These are what these companies' respective browsers/operating systems use to check if they have an internet connection and are not stuck behind a captive portal. Given the enormous quantity of devices out in the world that are relying on these URLs to stay up, you can probably feel comfortable using these in your tests. If they do go down, good chance you'll see news articles about it before you notice your tests failing.
If they ever figure out how to get Linux onto a phone with a good camera, I'd pounce on it. I would be willing to put up with app issues and weird bugs and all the jankiness that comes with early adoption, but I simply can't give up being able to take really good photos with my phone.
Those are probably the best I've seen from a Linux phone, but still far too much of a downgrade for my uses. Even some of the photos taken in decently well-lit locations are noticeably grainy, so I can only imagine how poorly it'd fair in worse lighting. Tons of props to this photographer -- I have great respect for the art of getting good shots out of low-end equipment, but it's not what I want for my daily carry device.
I'm not exactly a photographer, these are just random point-and-shoot photos I took with my phone. I am, however, a programmer who has spent a bit of effort on making the photos taken on this phone look reasonably good, which is all it took to get to "the best you've seen from a Linux phone" state. Imagine what someone who actually knows what they're doing could achieve there ;)
The reason some of them are grainy is that there's no denoising applied to them at all. The current GPU-based pipeline only does debayering, lens shading correction, highlight clipping, color matrix application, tone mapping and barrel distortion correction so far. You get the raws though, so you can apply whatever you want post-factum. Patches are welcome too.
The majority of the photos on the first page came straight from the phone with no post-hoc processing whatsoever (and if you click on the post about Millipixels, there's more of unedited ones in the replies). Only tram, train and kernel ones had some, and even then I don't think opening the photo in Darktable and adjusting a few basic sliders counts as "huge amount".
In that thread you're stating just months ago that Librem 5 owners no longer need to manually develop raw photos to get decent results. Truly atrocious video quality. I say the camera's not good enough for me, and I suspect the market agrees.
The tram photo is arguably the best output of the lot and it's got a ton of problems by contemporary standards. Fun photo, if you didn't buy the phone for the camera and you want to hot rod it to get the best you can out of it, great.
Send that to any of the typical phone reviewers and the camera's going to get panned, probably universally unless someone wants to stake out a contrarian position for publicity.
"It's not bad for a Linux phone" isn't the expectation, we need a genuinely reasonably good device with good components.
> and it's got a ton of problems by contemporary standards
No shit Sherlock. A processing pipeline written in a few weeks by someone who was just learning what color actually is while writing it, with lots of TODOs left around, can't compete with the state of the art? Who would have thought?
The post that started this subthread simply asked "Is this a good enough camera for you?", which is a valid question. For some (like me), the answer will be yes; for others, no. That's fine. I appreciate this link being posted because there's lots of really bad photos circulating out there coming from misconfigured software stacks that don't even do the most basic things you're supposed to do with pixels to get a viable picture, giving a much worse impression of hardware than deserved simply because of poor software. It was you who then suddenly showed up with your false "massive amounts of post-processing" claims and iPhone comparisons, as if anyone was claiming that it was comparable to iPhones. Take a short break and re-read what you wrote and what you've been replying to.
>> it's got a ton of problems by contemporary standards
> No shit Sherlock.
You're supporting my point. A device which first shipped seven years ago still requiring someone to properly configure a post-processing software stack to get viable pictures out of it (with no path to viable video) kind of makes my point for me. That the hardware can perform significantly better with post-processing is more of an indictment of the overall device than it is an endorsement of the camera hardware itself.
For me, from a camera standpoint (among others) the device isn't good enough to be a viable alternative to switch from Android (the post's topic), or from iOS (likely the most commonly used device family by HN readers). Even with your stack. As I said, your photos are the most convincing I've seen, at least the ones in the most forgiving circumstances. It's not personal.
As switcher targets go, the Librem 5 is a poor one. Ask any normal person to do what you call basic to get them looking like a midrange camera from a decade prior. And yet, when switching comes up, the same few suggest it as if it's a genuinely good option for anything but very niche cases.
> A device which first shipped seven years ago still requiring someone to properly configure a post-processing software stack to get viable pictures out of it (with no path to viable video) kind of makes my point for me.
Except it doesn't. The photos I showed are as they come by default out-of-the-box on the default OS the device ships with. You're just making things up again.
The (hardware) camera quality is about what you'd expect from a phone from 2019. Software processing to get JPEGs out of raw data is worse (with a huge room for improvement), but you get the raw DNGs and can employ any processing you want afterwards. Video recording, however, is the real Achilles' heel of the Librem 5, at it lacks both hardware ISP and video encoder (there are only hardware decoders). This means that any real-time processing has to happen on the GPU (replacing the ISP) and CPU (replacing the encoder). I can record indefinitely at 526x390 30 FPS and I expect 1052x780 to be achievable at a reasonable framerate with more work (currently it's at about 15 FPS IIRC), but I don't expect it to ever go significantly better that that. Of course, short clips (that fit into the RAM) and low FPS "timelapses" are possible to record at much higher quality, but these are fairly specific use cases.
That said, what's there already is enough for me to have plenty of fun with the camera anyway.
A Camp Snap is more than good enough to have fun with. That's an entirely different requirement.
I had an iPhone 11 Pro Max in 2019 and its camera quality was miles ahead of the linked photos, even if you assume raw output and post-processing. (The video, as you note, is even worse.) The link provided are shill posts of photos shot mostly under highly favorable conditions, with massive amounts of post-processing happening with much newer tech. I'm looking at casual photos straight out out of my 2019 device and they're vastly better than these.
Linux phones aren't going to get better by pretending the experience is something it isn't, and telling people not to believe their lying eyes. If everyone's cameras were suddenly replaced with the Librem 5's, the number of camera quality complaints would explode.
It's not good enough for anything beyond novelty use.
The link provided points to my random photos and videos I was shooting in varying conditions as I went through my life, posted on my personal social media account, usually with minimal to no post-hoc processing. Not sure what is up with you here.
Every time someone says we need a better open Linux phone, the Librem 5 and PinePhone crowd shows up to tell us not to believe our lying eyes, that our standards (which are already more open to tradeoffs than the general public's are) are wrong, that they love daily driving a device and therefore extrapolate that into some kind of viability as a commercial offering for anything other than novelty hobbyists.
They're fun photos and some of them are of pretty decent technical quality considering the device used, but switching from iOS/Android to Linux phone of the Librem 5 or PinePhone (/Pro) variety is a bag of hurt for anyone who isn't so committed they're willing to craft their personality around it with an almost religious-like zeal.
That's what's up with me here. Still, I appreciate your photos.
Edit: Check the comments. You've got a guy saying the PinePhone Pro is "there." Except, it's bad as a phone. Minor detail.
I stated the camera isn't up to snuff for me, your photos are the best counterexample anyone's posted, and they still aren't up to snuff for me. Also, decent video isn't optional in 2026. I remember upgrading a BlackBerry to finally get decent video. The whole thing is so fundamentally constrained that it will never be more than a niche novelty, barring new hardware and massively better supported.
It's just as if people weren't a single hive-minded entity and they all had differing needs and opinions.
Of course we need better phones. Librem 5, the one I personally still consider the best out there right now, is now 6-7 years old! It wasn't a low-end device back then, but it wasn't high-end either, and time flies. Yet at the same time I keep seeing people going "I wish we had XYZ" when XYZ was something I was enjoying all these years already. And you know, these better phones won't come out of the Santa Claus' bag. If all the people complaining about the need of better Linux phone had actually bought one, we would already have their successors in our hands.
There's still a lot of work to do, but the usual reaction when people are playing with my phone is being surprised at how usable it appears ("this is a Linux phone? it looks like a normal one!"), so I do think we need to put more effort into informing people of what's already there, even if it won't satisfy everyone's needs.
It's difficult to frame it as something that was mid-range in its time, as many, maybe most original backers probably didn't receive theirs until much later. COVID certainly compounded things and they can't be faulted for that.
Purism was a case study in how not to handle delays, and faulting consumers for being overly picky or not supporting the product financially ignores the broader context.
Some people may also object to buying a phone that struggled for years to...make phone calls. (And by some contemporary accounts still does.) It takes potato quality videos. It was an absolute mess as of 2025: https://forums.puri.sm/t/it-is-normal-for-the-librem-5-to-ha...
Its existence does not obligate anyone wanting a genuinely good device to exist to contribute to that project, and it's naive to think a lack of sales is what's prevented Purism from shipping a better device.
> as many, maybe most original backers probably didn't receive theirs until much later
Original backers got it in 2020-2021. It took until 2023 to catch up with preorders made already after the original campaign ended.
> Purism was a case study in how not to handle delays
Agreed.
> faulting consumers
I haven't faulted anyone. I just made a "if I had a dollar for each time..."-style remark.
> It was an absolute mess as of 2025
You linked to a post about a faulty unit and about how another unit with exact same software didn't have the listed problems...
> it's naive to think a lack of sales is what's prevented Purism from shipping a better device
It's exactly what prevented Purism from shipping the successor so far. It's also what prevented it from keeping a whole team of developers hired that worked not just on the specific device, but on the entire software platform and whose work is now being used across various devices. The Librem 5 project was a massive cost sink that left its fingerprints all over the place across the GNU/Linux stack, and more sales was exactly what would be needed to let it continue at that pace. The reasons why it didn't sell well enough can be argued, but the inarguable fact is that more sales would bring more progress.
Neither does mine. When I'm not rebooting into new kernels and don't let the battery go flat, I reach several weeks of uptime (eventually interrupted by one of the things mentioned).
Apple operates http://captive.apple.com, which returns `
Microsoft operates http://www.msftconnecttest.com/connecttest.txt, which returns `Microsoft Connect Test`.
Mozilla operates http://detectportal.firefox.com, which returns `success`.
These are what these companies' respective browsers/operating systems use to check if they have an internet connection and are not stuck behind a captive portal. Given the enormous quantity of devices out in the world that are relying on these URLs to stay up, you can probably feel comfortable using these in your tests. If they do go down, good chance you'll see news articles about it before you notice your tests failing.
reply