I thought it was pretty well established that JXL has a big room for improvement when it comes to encoders. There are some fundamental limitations wrt the absolute lowest bitrates, but that the room for improvement in encoders is rather substantial, especially with things like rate-distortion search heuristics.
The comparison is more about who has had the most man-hours and megadollars put into encoders quality than about the inherent limitations about the format.
With that said, I use AVIF, but I do have what can be said to be an irrational preference for JXL.
> but I do have what can be said to be an irrational preference for JXL.
Is it irrational to follow the collective "one format to rule them all" dream? Sure, the codec itself needs work, but seeing how long we've been stuck with the JPEG/PNG/GIF trio (even with better encoders like mozjpeg/jpegli or optipng/zopfli) with TIFF as a catch-all backup, getting the format right once and for all really matters.
I'm still weirded by the low amount of workforce in libjxl/jxl-rs; sure, too many cooks do ruin the broth, but too few too, in that domain.
Do you also happen to support the Motorola "shake to turn the flashlight on".
Every time I buy sometning that is not Motorola I regret it, and it is like 85% because of this feature. I don't know how they do it, but it triggers reliably, never misfires.
People always suggest apps that supposedly should allow the same thing, but they never work, which makes me think Motorola uses some lower level sensor access to do it.
I've used a lot of smart phones over the years. They always, always, have a pile of supposedly useful, ergonomic, gestures or shakes or whatever that do stuff. I have never actually reliably started using them and stuck with them until my partner got a Motorola device.
Shake to flashlight is the perfect combination of easy to remember, easy to trigger, and actually useful. It should be industry standard.
There are 2 features I missed when I went to Pixel phones from Motorola: shake on flashlight and twist on camera.
On a hike a few moons ago, as I approached a scenic overlook, another hiker pulled out their phone, twisted it, and took a picture. We had a brief conversation about our love for Motorola gestures.
If Motorola comes out with a lower priced phone with GrapheneOS and a Pixel quality camera or better, I'm there.
So, many people dont know this, but if you ever meet someone who says measles is natural and strengthens the body tell them about immunological amnesia. Measles in unvaccinated people will damage the immune system for a long time.
People that have had a measles infection will have a significantly higher rate of infection for something like 2-3 years post-measles. You can get things like varicella again. Even EBV (missing disease) can re-emerge, and often sensitivity to influenza increases dramatically.
If you have Bern vaccinated against polio, you will usually need another shot.
A friend of mine, an immunologist, use to say: the idea that an infection makes you stronger only ever (if at all) applies to bacterial infections. With viruses "the fewer infections the better" is true enough that I can stand behind it in my professional role. Even common virus infections can mess you up for a long long time.
My question is, did it provide POCs? I have had some security issues flagged that I tried my best to exploit, but failed miserably. Since I do not have any kind of privileged access the LLMs I use have been unwilling to actually prove that it can be exploited.
While I did fix most of the things because while they might not have been exploitable they were bugs waiting to happen, in all but one case I haven't been very impressed by any of the LARGE IMPORTANT SECURITY HOLES that were found.
No, the Mythos reports do not come with POCs. It's a lot of redundant prose explaining the bug it found (much more than I'm willing to read), it explains how it can be triggered, but doesn't generate actual exploit code.
But that doesn't matter for me. It shows me real bugs and less than 1% false positives. Some bugs are truly nasty and urgent, most are trivial and not exploitable at all, but they're bugs and so I'm fixing them.
Mythos isn't mythically superb for being the best LLM model, but only for not having these annoying guard rails that make all others randomly refuse to cooperate when you need them the most. Whoever invented guard rails - please stop! - they just slow the good guys down in the race.
While it is certainly good to do estimation oneself and not just blindly follow whatever the LLM outputs, keep in mind the recent PHP-related events ... Just because at some point one is not able to make a PoC, it doesn't mean, that no one else ever will be.
Wasn't that the correct decision? They put a deprecation warning on it to inform people it was unsafe in server context, while also keeping it.
It was unsafe and surprisingly easy to DOS servers using it incorrectly. And by "incorrectly", I mean how a lot of (most?) people used it with user data. An API that happens to be safe by accident when people use it should be deprecated.
The leader can be a lesbian and still far right. They are against gay marriage, against talking about homosexuality in schools. There were gay Nazis, even high up. Röhm for example.
Iirc Weidel herself has made statements regarding how the society should always promote heteronormativity.
He aligns with AfD on issues where AfD is with majority of the population (stopping mass migration, some remigration) but not on all policies which make them far right.
Also more than 30% of Germans support AfD these days. Hard to call that "far" anything.
That is the saddest part: there is already a fantastic green thread concurrency for dotnet: hopac. F# then got Task {} and people kind of left hopac behind.
ConcurrentML really brought dotnet's concurrency story into the 1990s.
The comparison is more about who has had the most man-hours and megadollars put into encoders quality than about the inherent limitations about the format.
With that said, I use AVIF, but I do have what can be said to be an irrational preference for JXL.
reply