What Players Actually Struggle With in Mirlo Volador
I've watched a lot of people play Mirlo Volador. Friends, strangers who found the game through social media, developers I shared it with for feedback. After enough of these sessions, patterns emerge. The same things trip people up, often within the first 30 seconds, and in ways I didn't anticipate when I was building the game alone in my apartment.
Some of these struggles led to changes in the game. Others are fundamental to how hand-tracking works and can't really be "fixed" — only documented and worked around. This is what I've observed.
The Camera Permission Moment
Before a player ever sees the game, they see a browser permission dialog asking for camera access. This is the single biggest drop-off point. It's not a bug I can fix — it's a browser security feature. But I can control what happens around it.
A significant portion of people deny the camera permission on instinct. The request comes from a game site they've never visited, and the kneejerk reaction is "no." I can't blame them. The web has trained people to be suspicious of permission requests.
For those who allow it, there's a subset who have previously denied camera access on the same browser, meaning the permission dialog doesn't appear at all — the request just silently fails. The first version of the game showed nothing in that case. Players saw a black canvas with no explanation and had no idea what was happening.
I rebuilt the permission flow. Now the game shows a clear loading state while requesting camera access, a specific error message for each failure mode (permission denied, permission previously denied with instructions to reset it, no camera found), and a retry button. It sounds obvious. I should have done it from the start. The bounce rate on the game page dropped noticeably after this change.
The Lighting Problem
Hand tracking with MediaPipe works best when the hand is clearly visible, well-lit, and in front of a plain background. The real world is messier than that.
The most common scenario that breaks tracking: a bright window behind the player. The camera's auto-exposure adjusts to the bright background and the hand ends up underexposed — dark, low-contrast, hard for the model to detect reliably. The tracking becomes jittery or disappears entirely. Players think the game is broken. The game is fine. Their lighting is the problem.
I added a tracking quality indicator to the UI: a small visual signal that shows whether the hand is being detected confidently or not. When tracking is weak, the indicator changes color. Players quickly learn to associate the indicator with their environment and make adjustments — turn away from the window, turn on a lamp, find a lighter background. This indicator reduced confusion significantly. Players stopped thinking the game was broken and started understanding that the camera needs decent lighting, just like any camera.
I also added guidance text during the initial hand-finding phase: "Make sure your hand is well-lit and fully visible." It's a small thing. But players actually read it, and it sets the right expectations before they start.
Going Out of Frame
New players move their hands too much. The natural instinct, when you first understand that your hand controls the bird, is to make large sweeping movements. The bird shoots to the top of the screen, then crashes immediately as the player overcompensates downward.
But there's a related problem: players move their hand so far that it leaves the camera frame entirely. When the hand exits the frame, the tracker loses it, the last known position is held briefly, and then the bird drops. From the player's perspective this feels like a random bug — the bird just suddenly fell. From the tracking perspective, the hand simply wasn't in the picture anymore.
The fix had two parts. First, I added a visual overlay showing where the camera sees the player's hand. This "mirror" view helps players understand the spatial relationship between their real hand and the tracked position. Second, I made the active zone of the camera smaller — you don't need to reach to the edges of the frame to move the bird across the full screen height. Small, centered movements are enough. This keeps hands in frame naturally.
The most counterintuitive lesson from building this game: less movement is better movement. The best players barely move their hands at all compared to beginners.
Expecting Touch Controls
A meaningful number of players, especially on mobile, try to tap or swipe the screen. This is deeply ingrained behavior. A game on a screen in a browser should respond to touch. That's the pattern years of mobile gaming have established.
Mirlo Volador doesn't respond to touch. The whole point is the hand tracking. But when someone's first instinct is to tap and nothing happens, they often conclude the game is broken and leave.
I added a prominent instruction card that appears before the first game starts. It shows a hand icon, a brief text explanation, and a "Got it" button to dismiss. This is the kind of onboarding I resisted at first — I wanted the game to be intuitive enough to not need explanation. It is not. Camera-controlled games are still novel enough that players need to be told explicitly what to do. The instruction card reduced early abandonment on first visit noticeably.
The "Calibration" Moment
Even after players understand the concept and get their camera set up correctly, there's a calibration moment that catches almost everyone on their first run. The game starts, the bird appears, and the player moves their hand slightly. The bird reacts more than expected. The player moves back, overshoots the other way, hits a pipe, and dies in under three seconds.
This isn't because the controls are too sensitive. It's because the player hasn't yet built an internal model of the input-to-output mapping. How far does the bird move when I move my hand this much? What does "steady" feel like in this control scheme? These questions get answered in the first few deaths. By the third or fourth run, most players have the feel for it.
The first pipe in the game now has a wider gap than subsequent pipes. This gives players one "free" pass to get their bearings before the real difficulty begins. It doesn't solve the calibration problem entirely, but it reduces the frustration of dying in the first two seconds and building negative associations with the control scheme before it's had a chance to click.
The Fatigue Factor
This one I didn't anticipate at all. Holding your arm up in front of a camera for an extended gaming session is tiring. Keyboard and mouse gaming keeps your arms relatively low and relaxed. Hand tracking requires your arm to be elevated and extended, which is physically taxing in a way that traditional gaming is not.
I noticed it myself during testing. After about 20 minutes of continuous play, my shoulder started aching. I talked to a few players who reported the same thing after longer sessions.
There isn't a complete solution to this — it's a physical constraint of the input method. But I added explicit guidance in the tips section: keep your elbow on a desk if possible, don't fully extend your arm, take breaks. The game is better played in shorter, focused sessions than in long grinding marathons. Accepting this and communicating it honestly is better than pretending the fatigue doesn't exist.
What Keeps Players Coming Back
The flip side of all these struggles is that once players get past them — once the controls click and they have a clean run — something shifts. The game becomes genuinely absorbing. I've seen it happen in real time: the confusion and frustration of the first few minutes gives way to focus and then to competitiveness.
The leaderboard is the biggest driver of return visits. Players who create a profile and see their score on the board are substantially more likely to come back than players who don't. Competition is motivating in a way that solo high-score chasing is not. Seeing your country flag on the board, knowing that a real person in another part of the world is 10 points above you — it's a simple mechanic but it works.
The other thing that keeps players coming back is the physical element. After a session of Mirlo Volador you've genuinely moved your body. It's mild exercise, almost. A few players have mentioned this as a reason they keep playing. That wasn't a design goal. It's a side effect of the input method. But I'll take it.
If you haven't tried it yet, the game is free and takes about 10 seconds to start. Be patient with the first few runs. The controls make sense faster than you'd expect.