Sure I care!

If they had any intelligence in that AI, then the improved Moon would show as the Big Cheese.

Never too late to let the model relearn all the goodies.

And Gromit's footprints!

Clangers and Soup Dragon for sure.

You can never photograph the Clangers, there is too much glare off the saucepan lids.

Isn't the "moon" actually the light in the tunnel?

It's one of the most obvious targets for this tech... it's reasonably easy to check that the think looks like the moon, and since the same face is always pointing towards us... you know what it's going to look like. The phone knows the time, the location, the orientation of the phone... and the available high res imagery of the moon at all phases is pretty good.

Got to be one of the easiest AI tasks ever... but it is pretty deceitful in many ways, and it risks the next dataset being trained on what an AI thought the moon should look like...

I'm sure it's not just a 'test' for a wider rollout of this kind of thing.

I mean who'd ever need to be able to AI identify common object in photographs? What use could that be?

Then one day it just so happens that every photo you take that includes (say) a soft drink can now has the logo facing the camera, or the competitors drink replaced...

But noone would ever think of that would they?

It is the continuous feeding in of AI enhanced images mixed with real ones to the training set of future AI image systems that could get interesting.

Sort of like when someone doesn't mute on Teams when they are using a loudspeaker and you get the feedback echo that goes on and on but which alters its frequency distribution over time.

"I am sitting in a room" indeed.

The phone knows the time, the location, the orientation of the phone... and the available high res imagery of the moon at all phases is pretty good.

It's only true AI if it goes "time: 10pm, location: city in Britain, OK, show picture of clouds lit by street lamp glow"

It would also [1]have to take latitude into account .

[1] https://www.planetary.org/articles/can-the-moon-be-upside-down

Who cares? I care.

They could advertise this as a "feature" and something you could switch on and off, but doing to surreptitiously like this is really sneaky.

They literally say in the advert "I'm not making this up", but clearly, they are.

Once this becomes standard our lizard overlords will be able to move freely on our side of the moon once more, safe from any intrepid photographer trying to capture their nefarious schemes. Think you've spotted something fishy on the moon? Take a photo to prove it to everyone... nope, just a normal picture of the moon!

I KNEW that wasn't my dick pic. Way too small.

You mean - Samsungs 'phone allows you to CONTROL THE EXPOSURE!

Dialling in 4 or 5[1] stops of underexposure, maybe - gasp - doing so by reducing the exposure time to help with the wobble!

Wow - I can see why everyone is claiming that this requires an AI.

[1] YMMV - go as low as you dare and bracketing is your friend.

" and bracketing is your friend "

As is a tripod.

Not necessarily. Do you use a tripod to photograph a flower, a dog or a building? Certainly not essentially. People forget that the Moon is just another well-illuminated sunlit object that doesn't require any remarkably different treatment to anything else similarly lit, and if hand-held (+IBIS/OIS if available) is enough for a sharp image, there's no reason for extra clobber. The reason why most newbie Moon photos end up hopelessly overexposed is mostly because they use auto-exposure based on a frame that's 99% dark. Spot metering is all it takes, and yes, bracketing provides a useful insurance policy!

With 125 ASA film, the moon's correct exposure is of course 1/125th at f16.

Call me old fashioned, but I expect a camera to produce an image of the light that falls on its image sensor, not some statistical average of what you photographed ought to look like!

"Wait, that's not Mavis in the wedding photo!"

The 'Sunny 16 rule' for photography is that for a subject in direct sunlight the correct exposure is f 16 at 1/(Film speed in ASA or ISO).

So for the Moon and planets in the solar system, an accurate exposure is actually quite short, as Mr Barnes has stated above.

https://expertphotography.com/photography-101-sunny-16-rule/

> Spot metering is all it takes

FWIW even using selective spot-metering I find that it helps to dial in -1 or even -2 underexposure as well, as you (well, I) don't really want even the brightest area to be saturated (and bracket around that exposure).

> Do you use a tripod to photograph a flower, a dog or a building?

Buildings, yes. With a decent mechanical head on the tripod to get it lined up nicely (Manfrotto 410 for ease of carrying around)

If the flower was about 0.5 degrees across and filling the frame, then again, yes, as that would be (quickly working from a picture of the Sun, which is - surprise - the same angular size) 1400mm focal length on my four-thirds camera or 2800mm equivalent for a 35mm ("full") frame. To shoot that hand-held, even with IBIS & OIS, is - tricky.! I'm willing to give it a shot, though, if you'll get me the M.Zuiko 150-400mm plus the 2x teleconverter, which has that sort of reach (at the cost of max aperture of, um, f8 or even f11 with both teleconverters in; rule of thumb, minimum handheld shutter speed is the 35-mm equivalent focal length times 1.5 (worse as you get older), but get back say 3 stops from the combo OIS & IBIS with my EM-1 (they promise 5 stops but, yeah), so say 1/500th second. That is just in the realms of doable but dang, that lens is heavy, I'm using a tripod for that flower.

I'm an amateur photographer since the age of 10 and an amateur astronomer since the age of 10.

I know that tiny lenses and focal lengths wont beat a telescope and a camera attached. This advert aims to suggest it will, which is simply laughable.

Other people wont know its crazy, and will think this is real and look to buy these devices and then find they have buyers remorse when it doesn't do what its supposed to do.

Anyone who asks "Who cares" must hopefully be trying the same trick as on Reddit, or is totally oblivious to the thing that does care, the LAW.

We have trading standards laws explicitly created to actually prevent thins kind of false advertising and the way Samsung get around it is by putting up tiny barely legible text that everyone ignores saying "image is simulated".

It reminds me of that other phone that claims to have more megapixels than my old DSLR but it turns out its only got half as much and the other pixels are interpolated. Again, there is a statement, a much clearer one in that case that tells you as such.

Basically, if its too good to be true, it is. Read the small print, save yourself being a mug.

> I know that tiny lenses and focal lengths wont beat a telescope and a camera attached. This advert aims to suggest it will, which is simply laughable.

Oh, come of it - the ad shows an image of the entire Moon that can fill a 'phone screen - this is not a major achievement in the world of photography.

Using a camera and telescope, an amateur astronomer would be embarrassed if the image only scaled up to a few inches across: our January AstroSoc speaker, by our local amateur astrophotographer, presented us with an image of Mars being occulted by the limb of the Moon, with Mars being quite recognisable as such. He also has a whole-Moon shot taken at the same scale.

I can get a moon picture that fills the screen. But it requires attaching my phone to the side of my telescope (a little 4" jobby) and lining up everything just right.

You're not going to get a full screen native shout out of a phone because it lacks hardware zoom, the best you'll get is digital zooming and that'll look progressively crappier as you zoom in.

Actually... My real camera has 12x optical zoom and some more digital (to make 64x in total at max) with the lens sticking far out the front, and I don't think even that's enough to make the moon fill the shot (but it's a old camera so great shots in the day, sucks at night).

> has 12x optical zoom ... I don't think even that's enough to make the moon fill the shot

Unfortunately, a spec like "12x" really doesn't mean anything on its own, without knowing at least the field of view at one end of the range, so sadly it isn't enough information to judge against anything.

FWIW, I just saw the ad on telly again - when she zooms in, the image is actually only half the width of the phone screen, held vertically - so that is about 1080/2, or 540 pixels. Presumably taken with the 10 MP (Telephoto) rear camera - can never get the proper specs for these cameras, so let's assume it is the same aspect ratio as the screen and therefore approximately 2240 x 4480, so the 540 pixels is one quarter of the field of view of that camera. The camera is stated to have a 10x optical zoom.

Now, that is not enough information to decide if the camera can make the Moon a whole 540 pixels across - need to know the fov or calculate it from the sensor size and lens focal length - but by the increasing number of downvotes I'm getting here, at least one of you lot can manage to dig out the relevant information to show how absurd the whole idea is. Yes, that is a challenge - show you working of STFU.

> You're not going to get a full screen native shout out of a phone because it lacks hardware zoom

Optical ("hardware") zoom has been a thing in 'phone cameras for a while, hasn't it? As just said, this Samsung 'phone is being quoted as having a 10x optical on one lens and a 3x optical on one of the other lenses.

Those that muck about with astrophotography use a few tricks too. Stacking for one.

For planetary work, stacking is an absolute necessity. Deep space you can get away with plain old long exposures - but go too long and stacking is needed purely to get the exposure times required (one night is simply too short).

But for the Moon, you don't really need stacking. You can fire off the multiple frames, at full resolution, and choose the one with the best seeing, but unless you are doing something extreme, stacking is a lot of work for not much payoff.

Now, taking advantage of the size of the Moon and the detail available, doing massive grid shots across the whole face (prepare to spend months on this one) is a great thing to do - and you will then run some of the same software used for stacking to do the de-rotation, scaling and alignment to put the whole super-res image together. This is what we were shown the results of in our January AstroSoc meeting and, wow, socks were liberally blown off. I could only ever hope to have that much patience!

Samsung are using this one example, of the Moon photos, to imply their camera/phone is amazing for all photos, and this is being used to have a lot of people spend a lot of money on Samsung phones.

In fact, the amazing photos of the Moon work for and only for the Moon.

This to my eye has crossed the line between marketing and lies, in part because the actual truth is so different to reality, and in part because the actual consequences of this, helping to deceive a lot of people into making an expensive purchase, are of considerable magnitude.

100% agree.

It is worring to see an apology of a blatantly misleading ad here in El Reg.

So, basically the marketing is working ?

The camera never lies, they say. That may be true, but what comes out of the camera and what people see afterwards are completely different things.

-> Samsung's marketing is deceptive.

Stage 1. Pretty much ALL "professional" food photography is faked. The "photographers" use everything from paint to shaving cream to make food look better than it is. So the camera has not captured the food, it has captured what the photographer wanted it to capture. That ugly looking hamburger the photographer starts with as a source prop, if indeed it is a hamburger at all, suddenly looks a lot more tasty.

Stage 2. The photos go into Photoshop. Adjust this, adjust that, do something with the curves. It isn't just done to create a look or to "correct" errors, it is done to make the product seem better than it is.

I've picked on food photography as it is so easy to find examples on YouTube showing what I have listed above.

Marketing is deceptive, eh? Who would have thought it. "At a steady 56mph this car does x to the gallon". Remember those adverts? They were egregious in their disinformation.

My Xiaomi phone's camera app has a mode where it can recognise people and "tweak" boob size, butt size, and a bunch of other parameters.

Part of my thinks it's cool that cameras are powerful enough to seamlessly manipulate the image in real time, but a larger part of me thinks it's shitty that some people are so insecure in themselves that they'd feel the need to use this.

Examples? Go into your burger slinger of choice. Look at the glorious pictures on the wall. Compare and contrast with the limp insipid piece of crap in the grease-damp box that they handed you.

The camera has been lying since day one.

Ever since images have been captured, people have been looking for ways to influence the result to their need or desire.

And with digital, it has simply exploded.

Including some cloud manipulation is par for the course these days.

Samsung is lying, bears in the woods, call me when something new is actually happening . . .

Why do all my family pics now look like some random sit-com actors?

Are you Basil Fawlty? Or Uncle Mort?

cousin it

No icon can do this justice

Slid down my pants, touched my toes and snapped away. My moon photos don't look any different!

Weird. Last time I took a photo of an arse, the AI replaced it with a hi-res image of Jacob Rees-Mogg.

Was it in black and white?

"Was it in black and white?"

Sepia.

I bought this camera to go dogging and all I get is the same picture of the moon.

There is no single “moon,” so the ML-assembled photo is misrepresenting the fact that the moon displays libration, up to 8° in longitude (from ellipticity in the moon’s orbit) and nearly 7° in latitude (from the inclination of the moon’s axis of rotation to the plane of that orbit). Except, of course, the evil geniuses at Samsung don’t know anything about the moon other than that there are lots of bona fide images of it. Prats.

A quick Google will confirm there are many photography advice columns that say, if you want a foreground scene with a good moon in the background, you'll have to do a composite shot (in other words fake It by mixing a different moon picture into the one you took of the foreground). All this is doing is automating that trick.