’t enough for a mechanism of this size, especially with such tight tolerances. I even printed the gimbal centers that have integrated gear sections on them, but boy… printed metal on metal doesn’t work too well. Even after a tedious amount of sanding and polishing, the porous printed surface just wasn’t meant to be used for gears. Well, at least not at this scale. I still tried to assemble the setup, and in the process, I managed to damage two of the motor shafts when I applied a little bit too much force during the pressure fit. At this point, I was fuming with boiling rage. I wanted to smash the whole thing to pieces, but then I reminded myself that if I’d gotten this far, this drawback wouldn’t stop me. I had to step back to advance the project. I had to accept the fact that on such a small scale, a mechanical gear transmission just wouldn’t work.
OK, no gears means no DC motors. Earlier, I had plans for using small belts to transfer power to the motors, but that’s another rabbit hole I definitely didn’t want to go down. But what else? To be honest, the solution is even better than my original plan. The obvious solution is… (drumroll…) electromagnets. They allow for the same precision and even faster, lag-free movement without any gear grinding. How come I hadn’t thought of it in the first place?
With all the benefits, of course, come their own problems. For example, Hall sensors are out of the question as they operate by sensing magnetic fields. In this setup, a 10x10mm cylindrical magnet would be constantly pushed and pulled by big electric coils; it is not an ideal environment to measure the rotation of the magnet. However, since the electromagnetic field does not require direct contact, the setup would be frictionless, meaning there would be no part deterioration. It would be a shame to use potentiometers to measure the movement of the stick. A rational solution I found is to use reflective optical sensors, which can detect movement against a patterned background. These sensors are small enough and just as precise as Hall sensors.
So, even though this problem was a major obstacle, it was helpful to realize that sometimes it is necessary to sit back, reassess, and admit when I’m not on the right path. Even if the gear mechanism had worked in the prototype, in the long run, it would have been nothing but trouble. The motors are heavier, they struggle to hold position or maintain specific torque, cost more, and on top of it all, you can actually feel the gears “grinding” under your thumb.
The new electromagnet setup will solve all these problems. It will be much more durable, with zero friction between components. Let’s see how it goes.
Stay tuned…
