I haven't looked up the history, but I would imagine that the "XL" is for the fact that it can handle images up to 1B x 1B pixels (1 exapixel).

Both lossy and lossless, and so much more.

I don't know what field you are in, but in my area, math, this is wildly not the case. They are much, much better. And the next line of the public frontier models are obviously even better judging from everything that we've seen drop from OpenAI and Anthropic. The large OpenAI release would by itself would have four or five results which would arguably be a Fields Medal if a human had done that, and quasi-RH which would if a human did it be almost definitely a Fields Medal result.

It's this.

And on the compression side, there's a lot of really cool stuff that lets it happen. For example, it dynamically resizes the blocks from 2x2 to 256x256, and they don't even have to be square. It may apply DCT in one, AFAS in another, etc - whatever is most efficient. It uses edge preserving filters to keep the edges of features clean. It can draw in splines and clone regions from one part of an image to another. Where it has to throw away high frequency noise (film grain, noisy textures, etc), it statistically samples the properties of the noise and saves that, and then regenerates equivalent noise. In "busy" areas, like say tree bark or whatnot (where the human eye isn't good at discerning changes), it throws away more precision there, and instead allocates it to details located in fragile flat areas, such as skin and shadows, that the eye is attracted to. It works in default in XYB colour space, based on the actual interactions of the cones in the eye, so it can aggressively quantize things which the brain cannot detect. And in the choice between fast but not-great Huffman coding vs. better but slow and patent-encumbered arithmetic coding, they instead use rANS, which is nearly as fast as Huffman but as efficient as arithmetic coding, and faces no patent challenges. Which is also why you can take a legacy JPEG and rapidly losslessly reincode as JPEG-XL and get a file that's ~20% smaller - you're swapping Huffman for rANS.

(That's the lossy mode - there's also a dedicated lossless mode)

yeah the Space Age technology of some screws and an O-ring were lost technology to the wearables space for the past 15 years... along with replaceable batteries that that allowed.

Just because it has a replaceable battery doesn't mean it's user replaceable.

Watches and such you often had to take to a watch repair shop to change the battery because they could make sure the O-ring didn't degrade as well as ensure it still retained waterproofness by making sure it stayed clean and likely lubricated. Heck, if you read the warranty they often voided the waterproofness guarantee once you pop open the lid and it was the company that replaced the battery who provided the guarantee if it stops working.

Sure, it's easy to do, but often you automatically lose the IPxx protections once you open the lid. Everyone accepted that because that's why you paid the watch repair company to replace the battery for you.

From the article: "Supermarket cold aisles are typically run by control panels that manage an entire store's fridges and freezers and can be accessed over the internet, which lets maintenance companies adjust them from anywhere."

Well, there's your problem. Stop doing stupid things like that.

But then if something goes wrong, you realize there's no one in the store who can fix it, right? The store manager would have no idea what the panel does. They're going to have to get a technician over to fix whatever it is.

Maybe you haven't done the whole "drive two hours to flip a power switch" thing, but you'll find the vast majority of issues people have are trivially easy to fix things like that. People don't interact with the panels, other than maybe accidentally by bumping into it or something and then changing a setting. You don't want someone who doesn't know what they're doing (i.e. a teenager who bumped the control panel while bringing in more pop tarts from the back) to fix it. But you also don't want to call in a $200/hr technician to push the button to turn it back on as well.

Those things pretty much have to be remotely monitored and often adjusted remotely because there will not be anyone in the store who knows anything more than "The panel has a beeping red light on it".

Granted, the security should be better - VPNs and such aren't terribly expensive things, and any reasonable supermarket with a backend inventory system already has a VPN to corporate so the HVAC techsn can access every panel through the VPN that already exist.

(One should also note this happened in France, and shopping habits in Europe are quite different from the US - namely in that many people shop daily - they pop into a really tiny supermarket for a few bits of produce and such and 5 minutes later they're back out with dinner ingredients. The weekly "big shop" you do in the US isn't really a thing.

When you take the time to understand the AI-written code ( often using the AI to explain its own code ), there is no problem and you can proceed significantly faster than without AI.

The problem is that that work to gain understanding is boring.
While I often had to do it in the days before AI, I did not become a programmer to spend all my time understanding code I did not write myself.

True, however I have used AI to help me fix bugs in code I didn't understand. But I didn't let it make blind PRs for me. I had it make the changes, then I reviewed all the lines that it modified and asked it to fix lines that looked wrong to me. It produced working code, but I didn't like the way it initialized a structure that was central to the process. I had to ask it to fix that by using what already exists because it created a maintainability nightmare.

But I knew that not from understanding the code at a deep level, I knew it from my years of experience that code that looks like that would lead to problems years down the line.

Now, since I was unfamiliar with the details of the code, I did have it produce documentation about what it did, as well as how it worked and other things.

If I did it myself, it would probably take over a month. But with AI, the task itself was complete in about 3 days, but it took me another week to figure out what it did and carefully review the code it changed and how to fix its issues. By the time I was ready to submit it for QA testing, it was a full two weeks.

I don't mind using AI, but I'm not the kind of person to use AI blindly. My coworkers have asked why I do stuff manually - and it's because it's faster for me to do it manually than for me to figure out how to tell AI how to do it, let it do it, then understand what it did so I can submit it. AI just becomes another tool in my toolbox. Heck, I still coded up a few things from scratch - it did not occur to me to ask AI to code up a trivial tool I needed.

What it really shows is just how young and unknowing new developers have become to choose a name that has an obvious history.

Or those who fail to learn from the past. I mean, people seem to keep reinventing the wheel over and over again, and with AI, they're doing it faster than ever.

It's like "I invented something new!" makes big news, when in reality you just reimplemented something old.

This is especially true in the retrocomputing arena, where people would reinvent the technology we had over again because they didn't realize we had a solution back then.

It's only a matter of time when people playing with a 386 or something reinvent HIMEM.SYS or EMM386.EXE and think they're the smartest in the world.

Especially because AI makes people dumb in that way.

You have no idea (and it is designed that way) how many of these systems exist or where they are located. Flock cameras are made to be easily seen. Others are designed to be hidden from public view.

They've been around for decades. The land borders have ALPRs that read the license plate for everyone entering the US. They're also used on toll roads - the ALPR is on the gantry you pass that reads your transponder. If you don't have a transponder it snaps a photo so they can bill you.

If you ever parked your car there are mobile ALPRs that check to make sure you don't park too long (since chalking tires is a 4th Amendment violation). This applies whether you're parking on the street, or in a parking lot - the city has mobile ALPRs as do the companies that are in charge of parking lots making sure you paid and your parking is still valid.

Flock is more than a decade old - they provided a service for private businesses to have ALPRs monitoring the parking lot of HOA streets.

The main problem is Flock is treating it as a network - little isolated silos of ALPRs is not a bad thing - and for decades that was the norm and no one really cared or minded.

But Flock treats it as a network - the bigger they grow, the more monitoring they can do. And then it becomes their advantage that they can search any vehicle from anywhere. But that goes way beyond what the cameras were initially installed for.

ALPRs that are isolated aren't the problem. It's when you start linking them together that things become problematic. Just like how individual government databases are fine in isolation, but things get scary when you start linking them together. You give the IRS lots of personal information, with the expectation that they will block access to that information from other departments.; Ditto the census and other agencies. While it seems utterly repetitive to keep providing the same information to each department, by keeping each database separate and isolated, it provides a lot of privacy.

But when databases are linked, things get scary and you end up with surveillance. It's why data brokers are a concern because they are buying up data from everything and then linking them together.

Frankly, I posted my question in hopes that some bright boy had come up with a better approach than setting up (and managing) our own CA.

Putin has to wake up smiling each day since November 2024. If Trump had lost the election, Russia probably would've lost this war by now... and Putin, his life.

Personally, I don't believe that Trump is an active Russian asset, controlled through blackmail or some other means. I think he's just an incredibly vain and stupid man who Putin can completely manipulate with ease. I mean, come on - how does a phone call that's supposed to be about pressing for information on that pneumonic plague death turn into "Hey we're loosening the sanctions and accepting 2.5 million barrels of Russian petroleum"?

I remember nothing of the sort. But I also don't pay attention to people saying things that are obvious horseshit.

Even if the actual Amazon was blown up, life wouldn't stop, and nobody but Amazon's insurers would be "defeated" - you are presenting a false premise or some other kind of bad-faith argument nonsense.

And it's not just the US - There has been a measurable reduction in bot noise among many secessionist movements across the west, including Poland, Quebec, and Texas.

They would be lucky if they actually got the pipes. It's far more likely the contractor took the money, chipped off a little to the inspector, and did absolutely nothing at all.

Good job, Comrade. Your rubles will be deposited just as soon as Yandex gets back up again... as long as Ukraine doesn't keep blowing data centers into scrap.