One key difference: this guy built his demo from scratch, whereas mine's a port of someone else's work. It's great to see another implementation, with its own techniques and features.
Through the nineties, Microsoft software engineers carried on a tradition of writing clever and distinctive software projects alongside the software they were primarily tasked with writing— this is where Solitaire and Minesweeper came from, for instance. These engineers also wrote "easter eggs" into their primary software, such as the beautiful flight simulator hidden in Excel 97.
Amazed by reading the above. I'm a sucker to read and learn from that clerverness.
That's really neat! I like the tunnels under the giant trapezoids, they make me think of those enormous trapezoid buildings in Blade Runner.
I can't take credit for Drivey's aesthetics, I'm just honored to have carried it to HTML5 ;) Somewhere out in Australia is the original programmer who I'm sure would love to see this, but who also loves avoiding the spotlight.
I just did a quick scroll of your Twitter feed and jeez you're prolific, this stuff is really cool. I'll try Hexpress this weekend.
Oh, I think I remember the original Drivey and was following the dev blog for a while back then. But I think in the end, it didn't go anywhere (no pun intended).
Some people are making a strong case (I think) for calling these types of projects "demos", but also, I think the inability to fit them in the arbitrary structure of existing nomenclature is a good quality for a project to have. :D
Have you considered using a small bokeh image in place of circles in your renderer? Their size is based on the (usually unchanging) optical properties of the eye, rather than the eye's distance to the light, so you'd fade them rather than resize them, and if you round their position to the nearest pixel you might be able to draw them with 2D canvas context speedily.
interesting that this is running very slow (like 1 frame per second) on Chrome, but runs very smooth on Firefox. I'm on the latest version of both for Mac.
The original was written in "JujuScript", with strong types and novel operators. I pretended it was Haxe, and ran around fixing compiler errors it till it compiled to JavaScript. I built out a small Three.js project to hook it to, built up the features, then refactored it into something I felt was maintainable. I tried preserving the organization of the original script as best as I could.
Sidenote: I belong to the "port it" school of software preservation. My friend who runs the BlastEm project belongs to the "emulate it" school. I've seen both approaches have been employed to preserve Glider, which I think implies how important that game is to people. :)
That's really cool, but I wonder why on my machine is maxes out all the cores on my CPU and still runs at something like 1 FPS on high
If I turn it to Low it runs okay but looks.. not great.
But I don't see anyone else having issues so, is it just me?
Same here. I've also experienced it with other JavaScript simulations. I suspect it has something to do with Chrome's hardware acceleration because it works smooth as butter on Firefox.
Good question. The honest answer is that I don't know and that's the problem.
A compiler would tell me what all the types should be and if they are being respected. I'd also write a lot of unit tests to make sure that the code is doing what it is expected to as well as enable refactoring more easily. The code isn't formatted consistently, so that also makes it hard for me to read, I'm kind of OCD about that and having tools like eslint/prettier, which do it automatically, makes this super easy to fix. Linting the code would also point out other issues that the compiler misses. Putting it into CI would ensure that all changes get checked and builds would fail, if there are issues.
So I guess that is what I mean when I say 'clean it up'.
Update: Getting a lot of downvotes (yes, I know against HN 'rules' to comment about that), but I don't get why people dislike my response so much and would love feedback on that (similar to how the OP asked for honest feedback).
What I did was pull the source code, load it up in IDEA, convert the .js files to .ts files. Then looked at the errors. Most of what stood out was that the author included a copy of threejs directly and used parts that TS can't infer types on. Not a huge deal, but makes following the code more difficult for sure. I don't think it would take too much time to update it at all.
If you compiled me, you'd know that you are a type who I respect ;)
I think you're right on many counts, and I appreciate the effort you took to see what TS could do for this kind of project.
If you create an issue on the repo, I can notify you if I make an attempt at using TypeScript to maintain it, in a branch at least. No guarantees, but I've considered it before, and it'd be worth my time once I have some to allocate.
>There are a few more things that I think would be fun to do:
First off, it's beautiful! Second, just because you can, doesn't mean you should :)
I think it's art, so be careful about adding more. The minimalism is beautiful. Just a bunch of moving lights, but they capture the feel of night driving! If you add everything mentioned as upgrades, it will be a simulation. An impressive engineering feat, but (I think) less as a work of art.
I wonder, what does someone who's never ridden/driven in a car at night 'feel' when they see this?
Well it is a new movie, at least by period standards.
Broadly speaking you've got the B&W era until the 70s, the "old classics" recorded in colour on actual film up till the 90s, then the period of questionable CGI and campiness up till somewhere like 2005 when what we feel like is new/recent cinema starts. The ongoing era of decent invisible CGI, quality digital cameras, and post-9/11 hopelessness.
It's why I still watch a lot of 90s/early 2000s movies, there's just something different about that era that feels nice.
Having grown up with the crappy "Jurassic Park somehow pulled off what no other movie of the time did" CGI of the 90s, the effects in the Marvel movies look pretty seamless up to the point where people's fingers start shooting green lightning or whatever. The nature of their plot holes and overall dumbness feels different, also. Much more polish, to the extent that I can almost ignore how stupid the whole thing is when I'm watching one. Almost.
I think we get that "goddamn years are going by" feeling with movies especially because they are connected to a specific year yet are disconnected from other memories (unless you brought your crush to one of them maybe)
I remember Alt-J performing a cover of "A Real Hero", another song from the Drive soundtrack, at Glastonbury festival years ago. The film and soundtrack were really cool.
I can't tell you what originated the aesthetic, but it's not Drive because some songs on its soundtrack were already part of the established aesthetic.
It's definitely one of the major works in that world, however.
I was thinking of this song exactly before I turned on the music; thought it might even be the same song for a second. I wonder why the (visual) aesthetic is evocative of this song?
It's hard to classify what it is. It's not a video, because it's generated dynamically. It's not a game, because you just watch. It's not a screensaver, because it's not the 90s. Maybe it's a "demo"?
This reminds me of Desert Bus, the legendary/notorious driving game by Penn & Teller. During the second half of the game, you're driving at night and the view is not dissimilar.
Of course, because Desert Bus is literally the worst game ever, the steering continually veers to the side and you have to keep nudging the wheel or you'll crash.
>The entire scene is created purely by drawing circles on a HTML5 canvas. It actually works a lot better than I expected. The biggest flaw is that the cars are totally transparent, so you can see lights from distant cars even where they should be occluded by closer cars.
Hmm, why not just draw a black rectange around any pair of car lights? It should work for givimg the impression of a solid car within the context of the video...
Edit: hmm, he does say this "This would be slightly harder than street light occlusion. Probably a first pass would be to render a black cuboid behind each car's lights, so that the cuboid blocks out anything that would be blocked by the car."
Depending on your OS it should just be a matter of starting a webview with the code as the default page and running it full screen. Then a small loop looking for mouse/keyboard events so it knows when to quit.
That was my first thought too. I was going to post "Tsk tsk, going slow in the middle lane, making people pass you on the right" but then noticed it was in the UK.
Does anyone have good resources for creating "generative assets" like this?
I have a decent handle on js but I have no idea where to start in terms of tooling. For now, I've been trying to build basic eye tracking to apply a "block" of color over eyes with an animation in said block on top of a real-time video stream.
Elsewhere, automating mouth animations from audio / even moving a character in a random motion when audio input is provided.
Any advice where to start would be greatly appreciated!
Beautiful. This is something I’ve thought about too. Thanks for making it!
One of the most beautiful light effects I’ve seen while being a passenger in a car is the reflection of oncoming headlights on the underside of powerlines, whether they be mains power or for electric trains. It’s a dazzling pattern, reminiscent of the wormhole scene in Interstellar.
Perhaps something to add to your “more” list? :smiling-emoji-with-eyes-closed-and-sweat-bead:
Edit: At first I thought there were also stars, but it was just dust on my phone…
If this is what most people see driving at night I'm jealous. Astigmatism makes every light source at night a big star-burst for me. Oncoming lights are near-blinding at some angles, the only way to cope is stare at the road lines on the opposite side.
I was pleasantly surprised the article actually had something resembling Night Driver. Such a simple concept, but it was a game that stimulated your imagination. That and star Riders were brilliant for their time.
Night Driver (1976) was the first video game with a first person 3D mode.[1] It's about as minimal a 3D system as you can get. The picture of the car is painted on, not on screen.
I loved using the paddle. Atari controls were not very durable, and the paddle's potentiometer degraded into a simple switch, which Night Driver handled perfectly too.
> I found it actually worked better if I omitted the x offset from the distance calculation (so just sqrt(y2 + z2)), otherwise objects near the edge of the field of view were weirdly distorted. But I wouldn't suggest doing that in the general case.
I think this is because you should actually just be dividing by y, the perpendicular distance through the screen. Probably adding z to the mix, while wrong, doesn't make it too weird because of the limited set of z values used.
If the result with just division by y looks distorted, then I would suggest fiddling with the value of k, which effectively controls how wide-angle versus telephoto things look.
> "It's hard to classify what it is [...] Maybe it's a "demo"?"
Exactly, those are demos, and while I believe it's slightly different there's a whole culture around it that I've never been aware until recently called "demoscene"! I even have a small "Demos" section on my website with a bunch of those, not to the same level of quality though since for me it'd be a bit more like "self-contained small experiments that resulted in something cool so I put it together as a demo":
- "Tree generation", specify a JSON/HTML structure and it'll generate a tree: https://francisco.io/demo/tree/ (disc: it was inspired by a broken demo I saw from someone else)
I would have classified it as a "simulation". Which is what the author calls it, but then tries to find other things to call it too. Simulation is fine.
There’s a long-standing tradition of people creating realtime graphics software on personal computers that doesn’t offer any interactivity. They are indeed called demos and the community is called the demoscene. It goes back to the late 1980s and had a golden era in the mid-90s on Amiga and PC. (Pre-Internet, watching and making demos was one of the few socially and creatively oriented things you could do for free on a home computer.)
To be pedantic, this isn’t a full-blown demo. Small demos are called intros, and a category of them is the size-constrained intro (e.g. 1kB or 4kB). So this could be either a small intro or an effect as part of a bigger demo.
It’s worth looking up some of the small intros. People can squeeze entire GPU-raytraced universes with music into a few kilobytes.
IMO the author could very much submit this to a demoparty in the "demo" category (assuming they'd substitute the music with a fitting original work).
These days, the only notable difference between demos and intros is file size. If it's bigger than 64kb, it's a demo. You can totally have a short, single-effect demo like this one, and there's plenty such demos out there.
Ergo, I disagree, I think this counts as "full-blown" for any reasonable current definition.
Fair enough. The definition of a demo in the '90s was definitely multi-part: multiple effects with usually still graphics in between (such as group logos or pixel art rip-offs from fantasy paintings). A single-effect demo that wasn't an intro would have been rare.
Man...I need to get into the Demo scene. I have been disillusioned with the tech world because of just how utilitarian everything feels. I have been looking for something like this forever
I have many many childhood memories of long journeys up and down the country visiting my grandparents sat in the back of my parents car at night with scenes just like this etched in there. And I'm from UK too so was delighted they did this the "right" way round :)
Brilliant work, love this.
And the music reminds me of Kavinsky too with the obvious association with Drive.
Also, I know it's not actually there, but my mind insists on filling in a "not quite black" sky over the road and the occasional tree or similar to the sides as shadows. Fascinating effect.
Im not sure the cars and lights need to be anymore than floating orbs, its quite a nice minimalist aesthetic with just the lights. I'm not sure it needs to be a game either, maybe the ability to steer the car would be good but any more than that might seem forced (and I'm not even sure it needs that). The only real problem I see is that cars pass through you from behind (the only time the floating orbs effect becomes an issue), but even that seems minor for what it is.
> What game can we make where the premise is that you're a passenger on the motorway at night time? It shouldn't be a particularly taxing game, I think the main experience should still be that you're just enjoying watching the lights, but it would be cool if there was some interactivity and some sort of goal.
If you wanted to make it interactive, maybe a Pokemon Snap like functionality where you try to capture photos of random environment features or creatures.
A man running alongside your car window, jumping over obstacles. Technically this game already exists, but it would be interesting to combine with this demo.
It would be awesome to do this with a bladerunner theme. Like, sitting in a spinner going somewhere, with all the different cars, spinners and maybe buildings passing by.
It's not just the driving on the left that gives away that this is not in the US, but also the fact that there are lights along the highway and reflectors dividing the lanes. Ha, safety precautions!
I hate US highways at night, they are so dark, sometimes I feel I might as well have my eyes closed.
Apple Music, and presumably the others, are good about recommending stuff in this genre. I started with Emil Rottmayer's 'Descend'. Timecop1983 is another good artist to launch from.
Strong agree for something to work to. I was into vaporwave a few years ago but really love the whole outrun, synthwave, %wave stuff for working. Darksynth if I am really having to get after it!
> The biggest flaw is that the cars are totally transparent,
This seems easily fixable, no? Just make cars opaque black and assign progressively decreasing z-index to each spawning light. (Unlikely anyone will leave page open long enough to reach min value, and you could just reset at that point.)
Someone with your level of optimism has an advantage.
The tricky thing is, this project's renderer is currently a queue of circles to draw to a canvas. It's under 100 lines of JavaScript. So any increase in complexity will require substantial changes, like abstracting over types of drawables.
Exactly. Whenever drawing a pair of red circles for tailights, draw ~10 overlapping black circles between them. Bam. Done. I don't think this requires naive optimism, just 5 minutes of additional effort. Maybe less.
I'm not the previous poster but thought this would be a fun challenge. No way it was 5 min though; took me at least 20 min to understand what to change.
Anyways, run this in your browser inspector to hot-patch the live demo so that each car has an opaque black circle as its body/chassis:
for (let i = 0; i < cars.length; i++) { console.log(i);
cars[i].headlights = cars[i].headlights.concat([ {xy: new V2d(0,0), z: cars[i].headlights[0].z, r:cars[i].headlights[1].xy.x, col: "black"} ]);
cars[i].rearlights = cars[i].rearlights.concat([ {xy: new V2d(0,0), z: cars[i].rearlights[0].z, r:cars[i].rearlights[1].xy.x, col: "black"} ]);
}
Well done, nice job rising to the challenge! I definitely did not care enough to do it myself, so I applaud you. I just tried it out and it's perfect.
To be clear, since of course this is the internet and one must be precise or else get nit-picked and "outplayed," I obviously meant 5 minutes for the author who already knows the layout of the code. Obviously. Any charitable interpretation would have taken that as a given. 25-40 minutes sounds more appropriate for a newcomer examining it for the first time.
Oh yeah I totally get you, I just added the disclaimer about it taking me 20 min because I didn't want people to think I was trying to brag/flex about doing it in 5 min, which I didn't do and don't want to try to claim any credit for.
> It's hard to classify what it is. It's not a video, because it's generated dynamically. It's not a game, because you just watch. It's not a screensaver, because it's not the 90s. Maybe it's a "demo"?
Yes, you'd basically be stuck going well under the speed limit otherwise, because you will 100% always find someone going really slow in the left lane and oblivious to blocking tons of traffic. Even when there's sign after sign saying "Unlawful to use left lane unless passing"
The red line indicates the hard shoulder but yes, even after the law change here in UK I have not seen much actual change on the road tbh regarding middle-lane hogging.
People are either ignorant, lazy or don't care as there is very little chance of being caught.
Well, that's beautiful. I remember a procedurally generated night city skyline demo that was on HN years and years ago, this is a somewhat similar endeavor.
Very cool. One 'flaw' that immediately strikes me as cool is that the cars are transparent — so, for instance, when a car passes you between you and the oncoming traffic, you see the taillights, but also the oncoming headlights/streetlights are not obscured by the car. So it's sort of disembodied headlights and taillights, which makes the effect even cooler... fun!
Thats’s awesome. Kind of reminds me of this vector-y 80s game I had as a little kid - the Tomy Turbo [1]. I’ve been trying to make something similar myself but keep pushing it aside. Anyways, nice work - it looks great!
Weirdly I had the idea to do the same thing, including doing it in js, a year or so ago. This is much nicer than what I'd pictured. Mine would've been set to Kamasi Washington's Clair de Lune.
Might be a good thread to ask. A few years ago I found a couple of sites that were like shadertoy but for 2D canvas shit like this. One I think was codegolf.tk (which appears to have disappeared), and I can't for the life of me remember what the other was.
> It's not a screensaver, because it's not the 90s
What's funny is that screensavers may end up coming back into vogue if OLED displays continue to have burn-in issues. Ray-traced flying toasters may be in our future.
I enjoyed reading this as much as I enjoyed watching the demo. Both the demo and the writing style are simple and calming, I didn't even notice that the cars were on the wrong side of the road until I saw the TLD ;)
Not exactly the same thing, since it's a high-paced action game, but I was always fond of the DOS game Skyroads, which has a similar aesthetic. (I believe someone has now made a web-based version called OpenRoads.)
It would be really neat to create a variation of this that simulates driving through mountain roads -- starry night sky, dark mountain silhouettes, the occasional cabin light...
I remember when I was very young I wrote something much simpler using PRINT and GOTO.
Hand drawn formula car (drawn using /\ [] and |) would scroll indefinitely in a simple loop.
Some people run their browsers with JavaScript turned off. Some folks disable CSS. Some even browse via Lynx, a text web browser. It seems like the author's met some of these folks.
> It's hard to classify what it is. It's not a video, because it's generated dynamically. It's not a game, because you just watch. It's not a screensaver, because it's not the 90s. Maybe it's a "demo"?
BTW, while we may think that driving is a universal thing, I instantly found it amazing how North American this is. From the integrated indicators, to cars passing on your right, to the road lights and even how the road undulates. Having said that, great project!
Edit: Well I got this entirely wrong and missed that this was meant to be lefthand traffic. (I caught a segment that looked more like there was an independent road passing an interstate crossing a city, rather than being opposite lanes.)
Depends on where you are of course, but reminds me of approaching Stockholm, just before it gets so thick that people start loitering in the left lane. Perhaps Salem, if traffic isn't too heavy.
Erm, the cars are driving on the left (red taillights on left side of motorway, white headlights on right)... wouldn't North America be the other way around, with cars overtaking on the left?
I happened to watch a segment where the cars on the opposite lanes looked more like passing on an independent road, so I missed that this was meant to be lefthand traffic. And it really looked more like an interstate passing through a city. – My failure.
Yes, but the cars are driving on the left in this demo (like in the UK), so on the right is where you do overtake legally - in the faster lanes. I haven't noticed any "overtaking" on the left, where it would be illegal in some countries that drive on the left in the few minutes I've watched it.
Why do people like to buy skins on Fortnite or Call of Duty? People like to own things to show status, taste or just express themselves.
If the question rather is, "Is owning an NFT the same as owning a physical item?". It's not exactly the same, but it's close enough to feel the same emotionally (you and only you can decide whether you want to transfer, sell or even destroy it). And it's strictly better than owning any other digital license because it's on a neutral platform with deterministic rules.
With respect to generative art in particular, it's a great fit because it allows a someone to own a unique instantiation of the algorithm. e.g. take a look at this collection: https://www.fxhash.xyz/generative/slug/take-wing you can create infinite variations on the same theme, but the artist has determined that they only want 250 pieces in total to exist. So when someone "buys" the art work for the first time, a completely unique piece of work is generated based on the random transaction hash. The algorithm still exists as a whole and anyone is free to view and enjoy it, but people still buy and trade the unique pieces because they enjoy it.
In fact, there's an upcoming project that really leans into the idea: https://twitter.com/tylerxhobbs/status/1571908670929133568?s... Anyone is free to create any number of outputs they like and share them. But only 999 NFTs will exist, some people people want to pay for the right to have their selection immortalized as one of the 999.
I don't understand the visceral hate (not from you, just in general). People who want to just look at it are free to do so, people who want to pay and own it do so, and the artists are paid directly (and get perpetual royalties). Who's losing out?
Anyway, my original comment was not even about NFTs, just wanted to bring attention to the fact that this is generative art, and other artists have similar posts about their work and how they use JS to replicate similar real world phenomena.
edit: I didn't realize you were the post author. Your work is amazing. I'm sure it will be appreciated by generative art collectors should you choose to publish it. Feel free to reach out to my username at gmail.com if you have any questions. I'm a serious gen art collector, and there's nothing I would love more than to introduce new folks to the space.
One key difference: this guy built his demo from scratch, whereas mine's a port of someone else's work. It's great to see another implementation, with its own techniques and features.