No, I think that's a bad example to choose. I do take your point, dedicated camera development is more conservative, but their market is different. If you splurge £10k on a camera system, you are likely a pro, and you want consistency, and longevity to the design and the operation of the system.

Thinking about it, dedicated cameras have mainly been innovating in video production rather than stills in the last 5 years.

Basically, all of Magic Lantern (a considerable feature set) should have been standard.

Also, it's a literal button. I'm paying $160 for a button with a timer on it.

Well, not Sony. Their mirrorless cameras have gotten much better autofocus and battery life and slightly better ergonomics, usability, speed and weather-sealing, but there has basically been no change in their video capabilities.

Except there is noise. There is always some dark spot that could do with a little more detail. There is always a bright reflection that saturates the sensor losing its color information.

With a decent burst and enough computational photography, you can recover all that information.

You have clearly never looked at unprocessed RAW data from the sensor or you wouldn't be saying something so completely wrong.

Secondly, strange that you don't actually link anything to demonstrate this noise advantage. Much less that it's "vastly superior" and "not even in the same ballpark". Hint -- because it isn't.

[1] - which is a revealing bit of cargo cult analysis. Sensor quality across "dedicated" cameras have enormous variations, and a lot are simply terrible. One of the reasons smartphones do burst and computational photography is for extended dynamic range, yet the iPhone 11 already has a sensor with more dynamic range than the significant majority of SLRs/mirrorless cameras. There are obviously better sensors, like the A7R II+, but there are many that are worse, sometimes much worse.

I told you the name of the website where you should look. How explicit do I have to be for you? Here's a link if it helps.

https://www.photonstophotos.net/Charts/PDR.htm

Compare an iPhone XS Max to a an A7R4 or something. At ISO 100, The A7R4 has 11.6 bits of dynamic range and the iPhone has 6.4 bits. HUGE difference.

> Without processing what comes out of the sensor is an ugly beast.

RAWs from a large sensor without denoising look fine. On my camera I only have lens correction and color correction - zero denoising. I've done statistical analysis of dark frames using LibRaw to make sure this is actually the case, and indeed I exactly replicated the results from PhotonsToPhotos.

> yet the iPhone 11 already has a sensor with more dynamic range than the significant majority of SLRs/mirrorless cameras

Hahahaha, please, show me one even APS-C camera from Sony, Fuji, Nikon, Canon, or any other reputable manufacturer made in the last 4 years that has a worse PDR than the iPhone 11.

Then you link to a dynamic range chart. Yet we were talking about noise. See, I looked at their noise charts where the XS Max (a generation behind, and significantly bested by the 11) beat the majority of SLRs, including contemporary SLRs. Tellingly you decided not to link that.

"Compare an iPhone XS Max to a an A7R4 or something"

Oh you know...an A7R4 or something. The fact that the A7R4 has one of the best sensors on an SLR is just, you know, incidental.

Fun fact - the iPhone 11 has 10 stops (10 "bits") of dynamic range, which exceeds the majority of SLRs on that site (and of those that beat it, few are "vastly superior"). The sensor actually has 12 stops of dynamic range but automatic pre-processing knocks it down 2 stops (though there are tools that circumvent that). "Hahahahaha".

"RAWs from a large sensor without denoising look fine."

You have never seen the actual RAW data, plainly evident from this claim. Do you know what a Bayer filter is? Do you realize the enormous amount of processing it does just to recreate edges and other artifacts? I suspect you don't. Do you know that every sensor has permanently lit elements, permanently dark elements, and permanently errant elements? That the camera has an evolving correction matrix around these artefacts? The amount of processing that turns that bayer sensor array into a so-called "RAW" file is absolutely enormous, and it automatically includes a tremendous amount of smoothing and corrections, even if you, the ignorant user, remain blissfully unaware. When manufacturers push out new firmware, changes to these processing steps is often one of the most noted changes.

So to repeat-

-the iPhone 11 has better dynamic range, single shot, than many SLRs.

-the iPhone 11 has a better noise profile than many SLRs.

And I'm just talking about the iPhone 11 as a very common smartphone. There are smartphones that exceed it now, the point being that many smartphones have superb cameras.

These are facts. There are SLRs that are much better, like the recent A7s, but any cargo culting as if the entire category gives you a superior result is simple ignorance.

This comment suggests you don’t even remotely understand the poisson statistics underlying sensor noise, or else it would be obvious why they “knock off” those two bits. The SNR of the iPhone sensor is fundamentally limited by the size of the capacitors at each photosite.

> You have never seen the actual RAW data

Yes I have, and my mention of “libraw” should have been a clue for you.

> Do you know what a Bayer filter is?

Yes. I also know about X-trans filters, foveon, etc. I’m fairly confident I can out-namedrop you here ;)

> the iPhone 11 has better dynamic range, single shot, than many SLRs. > the iPhone 11 has a better noise profile than many SLRs.

Give me a specific SLR in the last 4 years where either of these is true. The data is right there on photonstophotos, so go ahead.

You’re going to have a hard time, and if you understood the physics of how a CMOS sensor works you’d know why.

In a post where I'm talking about necessary processing (where you claim there is none) to turn the ugly sensor data into workable forms you attempt to correct me to tell me that it needs to be processed (I mean, you are aware of libraw so clearly you're an expert. funfact: I worked for two years on a professional codebase that specifically dealt with raw sensor processing), as if to demonstrate my ignorance. Remarkable absence of self-awareness on your part.

Your reply could really be posted as a counterpoint to your own first post where you decided to counter my basic statement that all CMOS sensors have noise and abberations.

"and if you understood the physics of how a CMOS sensor works"

Just to be clear, your authoritative site shows the iPhone (the last generation) being less noisy than many SLRs. And it's fact that the iPhone 11 has 10 stops of dynamic range, before any special bracketing, which bests many/most SLRs on that site. And yet we're still going. Amazing

But your core "if you only understood CMOS...reeeeeeeee!" claim is one that is often erroneously made. "larger sensor = larger sensor pit pitch = better, less noisy sensor!" Right?

I mean, this argument has been made against smartphones for years. To assure us that they'll always be crap because...like...physics or something. Though anyone who didn't decide to die on a really stupid hill has long moved on.

But here's the reality-

a) There are enormous difference between different CMOS technologies and implementations. Sony, for instance, has shot ahead of everyone else (though they bin and segment). There are many SLRs and mirrorless cameras with absolute trash sensors. Cargo culting is ignorant.

b) Many cameras with larger sensors spend the benefit on higher resolutions. The Sony A7R IV subdivides the sensor into 5x more pixels than the iPhone 11 (the A7R IV being one of the absolute best). If that was lossless you could simply process it to a lower resolution and have the best of both worlds, but it isn't lossless -- the circuitry and structure to support that resolution is substantial. Each pit has its own lines and power and memory channel. And even if they didn't, noise scales with the sensor size.

"Give me a specific SLR in the last 4 years where either of these is true."

You've already segued effortlessly from noise to dynamic range, humorously, and now you're making a baseless, nonsensical statement just hoping that no one actually confirms it. Your intellectual sincerity in this discussion is somewhere approaching non-existent, so I'm out. Cheers.

Charitably, you are confused, and uncharitably you are lying or misleading.

I gave you a simple prompt where you could prove yourself right with a single camera model, and unsurprisingly you’ve failed to take advantage of it.

> And even if they didn't, noise scales with the sensor size.

These noise sources are completely dwarfed by poisson/shot noise (which is the completely overwhelming noise source on most modern SLRs for “everyday” photographs).

The iPhone’s FETs hold a lot fewer electrons. There is no way around this. And then with technology like Dr-pix present in modern Sony sensors (including the ones they sell to Fuji), the gap gets even wider.

Is that a command? Am I your monkey? I humored your site, noted that it didn't confirm your claims (again -- you didn't link this proof about noise, and when I used your site to demonstrate that you were incorrect...you segued to DR where you could enjoy that this site has a single obsolete smartphone).

Curious that you decided to get the last word in again. Your own proof is flawed.

But again, you claimed that every SLR in the past 4 years is "vastly" better. "Not even in the same ballpark". Your net evidence so far is that a $5000 camera body, which is known to have a class leading sensor, has better dynamic range than a two year old iPhone. "Hahahahaha".

And you're showing your ignorance by not realizing that noise is the main determinate of dynamic range.

Is there a correlation? Yup. Is it a strong correlation? It is not. And when the cited reference explicitly has noise measurements, when someone so carefully steers clear of it...yeah.

If you're going to use a single camera as your example for the industry, I don't think that's fair. Canon makes great cameras, but my Nikon V1, circa 2011, is a camera "serious photographers" like to laugh at, and and outperforms your example. It was 5fps with a mechanical shutter, up to 60fps with the electronic shutter.

At the time, some people used the 30fps burst to make 4K video clips: https://vimeo.com/61441075

And FWIW, I used a command line tool to do image stacking to reduce sensor noise on my Nikon 1 cameras years ago. That was until I started using DxO, which does an adequate job of sensor noise reduction on a single RAW shot without as much work.

Also, the DSLRs I used more than 10 years ago (Sony Alphas) were able to do 5fps, but back then, nobody really took you seriously as a photographer if you used a Sony.

> while large cameras have powerful special-purpose image processing engines, they have relatively weak programmable processors. At the same time, they have more pixels to process than cell phones, because their sensors are larger. These two factors made innovation difficult.

from https://graphics.stanford.edu/~levoy/

Also I think HDR+ and other similar stuff uses only around 12 pictures.

Agreed! A dSLR with the same hardware/software sophistication of modern iPhones would be a marvelous product. If camera manufacturers are unwilling or uncapable to deliver this, I wonder if Apple would ever consider entering that market. Imagine: an iPhone with an APS-C sensor and replaceable lenses! The sad truth is that it would probably be a product with too long a lifespan for Apple to build it.

That's not entirely true, althoughI'm not sure I like the direction. A good example if Fuji's X-T* line, where, as one reviewer noted, "I had thought I'm buing a camera, but it turned out I bought a computer with lens attached". It does all creative tricks to compensate for inaccuracies coming from the hardware, especially the lens - but at the price of draining the batteries exctremely fast, and each lens need a firmware update. Is the final result good? Yes. Do I like it more than more standard solution from Canon and Nikon? Not necessarily. It seems cheaper though.

However, even on innovative features that it does have - like WiFi or face recognition, it implements so poorly that I kind of wish they hadn't bothered.

Fujifilm simply don't have the quality of software engineers that Apple and Google have access to. Nor do they place the same importance on software. The actual physical feel of this camera is sublime, it's a shame they don't think the software is worthy of the same attention to detail.

However if I compare pictures from e.g. Nikon Coolpix P950 which has small sensor but big zoom, the picture quality is not nearly as good as with phone cameras with smaller CMOS.

I wish camera manufacturers started to mount these cameras on a phone hardware, maybe even run Android directly, if they can't get their act together.

> Android Inc. was founded in Palo Alto, California, in October 2003 by Andy Rubin, Rich Miner, Nick Sears, and Chris White.[16][17] Rubin described the Android project as "tremendous potential in developing smarter mobile devices that are more aware of its owner's location and preferences".[17] The early intentions of the company were to develop an advanced operating system for __digital cameras__, and this was the basis of its pitch to investors in April 2004.[18] The company then decided that the market for cameras was not large enough for its goals, and by five months later it had diverted its efforts and was pitching Android as a handset operating system that would rival Symbian and Microsoft Windows Mobile.

Emphasis mine. It sounds like it was intended as an OS for cameras, but in the end it was not designed as one. However source from [18], confirms OP's memory - https://www.pcworld.com/article/2034723/android-founder-we-a...

>The creators of Android originally dreamed it would be used to create a world of "smart cameras" that connected to PCs, a founder said, but it was reworked for mobile handsets as the smartphone market began to explode.

> "The exact same platform, the exact same operating system we built for cameras, that became Android for cellphones," said Android co-founder Andy Rubin, who spoke at an economic summit in Tokyo.

I'm no hardware person, but I believe there are serious constraints on how quickly you can take images with larger sensors. The D6 can do 14 frames a second, and the D850 can do about half that.

Considering that Android was originally aimed at cameras [1], maybe that's something Camera OEMs can finally look at again.

[1] https://www.dpreview.com/articles/0013837043/android-was-ori...