Show HN: Fly By – retro biplane flying game (michaelteter.com)

1 points by michaelteter ↗ HN
Feedback welcome. Best on desktop. Works on mobile, but needs more polish there.

My concept, my guidance, and Gemini Flash doing the work. Took about 4 hours and 50+ conversations with Gemini.

Big Kudos to Gemini Flash (since 3.5 and later), because I have had great success with it on many projects.

65 comments

[ 0.17 ms ] story [ 59.6 ms ] thread
The scan lines are a great touch. It makes it feel truly retro
Nice job - it kind of reminds me of the rocket barrel levels in Donkey Kong Country: Tropical Freeze.

Feedback: Consider playing with the physics of the plane depending on your pitch, so that some parts of some levels might require you to deliberately pitch down for a few seconds to build up some "stored speed".

It basically does this. If you are in level flight or a dive, you increase speed up to a certain max (level flight max is less than dive max). If you climb, you bleed off speed. To reach high balloons, you often need to build up more speed before you can climb high enough.

There's probably room for improvement, but it's already enough fun that I can play it for several minutes at a time.

Now I'm trying to decide if there's anything else of value to add - without making it too much of anything. I think I'm about finished now.

It is fun. Now make it multiplayer over websockets/etc with some dev controlled signalling server (who peers register with/etc) and STUN check to get peer IPs locally.
Loved it. That's exactly what AI should be used for. Something that would take us one month at least, full of bugs, AI now delivers in less than a day, ready to ship

This is the essence, AI is a super power, the world is ours

Something that would take us one month at least, full of bugs, AI now delivers in less than a day, full of bugs
>This is the essence, AI is a super power, the world is ours

Don't me overdramatic. The guy who made Flappy Bird did it in a weekend without AI. A simple game like this wouldn't be difficult in any modern framework. That it can be vibe-coded in less than a day isn't really much value.

It now exists. It did not exist before I “made” it.

For me it has value, because it is something I wanted to play.

Well I dunno, man, this looks like the sort of amusing prototype I'd expect to take a couple of days - though not to understate the value of getting it in 4 hours. Iteration time is an important factor.

(Regarding the parting shot, I find myself once again, for the numerousth time, reminded of this curious lobsters comment that I'm having difficulty forgetting: https://lobste.rs/s/omq8rt/vibecoding_gets_emacs_patch_rejec... - perhaps not quite entirely directly germane here, I don't think, but, still, somehow, not quite entirely directly not.)

> Once the user is sufficiently (l)ionized, they're no longer receptive to neutral positions.

Every step in this argument (particularly this one) seems tenuous at best, and in need of citations.

(comment deleted)
Let us hope that the poster happens to see this, and is moved to comment, so they can address your points.
This was a nice game. It would be cool if you added some more purpose to landing the plane, like if the fuel and throttle were limited, making it actually useful to land it. Overall great game and enjoyable. Also would be nice to get some extra points for doing loop-de-loops or tricks with the plane. A shield power-up is also a great idea.
Yes, you're right - I need to improve the whole ground experience. The plane does gradually use fuel, but you're vastly more likely to crash before you run out of fuel.

Landing and pausing on the runway refuels the plane, but the control behavior on the ground sucks.

The fake scanlines and position of the UI elements make them difficult to read. The plane is too slow and its turn radius is too tight, the physics (if there are any) need to be improved to give a sense of weight and velocity to the plane IMO, and that probably necessitates a bigger playfield. Also it feels like it really needs a gun.

Overall it feels like an OK prototype but I agree it needs more polish.

That said, HN is traditionally for sharing code with the community. Maybe for AI submissions people should share their prompts? I don't know. I'd ask to see the code but since you didn't write it you can't really discuss it.

> I don't know. I'd ask to see the code but since you didn't write it you can't really discuss it.

Brusque but not inaccurate, Welcome to our future I guess.

> I don't know. I'd ask to see the code but since you didn't write it you can't really discuss it.

Huh? Why do you believe this? ever take a literature class?

Regarding scanlines, there is a setting to turn them off.

> Also it feels like it really needs a gun.

It took me less than 10 seconds into gameplay to pick up bullets for the (already existing) gun. Your comment reads as if you didn't try to play the game for more than a few seconds.

It's a fun little game; no need to be so smugly dismissive.

>Your comment reads as if you didn't try to play the game for more than a few seconds.

I made it a few minutes, I admit it was boring. That isn't my fault. In a better designed games these things would be more obvious.

I think it might be interesting to know age ranges for posters, because this made me smile as it reminded me of hours spent trying to perfect Barnstorming by Activision on my Atari, right down to the scan lines and flight dynamics. Not sure it’s supposed to be an actual flight simulator, so your point (if there was one) just seems like being a prick for no particular reason. Did skin come off your nose while playing?
I'm giving my honest impressions from playing the game and suggesting ways that I think it could be improved. If I ever posted anything here I'd expect the same treatment.
Yeah, I know. What I’m suggesting is don’t.
This is a forum for conversation, you don't get to decide who participates or how.
I mean, that’s like your opinion man.
Yeah, this feels like a vibe-coded game that people who played it for 1 minute are upvoting, but it just really lacks the feel of how even a mildly solid and entertaining game feels. I got bored very quickly.

I flagged it because of HN's AI content policy. I know it technically doesn't apply to submissions, but I think game submissions (or any others) should show more humanity, instead of being like "look at this momentarily pleasing thing Gemini coded for me".

I guess the 6 hours of intentional dozens of interactions I had with Gemini could be considered slop to some. But I created this to scratch an itch of an old game I enjoyed in the past. And I enjoy playing this one.

After 30 years of hand-crafting code in 8 languages, I think I can be excused from having to do the tedious work of translating ideas into executable code.

You would see about 50 conversations.

Contrary to what some negative comments suggest, this was not a simple one prompt game. I made a lot of choices and gave a lot of direction to Gemini.

If I get motivated enough I might try to make a meta doc describing the actual process with key summaries of different types of conversations.

If you pick up an additional machine gun powerup, the plane should fire more bullets simultaneously.
or fire bullets for longer, perhaps. it seems like in this case, neither happens.
This took me back to my first PC gaming experience at age 6ish. Circa 1988. Nostalgia unlocked!
Anyone else remember Sky Kid?

Great game and this reminds me of it. Thanks for sharing.

I like it, reminds me of the old arcade games. I'd rather use left and right keys instead of up and down but no big deal.
Maybe this is intentional, but the up/down controls felt a bit counterintuitive to me. Up makes the plane go down, and down makes it go up - is that by design?
That's how a real plane's control works, so many games do the same.
Indeed. Trim wheels when rolled upward trim for higher speed, i.e. nose down.

Wait, this agrees with GP's experience but not mine. Did OP invert it?

Yeah, that makes sense. I've never flown a plane or played this kind of game before, so it felt a bit unusual to me at first.
When you're in a cockpit you're pushing the stick forward to dive, up key on the keyboard is often what we use to represent that but it's not a perfect analogy.
I'd rather see AI being used to generate new games for old platforms (console emulators) than these faux-retro things with mixels, inconsistent graphics styles (some blocky bits, some curves), and rotation of squares.

Constraining the canvas to a low-res pixel grid would be an instant improvement. The recent post https://airport.apunen.com/ does a decent job of this but it, too, has an odd mix of modern aircraft sound fx and vintage biplane graphics. It doesn't feel very retro because it's full 3D rather than limited 2D sprites of a 3D scene.

I'd rather not see AI being used to generate new games for any platform. I'm not gonna read something nobody bothered to write, and I'm not gonna play something nobody bothered to write. AI has no taste and it's not a matter of post training because the Taste is the human component and actual effort put into something.
This is a silly take. You may as well refuse to play a game written in RPG Maker. Writing code doesn't make a game fun, playing it and iterating on experiments does. Generating code doesn't imply a disregard for the end product in the way generating prose does. Generating prose is generating the entire end product, so yeah, the end product was disregarded. Code is about as interesting to a game as a cutting board is to a meal. Some necessary infrastructure that you could reasonably disregard in order to focus on the end product. The game isn't better for having more carefully written code anymore than a meal tastes better for having been prepared on a fancier cutting board.

I really enjoy carefully and thoughtfully writing code by hand, I think there can be real artistry and creativity there, but no, it isn't the same thing as the game itself.

That was relatively fun but there should be a penalty for flying upside down and backwards, it should lose altitude or something.
I had a lot of fun, the scan line effect looks really good and the obstacles are interesting and challenging. I'd have liked to use my game controller, there's a browser API for it. And to have changed my control settings. I thought both mobile and desktop controls were good though.

I've been working on a browser game too lately. It's a pretty great environment to work in. Three.js + WASM have cured my desire to learn Godot, at least for now.

Love the flight dynamics, super fun game!! Now can you make GTA VII??
tried it and liked it but I wouldnt call it retro, at least by the looks of it
Triplane Turmoil was better.
Many games are better.

Many comments are better too :).