I've never read it. Does he have three robot daughters?

100%!

With the caveat that I'd not say lunar regolith is"like asbestos". Asbestos is unusually hazardous because it splits mainly along its long axis, so its fibres tend to get thinner and thinner over time (unlike, say, glass or carbon fibre, which tend to split transverse across the fibre). This turns asbestos into tiny needles that make it deep into the lungs, immune cells try to engulf them, fail and die, and then trigger an immune cascade in response. It's also not really the same as classic silicosis, either - on Earth we deal mainly with crystalline silica, but lunar regolith is mainly amorphous, which is less fibrogenic. It's less likely to cause long-term health issues and is less long-term stable (good), but the flip side to that is it's highly short-term irritating - not just from being fine and jagged, but also because (having not been exposed to moisture or the atmosphere) it's highly reactive.

Saying we'll get oxygen from the 0,5-1% of a poison in martian regolith, rather than bulk ice or CO2, is...

Well, it's take ;)

There is no reason to celebrate the existence of perchlorates on Mars. Also, I have no idea where you got the idea that perchlorates break down in water. Perchlorates are highly stable in water. Unusually stable relative to their high oxidation potential. It takes a lot of activation energy to break the chlorine-oxygen bonds. Which is how perchlorates long-term contaminate aquifers (even on Earth where microbes could potentially help decontaminate them - indeed, in the human body, the vast majority of the perchlorate that you consume leaves the body unchanged). Mars ice should be expected to be utterly loaded with perchlorates.

(Also, for the record, what we've found at least thusfar is not really "ice" as one may commonly think about it, it's probably better to refer to it as "permafrost" - lots of mineral grains in it)

potentially useful mineral dust to use hydroponically,

Mineral "dust" is very much what you don't want with hydroponics. With hydroponics you want a balance between aeration and water. Fine dusts offer basically no aeration. That would be a pot of rot (beyond being a health hazard to workers and consumers). Most hydroponics on Earth is done with things like rock wool, pumice, sand, that sort of stuff. So long as what you're using isn't toxic, you can use pretty much anything in hydroponics, so long as the particle size is reasonable (with a lot of flexibility, up to and including aeroponics - aka no growth medium at all - but never so dense / fine particles that the roots can't breathe).

(Different media offer different properties - coarser = better aeration but need to water more often & more vulnerable to temporary watering outages, for example. Some, like pumice and sand, are mostly reusable with cleaning / sterilization, whereas you usually don't reuse rock wool as its degradation is usually more than growers want to deal with)

That was a choice

Oh come on. What percentage of modern game sales are of games where the authors deliberately attempted to make it look bad? The overwhelming majority push for realism. And even for those that don't push for realism, almost none of those deliberately push for "looking like subpar or outdated CG".

If you want to jump back 5-10 years in gaming technology, you're free to, but the market does not agree.

. The problem here is that offloading design to AI like this makes everything look the same.

Unlike unnaturally smooth shading, poor lighting models, and limited texture resolution?

There is nothing fundamental about what AI can or can't make things look like. Anyone who has spent time with a modern image generation model (say, Nano Banana 2) knows that the only meaningful limit to what it can create is the limits of what you can imagine. To the degree that DLSS 5 might be too limited**, it's something that would be tuned out of future models, in cooperation with game devs (large game dev firms maintain good communication with hardware manufacturers like NVidia). Beyond model changes, it would be easy enough, for example, to let game devs add textual guidance for DLSS.

** I don't think your diagnosis of the problem is likely to be correct. I have zero doubt DLSS 5 can produce uneasy, disgust-inducing brown-foggy scenes - rather, it looks to me like it failed to detect that that was the goal here. Which again, is something that can be fixed in two ways: better models, and textual guidance.

I see you've fallen for the myth that the moon is real.

That's just propaganda pushed by Big Regolith.

All of these (endless) studies are so stupid. Someone buys a "lunar soil simulant" or "martian soil simulant", grows something in it, and writes a paper. But these simulants are *not* the same thing. They're designed to match (very roughly!) in terms of bulk elemental composition and grain size, but not *chemical* composition, nor trace elements, or even grain texture. For example, you're not going to find perchlorates or whatnot in them.

I mean, congrats dude, you've shown you can grow potatoes in Hawaiian volcanic ash. Stop the presses.

And what's even the point? At best you're showing "I took something inorganic and grew plants in it". That's literally the definition of hydroponics. You can grow plants in a pot ground up plastic Elvis dolls -what exactly is the point? The only thing you could prove is what e.g. perchlorates, arsenic, hexavalent chromium, sharp grains, etc do to plants - *but they're not testing that*.

And he's not even testing hydroponics anyway - if you're mixing it with organics, then you're just using volcanic ash as a soil amendment. Your average ancient Roman farmer could have told you that works.

Lastly, the "potato farming" bit of The Martian was mind-bogglingly stupid tripe, even by that book's low standards.

I'm not sure how much of an issue that will be, to be honest. A third of helium is for cryogenic applications where you have to make it that cool anyway; neon itself is a desired product, and specifically desired without having a large helium contamination; and also there's non-cryogenic means to separate helium from neon, as helium is vastly more permeable than neon.

But it is a good point that it's an extra potential cost to consider.

. Starfield fundamentally changed the character's looks.

Because the character hardly had "looks" to begin with? That's like charging that a realistic rendering of Cloud Strife "fundamentally changed the character's looks from FF7".

I think there's a legitimate complaint that they could follow scene art direction better - and probably will focus more on that in future revisions**. But complaining that it makes someone look more like an actual person instead of a CG blob isn't just an "OK Boomer" sort of thing, it literally runs contrary to what game devs have been trying for decades to pull off with CG.

** And it'll happen in both directions. Game devs will test with DLSS 5 on, designing to make sure that their vision is met whether it's on or off, and there will be - like there is with all modern graphics devs - a bidirectional communication channel between NVidia and the game devs.

The main thing that is "uncanny" IMHO is not the quality of the graphics, but rather, the way things move (something they deliberately didn't modify - they use the original geometry). So you get a hyperrealistic character but moving like a rigged armature.

Anyway, what's "uncanny" is "what you're not used to". If you took modern video game footage back and showed them to people 1-2 decades ago, they'd find it "uncanny". Same with CGI in movies. The whipper snappers here probably don't know this, but there was a massive backlash when CGI started becoming common in movies (seemed to peak around the time of the first 1-2 Star Wars prequels). And yet now you have all these people wanting to use CGI - which they're used to - and not AI.

Just like with CGI from the original backlash: AI tech will get better (early CGI was bad!) at the same time people gradually get used to it, and it'll become a non-issue in the future. You'll still have niche debates, like "Why did they make that character AI instead of having a human actor?!", "The AI in this particular movie was badly done!", etc. People still get mad at particular instances of CGI. But it won't be the unbridled universal anger against all uses that you saw in the early days of CGI, where, say, having a CGI explosion means that you're a corner-cutting scumbag director who hates both the art and your audience.

Honestly, I expect people getting used to AI will happen much faster than CGI. One of the biggest problems people complain about with AI already is not noticing that something is AI, and then being mad when they find out. Vs. with early CGI, you couldn't help but notice - it was in your face. And AI video seems to be advancing much faster than CGI did.

"Waah, I got used to janky video, and now I'm annoyed when it looks more like real life!"

Minecraft will never be the same ;)

*All* AI upscaling / frame generation is "AI content generation". The tweened frames do not exist. The data between the pixels does not exist. Its being made up.

There's not much difference between this and just scaling the same image down and then AI upscaling it back up.

I've seen the demos and I don't see how people are seeing the results as any more yassified than the originals. Even in the main example people keep picking, the DLSS 5 ON shots show more freckles and other imperfections than the original.

And as for comments that it's "depressing" that that future generations "won't even know this looks 'bad' or 'wrong' because to them it'll be normal."... man, these people are SO CLOSE to getting it.