Rumor: Samsung Galaxy S11+ sensor to use Nonacell technology

More interesting news: looks like Apple and Sony may use IBIS in 2020 instead of OIS.

Dec 20, 2019permalink

ewelch

If only the camera makers would care about computational photography like phone companies, imagine what our "real" cameras could do?

On the Samsung front, I only have to ask, will it fold or explode?

Dec 20, 2019*permalink

Ron Corso

I'm not a big fan of Leica but from what I read they are spending their R&D money on just that. If so, it could become the first real camera that has all the tech of phone cameras.

Dec 20, 2019permalink

Lawnmover Man

Many plugins in Lightroom, Photoshhop etc. already do computational photography.. To do the computational parts it in sw in body is not really the best option for most of the processing. Having good RAW-material to work with in Post is the best start of creating a nice photo.

Dec 23, 2019permalink

Samsung's excess of NR will kill it anyway.
Also the "great" plastic lenses.

Dec 20, 2019*permalink

Why don't you think Samsung will change the way NR is done? Now they have a sensor that is thrice as large as their old one, it would be stupid if they wouldn't.

Dec 21, 2019permalink

Playright

Aren't many lenses a form of plastic? A lot of lens 'glass' is actually resin with better light transmission properties than glass and lighter! Cheep plastic lenses are for toy cameras!

Jan 21, 2020permalink

Because Samsung uses too much Noise reduction.
Poor texture.

Jan 21, 2020permalink

winkalman

"The Galaxy S11 series is scheduled to launch in February 2020, so hopefully, we'll be able to have a closer look at then sensor and its performance then."

Maybe you can even be bothered to give the S11 camera (or the Mi Note 10 that's been out for a little while now) a full, proper review. The smartphone camera coverage here is pretty superficial; it seems to be comprised almost entirely of press releases and personal anecdotes.

Dec 20, 2019permalink

(unknown member)

You point it. Press the button without a thought and a jolly nice image comes out whatever. There you go an instant review.

Dec 20, 2019permalink

jxh

Check DXOmark, they have pretty deep reviews. Generally production items (except as noted) so nothing until the device hits the stores.

Dec 20, 2019permalink

Can someone explain why using ultra small pixels and combining them for a 12mp image is better than simply using larger pixels for the same size image.

Dec 20, 2019permalink

I suspect it's a marketing ploy.

Dec 20, 2019permalink

chetan crasta

I think that intelligently combining the read out of multiple pixels helps to identify and reduce random noise (like shot noise).

Dec 20, 2019permalink

More HDR tricks to make photos artificial and instagram ready.

Dec 20, 2019*permalink

https://m.dpreview.com/articles/3560214217/resolution-aliasing-and-light-loss-why-we-love-bryce-bayers-baby-anyway

chetan crasta
You "think". So you don't know the answer.

Dec 20, 2019permalink

APSCII

Because you retain some additional detail (above that rendered by a standard 12mp sensor) while achieving a similar level of noise. Additionally, you get the benefit of the much higher resolution when shooting in good light. This also allows for fewer lenses by way of using crop "zoom". High MP is definitely useful as a marketing tool, but there are also some very practical advantages to it. tbcass's assessment is misguided for this reason.

Dec 20, 2019permalink

petolino

It gives you the option of either high resolution in good light or low resolution (and lower noise) in low light. The downside of the hi-res mode is that the subpixels which make up a group are all the same color, so the demosaicing is more complex and less effective.

Dec 20, 2019permalink

jxh

You can do more algorithmically with multiple pixels in the raw than you can do with single pixels. Potentially even overlaying different mappings... i.e. scan twice with some pixels mapped to different bins then converted.

Telephoto is another good use. Grabbing pixels from the center of the image can provide pretty good performance, without the troublesome mechanical problems of trying to create a micro mechanical zoom (a production and reliability nightmare)

Plastic is not necessarily a terrible thing. Phone camera lenses are typically quite aspheric and can provide a correction that conventional lenses do through complexity. The technology was developed for DVR devices, which have extremely critical focus requirements (such as accurately focusing in different layers in the DVD). After years of development, they work pretty good (those of us who see the world through eyeglasses shouldn't be surprised).

Dec 20, 2019permalink

mmartel

This article should help:

You'll be improving your sampling of R, G, and B channels for each final pixel leading to better color accuracy and less aliasing and moire artifacts.

This should be especially beneficial for green stuff like grass and foiliage that can sometimes appear a tad clumpy due to relative scarcity of red and blue data to mix in.

Dec 21, 2019permalink

Jason Juan

This sensor size is larger. 1/1.33 inches. I believe the old one on s9 or s10 are smaller.

Dec 21, 2019permalink

GearGuy Of theYear

Nah, it was a translation problem.

They Asked a German Samsung rep: Is it true you are going to use a new multi-cell technology, and if so, How many will you use per pixel?

He answered: Nein

Dec 20, 2019permalink

Mister Anders0n

Cool. So more new tech for cell phones, while ILC makers keep pumping out the same crap. Oh wait, they took the mirrors out of DSLRs. What an amazing upgrade, LOL.

Dec 20, 2019permalink

Mister Anders0n

I've not seen any worthwhile improvements in ILCs lately, mostly just gimmicky features. IQ has not changed at all over the years. An "old" D750 will produce images just as good as the latest mirrorless craptastic camera.

Dec 20, 2019permalink

sh10453

They should've called it the Ninacell :)

Dec 20, 2019permalink

Sacher Khoudari

Will the lens be able to resolve all those 108 MP? Especially when taking into account the diffraction of such a small lens.

Nowadays, smartphone cameras are already diffraction limited, they cannot reach the advertised 12 - 20 MP.
https://www.zdnet.com/article/smartphone-camera-performance-what-does-the-sensors-megapixel-count-really-tell-you/

So will Samsung introduce a much much bigger lens to compensate? Or will they continue to supersample an already blurry image?

Dec 20, 2019permalink

dextonsweden

Is diffraction really that limiting?

My LG V30 has a 4.03mm lens with f/1.6 fixed aperture. For sure the 16MP in my LG V30 gives plenty more detail (at least in good light) than Samsung S10 (12MP and killed by massive NR).

The sensor in LG V30 is a 1/3" size. So with a 1/1.5" sensor, you would use maybe a ~6mm. If the lens was f2.0 or close diffraction would be even less of a problem.

I remember my Samsung S6 with 16MP sensor like 5 years ago. For sure it had heavy NR as well but way better details than Samsung S8, S9, S10.

Are they trolling me? I want more details - not lower noise.

90% of images are taken in good light, 10% of images are taken in lowlight where S10 would be in the lead.

Dec 23, 2019permalink

108/12MP 1/1.3" for S11+ ..
64MP for Huawei P40 Pro (rumors of 1/1.2")
12MP 1/1.5" for Sony.

Dec 20, 2019permalink

I prefer a sensor of same size, but 20MP.
No binning.

Dec 20, 2019permalink

madeinlisboa

Nothing like luring small minds with ludicrous and pointless features. What about a battery which lasts for a week?

Dec 20, 2019permalink

creepy.sol

You're aware that you can charge your battery, right? Moreover, battery tech is limited because batteries depend on chemical reactions. There's only so much you can do.

Dec 20, 2019permalink

DenImage

Just superglue an IPhone 11 and an RX100vii together and be done with it.
There's your real Smartphone camera.
Den

Dec 20, 2019*permalink

tokumeino

At the pixel level, yes. But not at the full sensor scale.

Dec 20, 2019permalink

Rob890

Another nail in entry level as well point and shoot cameras coffin

Dec 20, 2019permalink

madeinlisboa

Get a life

Dec 20, 2019permalink

Rob890

@madeinlisboa just like yours

Dec 20, 2019permalink

(unknown member)

There are actually very few point and shoot cameras left. Most fixed lens camera have full aperture, shutter speed and ISO adjustment with comprehensive controls and with most 1" you have to keep an eye on depth of field.
Plus limited reach on smartphones so I would be interested to know what exactly you mean by point and shoot. Plus entry level cameras have not a lot less flexibility than more expensive models. In fact may need more skill to get good results which is part of the fun of photography. Not instant automated image output.

These vague gloating point and shoot cameras are dead posts are so easy to type but convey very little really useful information or any hint that there is actually any real knowledge of the subject in hand. Just casual cut and paste internet garbage.

Dec 20, 2019*permalink

xeriwthe

ok boomerhater

Dec 24, 2019permalink

SolidMetal

I'd really like to see a DPR technical article about quad bayer sensors and pixel averaging. I think I understand the advantages in detail capture and double exposure but how would it help low light photography compared to traditional 12MP sensors of the same size? My intuition is that it should be worse.

Dec 20, 2019permalink

Eric Calabros

There is no advantage in low light, thats invention of marketing department. The invention of engineering department was "high digital zoom without low light noise cost".

Dec 20, 2019permalink

This sensor will not give advantage over native 12. Except HDR.

The Sony imx686 (probably 64MP 1/1.7") will, their new tech seems to get rid of gap between pixels and make much bigger light catching surface, also the AF that covers 100% of the frame.
It seems that it can prevent the AF "hunting" in low light.

Dec 20, 2019*permalink

panther fan

You do get the noise of a classic 12 MP sensor.

However you get more resolution if you need it, like for example when you digitally zoom. You also get an DR advantage in 12 MP mode because different pixels can be read out with different shutter times or gain stages (true for the implementation of Samsung and Sony sensors at least)

Dec 20, 2019permalink

Max Iso

So let me get this straight, by using smaller pixels, and averaging multiples of them, you get MUCH better noise performance? So why don't Canon and Sony just cut out a 450mp sensor, downsample it to 45mp and show it's better than anything else around?

Heck, can we do that right now with downsampling? If i take my 24mp M40 and averaging it down to 12mp, is it better than it would be at 12mp native? Something tells me this is a hype job, based around judging at the pixel level.

Of course a 9:1 downsampling, at pixel level, has less noise, bc you are looking at a larger portion of the light path, 9x larger in fact. Funny, we have been reading here on DPR, for years, about how more pixels means more circuitry and thus more noise (per given sensor size). Now we hear more pixel circuitry is the exact opposite.

DPR why have you been lying to us?

Dec 20, 2019permalink

Actually it works. What has thrown people off is cheap manufactured tiny sensor "point and shoot", mass produced cameras, where often is stuffed more and more megapixels (in a given size) just to pump up the spec sheets for the "my camera is better then yours" crowd. Many people can never fathom; sometimes less is more. B A L A N C E.

If you're like me; you know the engineers know what they are doing. The problem is not the engineers. The problem is many of you can't believe the manufactures would fail to make their best camera, due to greed. It's a simple failure of the "free market enterprise system". AKA controlled competition. Yes propaganda included.

1. A LIMITED benefits camera for every price tier. They are not even trying, on the entry level.
2. Producing a truly balance benefits and value camera would make them less money,
a) There's always a better new version.
b) If you could use it happily for Ten years then that's less sales.

Dec 20, 2019permalink

Of course today we have 4X more mega pixies than we normally need. So these tricks work, with enough extra pixels.

You should try to wrap your head around the fact; that most camera sensors don't do their full resolution very well *and* when you back them down(downsample) to where they are clean, at the "new" 100% then it's as if a 12 Mpx mamy work like a (mostly) clean 8Mpx, or a 50Mpx works like a clean 25Mpx. What have you. You could easily have a 24Mpx sensor beat a crappy unbalanced 40Mpx system (usually a smaller sensor). But as we see, sensors can improve.

Dec 20, 2019*permalink

99tollap

Standalone cameras don't have processors capable of doing this effectively. I meant Sony can't even put their pixel shift processing into their camera.

Dec 20, 2019permalink

https://m.dpreview.com/articles/3560214217/resolution-aliasing-and-light-loss-why-we-love-bryce-bayers-baby-anyway

You mean they didn't. Not can''t.

So see they can do; but don't do. Only enough so you will buy them.

Think of it like this. If you could make A LOT more money (MASS production) selling pet rocks, rather than real diamonds then you would sell pet rocks. We just have to learn to give up our bag of pet rocks, for the real diamonds.

Dec 20, 2019*permalink

PhotoRotterdam

The conspiracy theory addicts are out in full force today.

Dec 20, 2019permalink

badi

@MaxISO
The number of pixels doesn't really matter in that equation. What matters is the actual surface covered by pixels (the photosensitive part) vs electronics.
In older designs they were competing, meaning the more pixels you have also more electronics and wires you had on the surface of the sensor, so normally the noise increased (or rather said, the total amount of photons ending up on electronics rather than recorded was larger).
With current design (BSI, stacked, micro-lenses, "special materials" replacing the metal grid, etc), we almost reached a place where using smaller pixels vs more pixels doesn't really matter as the "total recorded photons" are almost identical in both scenarios (large pixels, or small lots of pixels). You would get more noise at per pixel level, sure, but if you would down-sample the images would be almost identical at any given ISO.

Dec 20, 2019permalink

Max Iso

@Badi, So if an ILC had a BSI sensor, it gets the same benefit as the phone sensor in the article?

@CarpentersEye, every camera i own allows output of a lower resolution. So if i set my M50 at 12mp instead of it's native 24mp, then it's "clean" ? So what's the magic difference between that and me just taking a 24mp shot and down sampling it in PS?

Dec 20, 2019*permalink

mmartel

Read this article:

I think you'll better understand the benefits of sampling more R, G, and B data for each final pixel.

If not, read it again more carefully.

Then consider why high res cameras we have seen recently have shown us better tonal range / color response than previous lower res sensors.

In other words, CanikonSonyPanaOly could probably benefit from doing this, too. And some do, with their high res modes.

But it does take a lot more processing power, memory, and of course new sensors.

Phones happen to have incredibly powerful CPUs, memory, and they can shoot photos without having to be troubled by CIPA battery life ratings and burst speed tagets (6+ fps) that would probably go bust if the same approach were taken in dedicated cameras...

Dec 21, 2019permalink

Max Iso

Sampling more points is well understood by most, has been for a long time and i don't know of anybody who disagrees. What i (and i assume most others) are challenging is the implication it's somehow catching up to large sensors through the magic of averaging.

The single most powerful factor in SNR, overwhelmingly so, is how many photons you capture. I hope it improves phones bc i buy them too, but every new thing that is announced is said to be the big breakthrough, then every time we see the same performance gap.

Im just here to insert a little sanity, even if it is against the current.

Dec 21, 2019permalink

No no mmartel. As usual some of us getting our dualities (when its both) crossed-up with our absolutes. People said sensors have certain (physics) limits; but then they didn't. Physics weren't changed. Our understanding and design did. The sensors changed.

People said too many pixel photo sights stuffed into to small an area would be to noisy. Now a plethora of pixies can (sometimes) compute a low light scene, in fractions of it's native res and in many cases still be more than enough (and with smaller systems).

Haven't we learned to never say never? I'm working on that too. Just remember, the truth is still the same. It's just we can't have it (all). Yet.

Dec 21, 2019*permalink

Boss of Sony

First smartphones got decent lenses. Then they got 1080p video. Then they got larger sensors. Then they shot in RAW. Then they got 4K video. Then optical stabilization. Then UWA lenses. Then they got AI. Then artificial bokeh. Computational photography. Then they got a periscope telephoto that reaches 125mm. Then astrophotography mode.

Now they're getting noise performance that matches full frame cameras.

Next they'll get decent ergonomics. That will be the final nail in the coffin.

Dec 20, 2019permalink

"Now they're getting noise performance that matches full frame" = ""Daydream''

Dec 20, 2019permalink

"Next they'll get decent ergonomics."
Another daydream...
A smartphone will remain a slick, thin brick...

Dec 20, 2019*permalink

SignumX

Then Canon starts upscaling its video to join the 4K revolution *rofl*

Dec 20, 2019permalink

bobogdan78

you hammered a lot of nails, keep up the good work!

Dec 20, 2019permalink

madeinlisboa

LOL. You could be a stand up comedic

Dec 20, 2019permalink

RedFox88

Artificial intelligence doesn’t exist, please stop using the term

Dec 20, 2019permalink

Except the fact most phones at iso400 already have very poor texture.

Dec 20, 2019permalink

Mister Anders0n

It doesn't really matter if the ISO performance is as good as FF, or even APS-C. Everyone shoots with their phone, and hardly anyone is buying ILCs anymore. That's what really matters.

Dec 20, 2019permalink

Articles about smartphones always brings out the Trolls. To answer your question since everybody eats at McDonalds and hardly anyone at 5 star restaurants McDonalds is all that matters.

Dec 20, 2019permalink

Phones has powerful CPU to do the noise reduction comparing to real cameras which has no power at all, so the OOC photos have less noise from phones for the *same size* of sensor.

But even for the same sensor size, real cameras can reduce as much noise as Phones can. How so? it uses PP from Adobe, DxO, and others' APPs on your PC which can be more powerful than any phones at least.

Dec 20, 2019*permalink

'noise performance that matches full frame cameras'
If you merge 30+ images then yeah that might be achievable. Impossible in handheld though.

Dec 21, 2019permalink

Mike FL,
"comparing to real cameras which has no power at all"
This is a somewhat exagerated. They also host dedicated image processors.
They are called Digic, Venus engine, BionZ, TruePic, etc...

Dec 21, 2019permalink

@Karroly.

Comparing to phone's heave weight dedicated image processors from Qualcomm, " Digic, Venus engine, BionZ, TruePic, etc" are just kids' game.

Below is just examples from 4 years old Snapdragon cam features from Qualcomm:
https://www.qualcomm.com/news/onq/2015/01/12/snapdragonsnapdragon-features-add-dslr-effects-your-mobile-phone

Now, Snapdragon adds more features such as, but not limited to:

- AI for noise reduction.

- can same time process up to 3 images from 3 cam/lens to form one final image, and here is an example for two cam/lens:

The phone shoots one photo by using Tele and Main cam same time, and Snapdragon uses Tele image as a map for Main cam's image to fill the Tele image b/c Main cam has the best IQ.

In the phones' sensor HDR, other than take multiple shoots, Sony's phone sensor has "Single Shot HDR" years ago...

Dec 21, 2019*permalink

Mike FL,
I know that smartphones have more powerful CPUs, but again, saying camera CPUs have "no power at all" is exagerated. Period.
And my entry level EOS200D features both several multishot HDR modes and one multishot noise reduction setting that can be used handheld with an OIS lens and in Live View mode for the shortest lag between shots. You are surprised ?

Dec 21, 2019*permalink

@Karroly;

I fully understand real camera VS phones.

I use Nikon full frame, but I use two Olympus body for travel light. By sing two body I do not have t change lens...

Some time I bring 1" sensor P&S such as SONY RX100.M3 and Pana ZS100, plus Oly TG2 for backup in case of rain, and use phones UWA as needed.

Powerful CPU & GPU is not the only thing about phone, but great R&D photography power from Qualcomm in chip level and Google & Apple in software level.

Now, back to Noise reduction, only SONY's cameras have so called " Selective Digital Noise Reduction" which is better than other camera makers, but still way behind Phones "AI powered Noise Reduction".

But none of the phone can beat 1" sensor camera, but "Daydreamer".

Dec 21, 2019*permalink

@Karrory, CPUs used by dedicated cameras are absolutely nothing compared to those that modern highend phones have. According to Wikipedia, BIONZ X processor found inside a7 III still uses Cortex-A5 cores - A nearly 10 years old ARM core which was used in the first generation Snapdragon S1 SoC by Qualcomm.

Next year most flagship phones will feature Snapdragon 865 with an ability to process 2GP within a second, shoot 4K HDR video while simultaneously capturing up to 64MP snapshots, apply varying amount of NR to different parts of the image, and so on.. On top of that, smartphone manufacturers are better at processing images with software thanks to huge amount of R&D budget they have compared to camera manufacturers. For instance, can your EOS200R do HDR shots without showing any sign of ghosting on moving objects? Most likely not.

Dec 22, 2019*permalink

Cameras, for long, use a kind of "selective noise reduction". The darker an area of the image is, the stronger the noise reduction.

Dec 23, 2019permalink

Also in areas with no detail such as blue skies, often more noise reduction is applied.

Dec 23, 2019permalink

@Karroly, that's far simpler than selective NR involving semantic segmentation that SD865 is capable of, which basically identifies objects(tree, face, hair, background, scenery etc) and applies different amount of NR. Actually iPhone 11 Pro already has this, and looking at sample photos it seems to be doing a great job with preserving the maximum amount of details where they are needed, while reducing the noise level to an acceptable level(noise level is still pretty high, but that's mostly due to its small sensor).

Dec 24, 2019*permalink

PortraitLover

by george, just use bigger pixels for mucks sake. that slight bulge wont kill anyone and there will be less cracked screens from accidental drops.

Dec 20, 2019*permalink

Jay A

I really wish these phone companies would stop pretending to be camera manufacturers and just decide to finally make a reliable, good sounding phone already. It amazes me that in the country which invented the telephone, the most common phrase uttered over the phones is " I can't understand you, you're breaking up." Cell phone technology keeps getting faster and faster at the same time call quality gets worse and worse.

Dec 20, 2019permalink

Clayton1985

Well I don't think the phone is causing the breaking up most of the time.... but what they really need to do is just change the name to something that doesn't have "phone" in the name since using a smartphone to make phone calls is probably not even in the top 5 in terms of importance for many people. I do still talk on the phone some but even then I hardly ever actually use my smartphone. A smartphone is a flashlight, plays music, navigates, takes pictures, sends and receives texts and email, remotely controls your camera, pays for your purchases and monitors your grill temperature .... and yeah sometimes is used for making phone calls too.

Dec 20, 2019*permalink

@Jay A
The phone was invented by an Italian, he just didn't have enough money to renew the patent.
Also, the first radio transmission was made by an Italian ;-)

Dec 20, 2019*permalink

AlexLex

@LensBeginner - well, not! Ever heard of Nikola Tesla, he's the man!

Dec 20, 2019permalink

Yes I do.
The first transmission of a message by means of radio wave has been performed by Marconi. Tesla did some experiments with a radio controlled boat, but nothing about radio communications.

Radio control is of course a form of transmission, but more related to remote controllers than to wireless phones, which are the subject of this discussion.

Dec 20, 2019*permalink

Jay A

Actually the invention of the telephone can be attributed to:
Charles Grafton Page, an American
Innocenzo Manzetti, an Italian
Charles Bourseul, a Frenchman
Johann Phillipp Reis, a German
Antonio Meucci, an Italian
Elisha Grey, an Amerian
and Alexander Graham Bell, Scottish born naturalised American who was given the first patent.

The pioneers or radio transmission had nothing to do with 19th century telephones.

Dec 23, 2019*permalink

@Jay A
You have to first transmit something on air before thinking about analog modulation and then about digital modulation, and even before that you have to think about it.

Dec 24, 2019permalink

Jay A

Well you first have to be able to use electricity in order to transmit something on air. So I guess Benjamin Franklin invented the telephone eh? :)

Dec 25, 2019permalink

Of course you have to leverage past inventions and use them as building blocks... but that's a much bigger jump than the ones I mentioned.
Mines are in the same ballpark, yours is _maybe_ in the same town.

Dec 26, 2019*permalink

Mariano Pacifico

35mm sensor of yesteryears is equal to nanocell sensor of today. Be aware ... the cameras are getting smaller and tinier ...

Well it was smaller in the days of Mr. Bond but not good enough.

Dec 20, 2019permalink

Akgbkd

It's no-na cell...

Dec 20, 2019permalink

Jefftan

just bull, better use a larger sensor, dirt cheap RX100 I will be better than this so called new tech

Dec 20, 2019permalink

Jefftan

for these miniature small sensor, real advancement will come from software not hardware

Dec 20, 2019permalink

Lightright

Cheap but fast android phone (crappy cameras doesn't matter) + Sony RX's/Canon G1X's = Win. Install the app that allows photo transfers and you're good to go for your social media posts. This has been my current setup but I use milc.

Dec 20, 2019permalink

badi

@Jefftan - the fun part is that this sensor is also huge in physical size. It is a 1/1.33" compared with today's 1/2.55" in Samsung/Apple/Google phones. It might not challenge real cameras yet, but that size increase in sensors will clearly be a sort of big jump compared to current phones.

Dec 20, 2019permalink

Clayton1985

@jefftan.... I think the problem is that your definition of "better" is probably not the same as all of the people that will buy this over the RX100 or any other dedicated camera. In any case, the smartphone will kill the 1" camera market.... that you can bank on.

@Lightright.... that might equal a win for you personally but that isn't what's going to win. Most people don't want a separate device and as smartphone tech advances fewer people will be willing to do what you're doing.

Dec 20, 2019permalink

LoneTree1

"Binning, by any other name..."

Dec 20, 2019permalink

beavertown

Will their tech in mobile sensor surpass Sony someday given that they are a much larger company than Sony?

Dec 20, 2019permalink

tonyC1994

Nanocell or nonacell?

Dec 19, 2019permalink

qzm

Nona means 9.. Tetra was 4.. IUPAC numberical multipliers..

But yes, they could almost get away with ether for marketing purposes no doubt.

Dec 20, 2019permalink

WilliamJ

Or is it none-a-cell ?

Dec 20, 2019permalink

warp101

Just waiting for the sedecacell sensor of 2024 😊

Dec 20, 2019permalink

Foskito

I thought it was Duracell...

Dec 20, 2019permalink

jaggedhorizon

" they could almost get away with ether for marketing purposes"

Would that be ether or hot air? ;-)

Dec 20, 2019permalink

Camera Conspiracies

And it'll still look worse than a Gopro.

Dec 19, 2019permalink

bobogdan78

actually it looks like m&m's, mmmm ...

Dec 20, 2019permalink

stevo23

So...it's not really 108Mp, it's 12Mp?

Dec 19, 2019permalink

Revenant

The sensor has 108 MP, which is used to create 12 MP images.

Dec 19, 2019permalink

Discombobulate

break each pixel into 9 but they are all in the same color. lol

Dec 19, 2019permalink

WilliamJ

It reminds me the proud counting days of the Fovea sensor...

Dec 20, 2019permalink

Probably uses oversampling for 108MP.
Those 12/48 and 16/64MP offer same detail amount no matter the mode you choose.

Dec 20, 2019permalink

Chris59

They are going to use Nonacells so that Italian grandmothers will be able to use Samsung phones too.

Dec 19, 2019permalink

SteBSit

fet del bu?

Dec 20, 2019permalink

zephyrprime

Or they could have just made a sensor with 1/9th the resolution and gotten even better low light performance due to fewer pixel walls. The megapixel wars just hurt performance in phones.

Dec 19, 2019permalink

J A C S

Psst... This will be the next big thing - a 12mp Bayer sensor!

Dec 19, 2019*permalink

xeriwthe

if they can give you the benefits of extra resolution with small loss in low noise performance I'd say not bad, it does make processing RAW images for those who like that route, a PITA

Dec 20, 2019permalink

zonoskar

They could have done that, but then they'd lost the ability to do multiple exposure settings for each pixel in a cell to give 1 exposure HDR, not the multiple exposure HDR we've all gotten used to.

Dec 20, 2019permalink

J A C S

Even my old S6 was doing this already - HDR images had half the resolution, roughly speaking.

Dec 20, 2019permalink

scoobysnapstories

i believe in samsung, sucks they stopped mirrorless slr cameras.

Dec 19, 2019permalink

Chris59

Hear, hear!!!

Dec 19, 2019permalink

Greg Silver

Honestly with their initial tech into the digital camera arena, if they would have pushed through I think they world have been on par our surpassed Sony.

Dec 20, 2019permalink

Greg Silver

Honestly with their initial tech into the digital camera arena, if they would have pushed through, I think they would have been on par or surpassed Sony.

Dec 20, 2019permalink

Mirrorless SLR !
Hopefully, such a monster does not exist.

Dec 20, 2019permalink

Their mobile sensors are inferior to Sony's.

Dec 20, 2019permalink

CarlitosM

What's the difference to binning?

Dec 19, 2019permalink

Samsung binning... So it is better !

Dec 20, 2019permalink

https://eskerahn.dk/wordpress/wp-content/uploads/2018/07/IMX586.png

None. This is basically 9-in-1 version of Sony's Quad Bayer :-)

Dec 20, 2019permalink

Fly18

Niiiiiiice, Samsung.

Dec 19, 2019permalink

dr.noise

That's great. Let's make 100x pixels but combine them.

Dec 19, 2019permalink

Androole

I don't really get understand the nonacell layout...

In a quad-bayer design, you can readout a group of 4x1 small RGBG subpixels that are adjacent to each other, or you shift your sampling array left and up by 1 subpixel and then you've got a group of 4x4 RGBG big subpixels.

With a 108MP nonacell layout, there's no way to get any colour information for many of those pixels when they're sampled individually in the 108MP configuration. Sure you can get the 12MP version fine, but why even bother with such tiny subpixels in the first place then>?

Dec 19, 2019permalink

onlyfreeman

So marketing can write 108 MP on the box I suppose. The marketing power is probably worth more than any technological advantage.

Dec 19, 2019permalink

Androole

I have heard some people defending quad-bayer (and similar) layouts based on the idea that theoretically the alternating lines can be read out at different exposures in order to enhance dynamic range.

But to my understanding there are no cameras that actually implement this concept, because it's much more complicated than doing whole-image multi-sampling for HDR, which doesn't require algorithms optimized for a specific sensor and kind of readout circuitry...

I'd love to see some evidence of it being used in that way aside from a whitepaper accompanying a concept sensor announcement.

Dec 19, 2019*permalink

s1oth1ovechunk

The nanocell has the same layout as quad-bayer, tetracell.
Converting to 12MP is probably just a more advanced, direct, fast demoisaicing that is more complex than a simple box filter over four pixels.

I assume the point of having more pixels without having a matching bayer array is that you can do fast box filters with simple math while also being able to change the sensitivity of neighboring pixels to capture more dynamic range.
They could also be solving problems that tiny 108mp bayer color filters may cause. Maybe making the color filter array too fine leads to other issues, issues not present on larger sensor cameras with high megapixel counts.

There must be reasons, otherwise manufacturers would have just done the simplest thing...

Dec 19, 2019permalink

iserp

I guess it has advantages with respect to aliasing, given that the color information comes from many more points in the sensor.

Dec 19, 2019permalink

zephyrprime

Because customers are stupid and they will pay extra for a phone with big numbers like 100. This is literally the reason why. Having a 100MP sensor just provides worse performance at higher cost than a 12mp sensor in a phone.

Dec 19, 2019permalink

s1oth1ovechunk

Zephyrprime, that's not true.

Dec 20, 2019permalink

cantanima bis

@Androole "But to my understanding there are no cameras that actually implement this concept"

Maybe not in the present tense, but that's basically how Fuji EXR sensors worked: it was equivalent to two interleaved Bayer sensors, turned 45 degrees, with separate wiring, so that one of them could take a shorter exposure than the other. It was sort-of successful for a while.

https://en.wikipedia.org/wiki/Bayer_filter#Fujifilm_%22EXR%22_color_filter_array

...well, I look further down that page and see that there's something else that you might be talking about, though I'm not sure what the actual difference is in principle, even if the practical implementation differs in details.

Dec 20, 2019*permalink

@Androole, that's not true, Quad Bayer HDR is if anything less complicated than a traditional HDR method due to the fact that image alignment algorithm is almost entirely unnecessary with such one-shot HDR. The true reason why no manufacturers haven't implemented this concept is that multi-frame HDR is simply better in lower lighting conditions: one-shot HDR splits 4 pixels into 2 high exposure pixels and 2 low exposure ones, therefore effectively reduces the pixel size by half. It's a feature that shows the true potential when recording video but shooting still photo.

Dec 21, 2019*permalink

BigBen08

I wonder if there will be a S11 Note, or something like it?

Dec 19, 2019permalink

Leitz

Yes there will be a Note 11

Dec 20, 2019permalink

panther fan

Nice so it will be

-A 12MP camera in low light
-A 27MP camera with single frame HDR in normal light
-A 108MP camera on the advertising sticker

Dec 19, 2019permalink

JochenIs

and how much MP can the lens resolve? how much can it resolve with those fingerprints on it?

Dec 20, 2019permalink

Chien Lin

I use a 64MP phone to shoot, and the resolution is about 20MP FF.

Dec 20, 2019permalink

The way pixels are arranged will only allow you to shoot in 12MP mode. This also means that there won't be 27MP because it's not Quad Bayer, and 108MP will be a dumpster fire that you wouldn't be even bothered to zoom into.

Dec 21, 2019permalink

Onteo

panther fan, by how much would you quantify the reduction in noise coming from a 9 frame AI composite? Also, would you think the resolution would be much closer to 12MP as it gathers from way more pixels?

I'm really curious about this, can't wait to see some real world pictures.

Dec 21, 2019permalink

Mariano Pacifico

Samsung got the technology Apple has better implementation

Dec 19, 2019permalink

SHood

You could say the same thing about Sony as well. Sony Smartphone's always have great hardware but are hampered by poor sofware.

Dec 19, 2019permalink

beatboxa

Cool story.
But is it missing a headphone jack, and does it have that disgusting hole-punch?
If so, I'm sticking with my S9.

Also, give me more area, not more debayering pixels, please.

Dec 19, 2019permalink

Toni Salmonelli

I would really like to see this sensor tech in the next generation of Fuji X cameras. Albeit with a much much lower pixel pitch.

Dec 19, 2019permalink

What for ?
The advantage of a sensor with so many pixels in a smartphone (read pocketable device) is digital zooming.
For an ILC, the pro is : you can avoid carrying long and heavy telelenses.
But the cons are : IQ and S/N decrease as the digital magnification increases.
So, why buy an expensive Fuji camera in the first place to end up with crappy zoomed images ?

Dec 20, 2019permalink

Toni Salmonelli

According to Samsung, the new sensor design with that new barrier material between pixels offers advantages compared to conventional BSI sensors with metal barrier also for big sensor.

I actually are not interested in crazy pixel counts. IMHO manufactures should stop increasing pixel count. I personally do not need more than 20 MP.

Dec 20, 2019permalink

Toni Salmonelli

Ah wait. I am not interested in this pixel binning technology but only the other advancements of the new sensor tech. Forgot to write that in the first place.

Dec 20, 2019permalink

Binning 4 or 9 pixels instead of having only one pixel of the same overall size does not bring any advantages in term of S/N. Instead, "breaking" a big pixel into 4 or 9 smaller pixels DOES add gaps, even if they are very small, which means less photons captured.
Also, looking for new technologies to reduce the gap between pixels is something that all sensor manufacturers are working on. Just be patient... ;-)

Dec 20, 2019*permalink

gskolenda

There is nothing wrong with high pixel counts! As long as they have excellent dynamic range, good low light, and low noise. I had the Nikon D850 loved it but it's size and weight of the camera and the lens's. But Oh how I loved being able to Crop, and when I would down size, the image even got better!

That's why I love High Pixel Sensors! Smartphones need some serious ergonomic's! I hate how clumsy they are to use, touch the screen by accident and then you have to fix that before you shoot, not my cup of tea!

Dec 20, 2019*permalink

"There is nothing wrong with high pixel counts!"
Oh yes there are !
More data to transfert, which means slower sensor readout, and a camera more prone to rolling shutter effect,
And bigger memory cards,
And PCs with faster processor, more RAM, bigger HDD to post-process bigger RAW files.

Dec 20, 2019permalink

gskolenda

Those are all known facts, and a decision one makes when buying a High Pixel count sensor. I keep all my equipment at the latest tech as much as possible, and never had a problem with my D850's large file sizes. You can't expect to cook a gourmet meal with poor cooking tools!

Dec 21, 2019permalink