Show HN: Fly By – retro biplane flying game | Hacker News Reader

Show HN: Fly By – retro biplane flying game

michaelteter.com

85 pointsmichaelteter67 comments

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.

Like many side scrolling airplane games, the object controlled in this game behaves a lot more like a rocket than an airplane. I suggest learning a little about the physics of airplanes[1] to improve the feeling of flying.

I made a prototype[2] once (not working on mobile, unfortunately) to show what can be done, and it takes getting a lot of details right![3]

[1]: http://av8n.com/how/

[2]: https://xkqr.org/flightle/

[3]: https://entropicthoughts.com/sidescrolling-flight-simulator

Second this. I've been trying to build a casual flying game in unity and getting the physics right is hard.

What I found particularly hard is simulating accurately how gravity works against the aircrafts momentum, and trying to get stall and recovery right

The goal was to make a game similar in look and feel to an old console game I played when I was a kid.

Just like with many arcade racing games, the physics are not intended to be hyper-realistic.

I do love realistic physics in flight/driving sims, but that just wasn't my goal here.

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.

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.

For further inspiration, check out Wings of Fury, an old DOS game. https://www.myabandonware.com/game/wings-of-fury-ta

Apparently a newer 'inspired by' successor is https://store.steampowered.com/app/1371710/Rogue_Aces_Deluxe...

It's no Sopwith

The Sopwith Camel, flight simulator memory unlocked.

I was expecting this before I clicked on the link and I was not disappointed.

https://en.wikipedia.org/wiki/Sopwith_(video_game)

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 (edit: OK it does.)

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.

Literature is written by human beings, and if the author was in the class I expect they would be able to discuss their process and methodologies.

But the entire point of using an AI to generate code is not to pay attention to any of that. The author of AI generated code did not write said code and cannot therefore provide any insight into the process or methodology beyond "here is the prompt and model I used." Nothing that would be of value to the few of us left on this site who actually care about such things. Nothing, in other words, to gratify intellectual curiosity.

Also regarding scanlines, fake scanlines shouldn't obscure UI elements or make them difficult to read. Actual scanlines on actual CRTs didn't do that.

Most literature classes discuss works where the author is unavailable.

The point of AI is not to generate code and not pay attention…

>Most literature classes discuss works where the author is unavailable.

This is not actually a literature class, and the author, as such, is available. And the purpose of Show HN is to have a conversation with the author.

>The point of AI is not to generate code and not pay attention…

Yes it is. That's what "vibe coding" is. That's what all AI workflows converge towards. That's how most people use it. And that's what makes discussing AI generated posts so limiting.

The author did not claim to have “vibe coded” anything.

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

You are right though. It is exhausting since some people just can’t get it through their head that you can use AI and still understand the output.

Nobody's generating this in four hours and understanding anything, but if I'm wrong I welcome the author's replies. It would make more interesting conversation than usual for these threads, which I guess are just supposed to be for kudos now.

I am the author.

Is code the art? Or is the product of that code, and the way people feel when they use that product the point?

At this point, after enjoying the craft of writing code for decades, I now just want to build things I like. At this point I care zero about the details of the code as long as the product meets my requirements.

Until you try building a complete (if simple) product, you may not realize how many decisions are involved which have nothing to do with code but are critical to the success of the product. That is my focus.

I agree with you the game is mediorce and feels like a single prompt AI generated game (whether or not it actually is)

But, reguarding AI code ... a movie director directs others (actors, camera people, editors, set builders, ...) to make a movie. They don't do those things themselves. Yet we still ask the director about the movie they directed. I see AI as similar. I direct it. To me there is still value in knowing why a director made certain decisions and their process for getting the result they wanted.

Because getting slop is easy and slightly novel ATM, people post it and it's crap, but I believe some people will use AI as the tool it is and put in enough effort *directing it* to get good results and be worth respect.

> 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.

edit: OK I tried it again, got the gun, it autofires which still isn't fun. If I actually got to control it and had less ammo it would be more challenging. If that's a possible mechanic I couldn't see how to activate it. Still boring.

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.

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.

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.

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 (and probably with less than 4 hours' continuous effort from the human). 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.

Let us hope that the poster happens to see this, and is moved to comment, so they can address your points.

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.

suggestions: switch L-R directions like Defender dogfights with the red baron vertical and horizontal (have to stall through them) hoops ability to shoot up buildings levels with different goals... then you fly somewhere else to the next level. aircraft carrier landings other/faster planes

some of the text in the UI is too small, old games just had 1 font. braking at landing is confusing... too easy to accidentally switch to full throttle

That was relatively fun but there should be a penalty for flying upside down and backwards, it should lose altitude or something.

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.

This took me back to my first PC gaming experience at age 6ish. Circa 1988. Nostalgia unlocked!

Love the flight dynamics, super fun game!! Now can you make GTA VII??

Cool, but work with AI to figure out gravity and stalls better.

Anyone else remember Sky Kid?

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

Concept and visual look is great! Controls need a little work.

stall warning, angle of attack and wind dynamics in a balloon popping game, that is a lot of commitment to the flight model. respect.

What is the wind dynamics? I would have expected to climb when the headwind increased but I didn't. I also didn't manage to stall, only invert when I went out of bounds.

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.

What a cool game!

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 :).

The scan lines are a great touch. It makes it feel truly retro

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.

Awesome!

Badass.

nice!

Reply on news.ycombinator.com