@pbmonster's banner p

pbmonster


				

				

				
0 followers   follows 0 users  
joined 2024 May 13 11:54:07 UTC

				

User ID: 3048

pbmonster


				
				
				

				
0 followers   follows 0 users   joined 2024 May 13 11:54:07 UTC

					

No bio...


					

User ID: 3048

For casual snapshots of groups of friends etc. phones are great.

To be honest, I mostly take photos of my kids, and sometimes of people climbing/skiing. Both situations can have challenging lighting and object that don't stop moving around. The phone software has been absolutely amazing at eliminating motion blur and/or underexposed images, something I've previously struggled with even at 1/120s. And yeah, I often miss having the tele lens, but I've gotten used to moving in to take the shot - or with having a shot of nice landscape that has some action in it.

Most importantly, the ergonomics of taking photos on a phone are shit tier compared to any halfway decent camera that has a viewfinder and where the body has been designed for the task of taking photos.

True. Getting a phone with a hardware shutter button is absolutely essential. The rest can't be helped, I think.

I've gone the other direction and bought a real camera a while ago. I got fed up with the overcooked processing phones do.

I know what you mean. The good thing is you can turn that off - either feature by feature, or all of it. Or use an alternative camera app if you want to set exposure and ISO yourself (and those apps aways only produce traditional stills), export un-edited stills from the short videos the main camera app takes before it starts AI-editing them, or tell it do AI-slopification by default but also always save RAW images. At least that's the state of the art on Google Phones.

Luckily there have been good lightweight mirrorless cameras on the market for over a decade.

I'm sure for many use-cases a modern mirrorless takes far superior pictures, especially when used by a experienced photographer. But the AI has been amazing for normies.

It's good enough to tell similar but distinct varieties of flowers apart too.

In my experience, Google Lens generally can't, but apps like Flora Incognita (which instructs you to take images of the leaves, the flower, the stem, the bark, ect.) can. Flora Incognita also tells you a certainty percentage, which is really helpful.

In my garden, Google Lens has an almost comical inability to distinguish my carrots from yarrow - and it won't warn you that it's less than 50% sure. If you only feed it flower pictures, flora Incognita has trouble as well, but tells you it's less than 40% sure until you take pictures of the leaves and stem.

I have barely changed how I use a smartphone since I bought my very first over a decade ago

I have the same feeling. But for me, something else changed: the phone takes on more and more additional duties. Within the last two hardware generations, I've completely stopped bringing my DSLR camera and my outdoor GPS unit (for mountaineering).

Modern smartphone camera sensors and lenses are decent (when compared to compact cameras), and modern camera software is - frankly - completely insane. The combination of the two now easily beats my skill level on a DSLR camera that has orders of magnitude more sensor area, lens diameter and aperture diameter. I'm generally a software skeptic (progress in software development over the last 10 years has resulted in very little "real" value being created), but camera software amazes me. Instead of taking a still image, phones now always capture a short video instead and distill the final image in post - a mostly automatic process that results in sharp, correctly lighted and color balanced shot.

Replacing the GPS unit was more trivial. OLED displays are more readable in the sun, the GPS chips got a bit faster, and again, the quality of the software/data got orders of magnitude better (the free offline geo-data available today is vastly better than the commercial data of a few years ago). Route planning and terrain analysis also got so much better. Used to take a PC and skill and experience, now everybody can do it on the phone with 30 minutes of instructions. Also, if you have the right smartwatch, you won't be taking a device out of your pocket at all anymore.

Is this true? Perhaps my perspective is skewed by living in San Diego

Probably, it's the location I would expect to look the best in that respect.

I personally know dozens of people who are fully fluent (in the sense of being able to competently converse about a wide range of topics) in both Spanish and English. When it comes to second-generation Latinos in most parts of the country, or at least in the Southwest, my perception is that bilingual fluency is actually very high.

In my experience, many of those people actually can't e.g. do their standard white-collar job in their second language. If you want to have a country with dual national languages (as opposed to making Mexico an imperial possession as someone suggested below), you need a lot of people who can do that well, since a lot of national/federal institutions need to be run in both languages.

At least you would need those people, traditionally. A lot of that might shortly be superfluous, since language models work well across languages and federal institutions might be a thing of the past!

But just imagine being in the Army, and working alongside an integrated Mexican auxiliary battalion - or integrating US special forces into a Mexican-led theater. You'd really want at least everybody above O-2 being bilingual. Messing around with interpreters under fire is pretty much unacceptable.

I agree Mexico is the much tastier target. In my personal assessment, Mexico is more culturally compatible with the US than Canada.

That's an... interesting proposition. To start, how do you expect the language integration to work out? Just have dual national languages? National languages on the state level? How do you feel about Spanish slowly (or not so slowly) creeping north, possibly displacing English in the southwestern states within a few decades?

Language is extremely important for the national consciousness. And unfortunately, both the old stock and the new citizens don't exactly have a great history/culture of bilingualism. The number of people being actually fully fluent in both languages is currently extremely low (when compared to existing countries with multiple national languages).

My prediction is you'd have independence movements solely based on language, and quickly.

Sure, but unlike the Suez, there's mountains/hills in Tehuantepec. Hundreds of feet of elevation difference, a canal would need many dozens of locks, maybe hundreds.

I can see a high capacity rail line, but digging a canal to rival the one in Panama is madness - especially as long as the one in Panama exists, and acts as economic competition.

Yeah, I was seriously considering picking one up for cheap in order to benefit from the image damage the brand took from the recent politics... but the cars just seem to suffer from inconsistent quality.

Some drivers really love their Model 3, have zero problems for the first 100k miles and give glowing endorsements to everybody who will listen. Lots of taxis use them, which is usually a good sign. Others are at the dealership all the time, and often for dumb expensive stuff.

Coming from Toyota, I don't like the risk. I expect my used cars to be solved problems.

For quick object identification while looking at the night sky, I like Google Sky Map. Point your phone at the sky, and the screen will show labels for everything you see. When it came out, it was one of those magic moments of modern technology for me. It uses GPS, the phone accelerometers and the clock to know exactly what you point at, and the UI is pretty.

If you get her hooked, there's no way past Stellarium for proper nerding out during the post briefing. You can put your point of view anywhere, run the time forward and backwards, have planets/starts leave traces as they move, ect.

Sure, but there's always false positives, the only question is how low the rate is.

And your kids are going to live with you for 18 years, so you need quite a lot of 0s to make reasonably sure the turret isn't going to blow their heads off by accident.

It's a difficult problem. Maybe one brings over a friend. Maybe they wear silly monster masks.

The tech exists. (Semi-)automatically killing/maming people really isn't that hard. It's different levels of illegal, depending on locality, but the real problem is false positives.

Most people just don't want to live in a mine field, no matter how many fancy hightech safeties there are.

No, if your opponent shoots himself in the foot (and splatters blood on you, which is irritating and expensive), you respond by shooting him in the kidney and try to dodge the blood spray.

When the US put tariffs on Chinese steel and aluminum, the Chinese put tariffs on American soy beans, which directly hurt the republican heart land and didn't really hurt their own farmers all that much - Brazil was still selling soy beans after all.

Effectivity, this is a step beyond the standard prisoners dilemma. You can choose where you defect, and the gain/loss can be very asymmetrical. It also can be more than monetary gain/loss, since the payoff might be in geopolitical position or voter opinion instead of trade balance.

The Euros keep failing them in large numbers in their road-worthiness inspections at around 4 years. Depending on the exact country, it seems 20%-30% have "substantial deficits" which require major repair work. Worst EV in class, every single time.

The biggest problem is certainly brake rust from under-use (which you can mediate yourself, and Tesla could probably fix that by software update), but the reports I've seen also all mention suspension problems and faults with the headlight systems.

I wanted a used Model 3, but major repairs at 4 years is kinda scary. I've driven my Toyotas all well past 15 years of age, and I'm not confident the early generations of the Model 3 will get anywhere close to that.

Interesting. Are hard, clear fruit spirits (usually double-distilled straight off the fruit mash and not treated or mixed after) also called Snaps in Sweden?

Anyway, I can warmly recommend drinking those after dinner, neat, as well. I like plum best, but Williams Christ (a type of pear) is a well-deserved classic, too.

Cool, thanks! My ePub is identical to this.

I am vociferous in my endorsement

That's enough for me to give it a try! After 15 minutes of research and several false starts, I've now settled on the Zelsky translation. I found an ePub with 2334 chapters. Does that sound about right, or do you recommend something different?

I've only learned about xianxia through this thread, and I'm intrigued. Have you read "Cradle" by Will Wight? Is that "western xianxia"? How does it compare to Reverend Insanity?

I'd call Cradle "power fantasy slop", but reading stuff like that is my guilty pleasure. Maybe I should give the "eastern OGs" a try...

Yeah, I would. Destroying the Nord Stream pipelines is easier to defend. You can argue it was more about preventing your common adversary from selling than your good ally from buying. You can argue you're just making sure your good ally is following the agreed-upon embargo (with the stern implication that you both knew that this ally was always in danger of smashing the defect button if the economy got rough enough).

Then, of course, you can also always pretend that it was the work of an Ukrainian crack squad, that they (tragically, really) slipped their leash, and that there's really not much you could have done to stop them in the first place. Your intelligence counter-parties and the political elements they advise will see through that, but they'll understand. Support it even, maybe. The public won't see through the lies, and if they do, they'll have forgotten all about it a week later.

Germany imported less from Russia in 2022 than they do from the US now, and it caused a minor energy crisis and cost spikes when they stopped importing Russian gas. They had to build terminals to receive US LNG. Or am I wrong about that?

No, absolutely! But all those new floating gas terminals are agnostic to who's LNG carrier docks and delivers gas. Any country with gas liquidation tech can now sell to Germany - and that's most of the counties with gas wells.

Specifically, the US doesn't operate a single large LNG carrier. Those are built/owned/flagged/operated by third parties, and they can just pick up gas for Germany from somewhere else.

Gas delivery by LNG carrier is a mature global market. Japan, South Korea and India have historically imported a lot of their energy needs this way. Now the EU does, too.

the United States can probably credibly threaten to throttle German and thus European domestic arms production if they so choose

Not by LNG exports, at least not without significant direct embargoes. Qatar, Norway, Algeria, Canada, ect. all ship a lot of gas, and would supply gladly. The US could increase global gas prices by not selling to anyone, but the Germans would spend too keep their at least their MIC running. And at the end of the day the US really likes selling gas...

No, if the US really wanted to put the hurt on the EU, they could stop selling them chips and sensors.

But I don't think those steps are very realistic, measures like that would be unimaginably antagonistic.

pulling out of Europe means that the Europeans will have to arm themselves further, which might actually prove fairly lucrative to the United States.

If that's your goal, you need to pull out very, very carefully.

The only reasons the Germans are begrudgingly buying any F-35s and FA-18s in the first place is that the US isn't certifying new EU aircraft for B-61 delivery, and the non-French EU really wants to be part of NATOs Nuclear Sharing program. If the US pulls its nukes from Germany, Belgium, the Netherlands and Italy, I don't see those guys buying American aircraft ever again.

And what else do you want to sell them? Outside of an actual defense emergency (where they would absolutely buy everything on offer - as Poland is doing right now, because they correctly perceive the situation to be an emergency already), they are more than capable of arming themselves with domestic systems, and would do so for now pertinent strategic reasons - and a whole lot of spite, of course.

There are no dense, cheap cities in the First World.

Ah, come on! First and most famous example is Berlin - it's still relatively cheap today, when compared internationally, but it was fantastically cheap for 25 years.

Germany is actually full of examples like that. Dresden is following the Berlin playbook, and Leipzig is the new cheap/dense/hip city for now. There's also Dortmund, Dusseldorf and Essen, but those are cheap for a reason (they're ugly dumps, but they are dense - and there are jobs there, so they aren't cheap just because the region is totally economically destitute).

There's also Vienna, which is an interesting case, because Vienna has cheap housing mostly because the city owns tens of thousands of apartment units and is actively using its market position to push down prices. Austria also has a couple of other cities that are cheap and dense, and so does Italy, but going in detail isn't that useful if nobody has ever heard of them.

My conclusion is that you can have cheap and dense Tier 1 cities if you expend some effort, and there's great value in boosting your Tier 2 cities.

There's probably also

  • There's a decent chance that any medical event they would be on the hook for happens at a foreign hospital, which, on average, is going to be a whole lot cheaper than a US hospital.

For reference, $500/mo for health insurance is very expensive even in HCOL and high-GDP-p.c. countries like Switzerland, Liechtenstein or Norway.

my intuition is that one could train a model with lots of sheet music-audio file pairs and then feed it the latter to generate the former

Yeah, that's the way. Once you run out of training data, you can probably also do self learning by transcribing music without available sheet music, transforming the generated notation into sound through a synthesizer, compare the results (this needs another model) and then try again. Once you run out of music, you can continue with synthetic data (since current models can already make fresh sound files of high enough quality).

The devil is in the details, of course, e.g. current software transcription aids work much better for solo piano than for any other instruments (there not many different ways to modify the sound of a note on a piano). Guitars, on the other hand, are notoriously hard to transcribe. They kind of make up for it by having tabs available for a million songs, so at least there's a lot of training data. But the relationship between tabs and final sound is much less straight forward than for piano.

Can it output sheet music? Can it output notation? Part of the problem is the synthesizer isn't that great. Ideally it would output notation is something like Frescobaldi.

No, and it fundamentally can't right now. Those models are trained on raw music, not on notation. During creation, the model isn't "composing" like a human would, in the same sense that an image model isn't actually sketching, drawing and painting - the final image is directly condensed from the diffusion process.

But this is clearly the next step in the value chain. Once audio creation models can input and output notation, they will completely change the creative process - in the same way that video models will become valuable once they can input and output an entire 3D scene into/from Blender. But this step is difficult, there is orders of magnitude less training data in all those cases (you need specific sets of music + notation, video + 3D models, ect.

Music is, of course, simpler than 3D in this aspect. You can run AI audio creation through the usual transcription aids or quickly rebuild a beat you like in Abelton by ear/hand.

Cut off the moldy parts, eat the rest.

This is true for a large number of types of food, but bread is not one of them. Mold on bread goes deep, fast, and eating the deep mycelia is quite toxic.