I own several "real" cameras: FF, M43, and over a dozen lenses. Multiple camera companies. On paper many have very impressive specs. The problem is that camera companies focus primarily on raw specs with software being a secondary concern. To give one concrete example, I can press one button on a cellphone (almost any) and it will capture more dynamic range than my $2K DSLR.

How can a tiny sense in a cellphone capture more dynamic range can a massive sensor? Software. It is stacking multiple rapid exposures using an eShutter and computationally merging them. You can do that with a DSLR, but it is time-consuming, outside of camera, and produces worse results (e.g. motion in the frame will ruin your stack).

It is super aggravating, but it isn't actually surprising if you look at the amounts of money spent on photographic software alone compared to almost the entire camera industry. I've given up for most things, a cellphone can beat all but the absolute top end dedicated cameras and even then only for niches.

What have made interchangeable lens cameras increasingly irrelevant: the workflow. On a phone, somewhat obviously, I take a photo, and post it to Instagram, instantly. On a stand-alone camera? I…

…uh, well if I’m shooting RAW I need to either have RAW/JPEG turned on, or convert to JPEG in camera, and then save that image, and then use the awful built-in wifi to connect to a shoddily made app that will randomly not connect for reasons, and then hope the transfer finishes, or use a SD-to-Lightning adapter and dongle up my phone.

The DSLR/mirrorless workflow is still, conceptually, film, except replace a darkroom with Lightroom. It’s terrible for run-and-go casual shooting, which was a huge part of the lower-end market.

Personally, I like the control a more manual camera offers, but appreciate the quality of my phone's camera at times too. Generally, I agree, technology that tries to outsmart us feels out of place. Technology that tries to empower us feels just right.

Given that the focal length of the lenses is different, it would be hard to "fuse" the images together, though I suppose it wouldn't be impossible. All this to accomplish what? A higher effective aperture? Or are you imagining some strange blending of foreground and background in focus, with middle-ground out of focus?

however, there are lenses available that combine multiple focuses, focal lengths, and framings by projecting multiple images onto the sensor, to be captured with a single actuation of the shutter.

modern cameras also include features for automatically integrating multiple captures to expand details such as depth of field, dynamic range, resolution, and so on. your phone camera does this too, it is just concealed and non-obvious.

Yep. And there's a whole culture around it, too. The manufacturers who try to deliver improvements - Olympus, for example, have quite a bit of in-body smarts - generally get shat on for doing so. There's a lot of emotional investment in the "digital darkroom", make everything hard mindset.

Because of what you said right here. They kept trying to appease their highest-end clientele without actually thinking about the job to be done for 98% of photography, which is to post kid and cat photos instantly.

But when I compared my DSLR with my more modern cellphone - well, the colors were just nicer on the cellphone camera.

Just because the DSLR would have better software wouldn't force you to use it. If you want control do that, if you want incredibly high quality point-and-shoot photos, you can do that too.

For my particular weird niche of photography: there is no way a phone is going to get anywhere near the richness of tone and color I get in very low light.

https://www.flickr.com/photos/perardi/albums/721577201444334...

Just out of curiosity, are these flat out of the camera or after processing?

And how is the performance when editing on the phone? Those are some big files.

I’ve only tried editing Nikon Z7 RAW files on an iPad Pro, and I honestly don’t remember which generation of Pro. It was OK, but not stellar.

The file size, oof, yeah. That’s a good point. My lossless compression RAW files from my Fuji are ~30 megabytes. Given an iOS device averages 128GB of storage (my rough ballpark guess), that’s going to fill up quickly.

Fuji. It’s like trying to connect a Palm Pilot to wifi or something.

(I'm actually not sure why this is, since pro video workflows handle it. I think it's because phones and movie theaters have HDR displays but PCs don't.)

Presumably they can do something like deep fusion for noise reduction/superresolution even better than you already have from a real camera, but nobody seems very interested in doing it. Adobe can't even improve their image resizing for some reason (backwards compatibility?), when they come up with a new resizing method they have to put it in a different place in the UI and give it a silly name like "neural enhance RAW".

I do have to go through the step of post-processing…but I'm doing that with JPEGs from the phone regardless.

Completely tangential rant: #nofilter is dumb. Your camera is actually producing a low-contrast black-and-white image that is then demosaiced and interpreted according to an engineer somewhere who is trying to make it look fairly literal. There's filters all the way down.

Wait, where was I? Right. Post-processing. So once I've editing the RAW file in the Photos app, or in Lightroom…boom, it's right there for me. It's on my phone, it's getting backed up to iCloud, and I can trivially post the photo anywhere.

Whereas, on a stand-alone camera, I have to jump through a series of connecting hoops. So RAW doesn't have much impact on the workflow.

I use a Fujifilm X-Pro3, and while I have a variety of lenses, I almost always have my 23mm f/1.4 mounted. With the crop factor that gives me a "traditional" 35mm focal length. I almost always shoot with fully manual exposure, and only use autofocus about half the time.

I chose this kit specifically because I have full control over the image with physical controls to make changes - I can change shutter speed, aperture, ISO, and even dynamic range without taking my eye out of the viewfinder.

The X-Pro3 is pretty much made for me and my use case. While it has a "screen", it's folded into the back of the camera and isn't visible unless I manually flip it down. It can't practically be exposed when I'm using the camera normally. It's handy if I want to shoot at waist level or if I want to show someone standing next to me an image... but that's it.

I have nothing against "computational photography", and it definitely has its place - but for me, as for most photographers, that place is in post-processing. Adobe is responsible for that stuff, not my camera. I expect my camera to produce a RAW file as an artifact, not an "image".

I honestly do not know. They definitely do for still photos—Nikon has had chromatic aberration correction and lens distortion correction for quite a while, and it’s keyed to individual lenses.

If they aren’t doing that in video, I suspect it’s a pure horsepower limitation. Reading the data off the sensor, and applying the corrections in real-time while writing to card may be too computationally intensive or bandwidth-limited.

https://www.sony-mea.com/en/electronics/support/articles/000...

Pretty much this, and this is what some people already do with pancake lenses, correcting the result in post. However there's a problem with this: for the universal vendor-supplied profiles to work well, the lens must still be made within pretty tight tolerances, so the cost reduction is limited. And proper manual calibration is really expensive and tedious, which defeats the point.

I don't know if there's a niche, but I'd love to see some startup doing the full-fledged lens and body calibration (STF/OTF and stuff) on demand to make high quality custom tailored profiles. That way you can use really cheap and compact lenses to get the same result; software correction can do wonders nowadays.

examples:

  - exposure stacking
  - focus stacking
  - time lapse 
  - burst
  - panorama (or any kind of increased resolution)

You’re forgetting that today everyone is a photographer. It’s not like the 90s or even early 2000s where a photo was an event and even owning a decent camera was the exception, let alone actually lugging the thing around consistently.

I grew up on analog photography, and one of things I really appreciate about recent Fujifilm cameras is that you don't necessarily need to use Lightroom, Photoshop, etc. There's an "X Raw Studio" app, which lets you develop RAWs in-camera through a USB tether to your computer. It has a very, very limited set of RAW development choices, but for me it brings back a little bit of the feel of working in a darkroom. Limited options. I rarely use actual postprocessing software these days.

In what aspect ? Artistically a 2010 dslr would shit on any single phone on the market

If your goal is to take pictures to document your life then yes use your phone, 100% the better choice here, but if your job or passion is to make pictures a proper camera will always be better, even my 1965 leica produce better pictures than the current top end iphone.

The problem with DSLRs/mirrorless is that they're proper tools and just like with a brush and a canvas you have to put time and effort to master them. No amount of computational photography will make you a good photographer and if a modern 2k$ camera is the bottleneck it certainly is more of your fault than the camera.

I also think the cross section between casual users who want an iphone like experience X people who drop 2k$ on a camera is extremely small. Different tools for different jobs

This is a good analogy, I will have to borrow it.

I got friends who ended up buying expensive DSLRs after they saw my photos taken with f/1.8 lens. But they had kit lens and were complaining that their cameras. I told them to get f/1.8 lens. Then they were complaining about zoom. Also they could not use maximum aperture outside. I helped them get ND filters. Now they are complaining about that. And forget to take ND filters indoors, get way too many photos incorrectly focused etc.

Most people expect pro cameras to just act just like cell phone cameras.

And the internet. Just search for whatever questions you have and you’ll get tons of results. Now that it’s all digital, you can try a bit of everything and see what you like (and ignore what you don’t).

But I learned a lot more by joining a local photography meetup. They would hire models to practice portrait shoots with and the main people were very helpful. They charged $50-$60 per shoot though.

Fair warning, back then I ran into a lot of photographers who did photography for living and were pretty grumpy about dwindling incomes. Don't let their snarky comments discourage you from going to meetups.

The camera you have with you is better than the camera you don't. A smartphone camera which delivers 90% of the results while being on your person 100% of the time is therefore a strong contender.

> No amount of computational photography will make you a good photographer and if a modern 2k$ camera is the bottleneck it certainly is more of your fault than the camera.

- Posts about trade-offs and the loss of competitive advantage between DSLRs/mirrorless Vs. smartphones.

- Replies that they aren't a "good photographer" because they hold this opinion. And are therefore clearly the "bottleneck" in quality photographs. QED?

This is exactly why photography never evolves (see also film). The community is actually the problem. More specs! More specs! Ignore technological advancement! The camera market shrinks 10% YoY, but no changes, no listening to the market, those who point out the flaws in the strategy aren't "good photographers" who are just bads "bottleneck[ed]" by their lack of skill.

CPU and GPU power has doubled, and power consumption has halved, computer vision has had a Renaissance, use it onboard? Na, that's only what 95% of the consumer and prosumer market wants, let's just continue to serve the 5% and keep the training rolling towards bankruptcy.

If your equipment need RAW, you've just shrunk your market share by 80%. If you cannot post instantly you've just shrunk it by 90%. So fight to be the King of the tiny kingdom that is the remaining 5-10%. RAW is a great facility to have, but it is just that, a facility. When it picks up the slack for the antiquated onboard abilities it is just a crutch at that point.

It delivers 0% of what I, and many, want. It's like saying a renault twingo is 90% of a porsche 911 because it can go from A to B, it's not and people who say it is simply don't know what they're talking about or don't have the perspective to understand the need for such tools.

> This is exactly why photography never evolves (see also film)... More specs! More specs! Ignore technological advancement!

Photography _never_ evolves ? Er ok, well if that's what you truly believe I don't see much room for argumentation

> If your equipment need RAW, you've just shrunk your market share by 80%. If you cannot post instantly you've just shrunk it by 90%.

> The camera market shrinks 10% YoY

It's fine, DVD don't sell much anymore either, people move to what's best for them, a dedicated camera isn't and if a phone does 90% of what they perceive is enough then let it be. Nobody's going to buy a sony a7rIV to take snapshots of their lunch and post it as a 480p image on instagram.

People use phones because they have phones with them anyways, not because they have computation photography assistances. If it was as easy as you imply don't you think we'd see viable products already ? There is no market for camera oriented phones nor for phone like cameras

Well, it generally is, given 90% or more of a typical 911's driving is probably not on the track or strip. Think of how many rich retirees have them simply for the build quality and status that comes with it.

What everyone's missing is that smartphone cameras are a completely different use case than dedicated cameras; whether it be your hobby or profession, a DSLR/other standalone camera is something you take because you're trying to take stunning pictures and you're willing to tinker with the photo before and after it's shot. A phone camera is designed to look stunning, and while you can shoot in RAW, it's designed to get shots for people that only take pictures to later remember the time they spent in a place - these are the same people that would've instead purchased a cheap pocket camera for their road trip and would only see those photos again when they were scrapbooking.

I can 100% guarantee that it won't when I'm holding it.

Sure, you can invest time to get good at it and the money into the tools to apply those skills. I have other priorities so I don't and the very best option is a smartphone camera.

It makes no sense that point and shoots aren't attacking this. Even "cheap" ones have a massive edge in terms of (imaging) hardware over phones.

Someone like Apple or Google should create a dedicated camera and just knock that entire segment out from under the dinosaurs.

what they won't do is pre apply filters to make the colors super poppy and generically "artsy" like the phone will.

Of course they do. Fuji’s X100 line has a whole slate of film simulations (mimicking old 35mm film). My Olympus bodies have 10-15 art filters built in. They just don’t have fancy names like Instagram - they tend to be more descriptive.

With a monochrome sensor you can get already exceed 50% quantum efficiency, requiring only a few hundred photons per pixel to get decent quality, and around 10000 in the bright spots to reasonably match what a TV can reproduce (dynamic range and perceived noise).

But with a color sensor you can get recognizable/useful colors down to about 100 photons per pixel. Yes, it will be noisy, but at that light level (assuming a remotely-fast lens) your eyes will give you a hard time when you even try to read normal text. Should be enough to walk without tripping, though.

A faster aperture will let in more light, so you will not need as long an exposure to get the same amount of detail. It can make a meaningful difference in low light shots.

Adjust your aperture to get the appropriate amount of bokeh, of course; but generally I prefer as wide an aperture as possible in most situations, as long as the focus is appropriate.

A high-quality P&S (like the Sony RX, Fuji X100, or Ricoh GR) is going to outperform a phone camera in just about any edge-case (low light, motion, natural bokeh). Even more so if you shoot raw.

Somebody said it best in a sibling comment - a high-end camera is a tool, like canvas and brush, and you need to know how to use it, practice with it, and get in the zone when using it.

Nobody is arguing that cell phone cameras aren't amazing. They are, no question. And they're almost always in your pocket/purse, so even better. They're great for generic snaps of life.

But there are things they simply can't do - wildlife photography and outdoor sports (not enough zoom), low light - toddlers playing inside or indoor sports (sensor too small, stacking exposures with night mode doesn't work with fast moving subjects), natural feeling bokeh (portrait mode has some odd edge cases). If you don't care about any of those, then a cell phone camera is likely all you need. If you do care about those, then a high-end super zoom, mirrorless, or DSLR is going to make a massive difference.

Look at the discussions about low light below. Why the heck should I know about f stops when my camera could just measure and adjust accordingly (including giving instructions to hold) like the phone does?

But the bigger the sensor, the slower it scans. Scan & stack is not yet feasible for bigger sensors.

Besides, smartphone sensors need to do this to overcome their limited dynamic range. Bigger sensors are not typically limited by dynamic range, and hence don't need this.

There is a massive gulf between smartphones and standalone cameras there in the outcome.

My point is that there is literally a product segment called "point and click" that is terrible (relatively speaking) at doing that thing.

All the replies here are focusing on how cameras can produce better results. The difference is my phone will in my unskilled hands.

Btw, why do people keep saying "DSLR" to mean "real camera"? "DSLRs" are not at all the ideal of picture quality, actually they're worse than mirrorless but when they were "SLRs" they were worse than rangefinder and view cameras.

I'd be happy with a mirrorless shell that I snap a phone into to control the rest of the workflow. Give me that iPhone auto-AI picture magic (and some tunables to turn it down when needed) and I'd be happy. Well not in bright sun, but whatever.

I was just looking for an EyeFi the other day, which are apparently no longer a thing. The Wifi workflow on my A6000 is just so bad as to be unusable, so I end up batching photos into a big editing session every 1-2 weeks, uggh :(

I don't believe it's true. Dedicated cameras never "ignored" the means to get better dynamic range and signal/noise ratio with stacking. It's just not needed most of the time, being a very specific trade-off for certain scenes. And RAW processing software is pretty much embracing computational photography (calibrated optics correction, noise reduction, highlight reconstruction etc, including things like auto retouching and tonemapping if you don't want to do it manually). It hugely improved over the years, being able to pull out much better results from the same RAW shot.

What they're lacking is miniaturization and speed. Big sensors and many megapixels mean slow reading and writing speeds; my smartphone can record RAW video at 4K@60fps which is out of reach for my camera. Modern super-sharp lenses became huge. Electronic viewfinder and variable aperture mechanism can't be made reasonably small. Thankfully, the need for mechanical shutter seems to slowly go away as recent Nikons show.

I hope there will be more super-compact cameras like Sigma fp in the future; combine it with a pancake lens and proper calibration ("computational photography") to compensate for the lack of corrective elements in pancakes, and you got yourself a pocket camera with vastly better capabilities.

Smartphones can sometimes get nice results and are pretty good at motion compensation in deep stacks, but they are slow, un-ergonomic and very restricted in what you can shoot well with them. And the people who use dedicated cameras tend to shoot RAW with smartphones for best results as well. (stacking is available in bayer domain so it's not a problem)

I respectfully disagree. No dedicated format for stacks (instead just filenames), no first party software to combine the stacks, and absolutely no on-board combination.

Heck on a fairly recent camera I own, you cannot even configure the camera to always take a stack cellphone style. You have to bind to button to open a menu, to select stack size, every single shot.

> And RAW processing software is pretty much embracing computational photography (calibrated optics correction, noise reduction, highlight reconstruction etc, including things like auto retouching and tonemapping if you don't want to do it manually).

True. However, the market has shown time and time again that onboard is more popular than post-progressing. People want to snap and post, they don't want to snap and then post the next day/week after they've post processed it in PhotoLab or Lightroom.

Why do we need stacking? Because it allows you to do exposure bracketing to get better dynamic range.

Now the question is, in a high end mirrorless like, say an a7r4, you get something like 12-14 stops of dynamic range (expressing this as a number is kind of ambiguous). Do you really need one or two stops of dynamic range with bracketing? I don't think so.

If you expose smartly you can absolutely make use of the incredible dynamic range.

The only problem I see is on iPhone, it shoots in HEIC which is a great format because it lets you encode much higher dynamic range (and the software support is there to display it correctly). Check out a recent iPhone and take a picture with the sun and something medium-dark and you'll see that the sun is eye searingly bright whereas the medium-dark subject is still visible. I wish there was a flow that lets me produce HEIC images (AFAIK even Photoshop doesn't allow HDR workflows). A lot of the pro imaging workflow is only "aware" of printed end results and low-end digital images like Instagram or JPEG web publishing. Very frustrating in my opinion.

This is false. My 2016 (I think?) Olympus Pen-F does multiple-exposure combination on-board. The results aren't always the best, since you can't control much, and it's JPEG only, but it does do it.

The reference manual: https://www.manualslib.com/manual/1081745/Olympus-Pen-F.html...

It reads the 4000 pixel rows in parallel, from the bottom and the top. There's no hard reason why this couldn't be extended to a larger sensor, with the per-row time at-worst linearly increasing with the row count (due to capacitance on the column-lines). But even then, the row count would only have quadratic influence on the readout time, and the column count not at all. Combines to, for fixed aspect ratio, a constant number of pixels-per-second.

So going to the 8k equivalent of 8000x6000 pixels (48MP), you'd quarter those 300fps and still have 75fps left. Assuming you take wide-screen video of 8000x4320 (some sacrificial columns that don't matter for speed and double the rows of "4K" (2160p)), the reduced row count would take that up to ~104fps.

Oh, that chip also has a pipelined electronic global shutter, so you can take the long exposure of a HDR stack while the short exposure is being read out. Also no jello and shutter down to 1/50000.

How big of a sensor were you thinking, that RAW video recording would be inconceivable for the big sensor? And yeah, sure, you'll have to do something with those many gigabits. It's about 36 Gbit/s for the above example. A PCIe4x4 connection would do, but you'd need about 2-and-a-half PCIe4 M.2 SSD's worth of sustained sequential write to handle uncompressed RAW.

I very much prefer the rendering of a "big camera", even with less dynamic range, than what can be obtained from an iPhone.

Of course, as others have said, the iPhone quality is still amazing for what it is, especially given that you can carry it in your jeans pockets. Even the tiny Olympus Pens are worlds away from that. And just as always, the best camera is the one you have with you.

I also agree that it would probably be great to have a "big camera" with the smarts of an iPhone. But there's probably not that big of a market for that, so the manufacturers don't invest in it. Although I'm surprised Sony didn't do anything, since they're in the camera phone business, too. At least, much more than say Nikon, Canon or Olympus.

Until you zoom in, and then it's all a big smudge. Unless you are talking about $1k+ cellphones only, which is outside of the price range of most.

There are two photographs that come to mind where the iPhone utterly disappointed me. One was a morning park scene, rays of light filtering through leaves. The leaves just looks liked blobs. Another was on a boat at night; with my eyes I could clearly see texture in the buildings on the illuminated skyline, reflections on the water, and texture in the water. The iPhone shot was just a black screen with some blurred specks of light.

I know I could have gotten both of those photographs with my A7ii. I have handheld nighttime pictures with stars in the sky. A big sensor gathers a lot of light, and even more important than getting close is collecting as much light as you can.

They do make clip-on zoom lenses for phones but I haven’t actually used any, not sure how much they help.

With all that my 13 Pro is about as capable as my old Nikon D3100 which I bought over 12 years ago.

I am actually considering buying a Nikon Z50 so I can actually go back to an era of more control.

They have "periscope lenses" in some phones

The iPhone 13 pro has a couple of annoying things already in that space.

People buy DSLRs for many reasons, none of which include unremovable sharpening filters, automatic replacement of faces with leafs or turning a bright orange sky blue. Or replacing the blurry 300px moon you just shot with a 3mp stock photo.

It’s probably the same as AOT vs JIT — the proper solution would be to move more data to the computer where we have faster hardware and less resource-constraints, which would be the equivalent of profile-guided optimizations in my analogy (things like movement, maybe even some basic description of the scene based on NN and previously seen images)

https://chdk.fandom.com/wiki/CHDK

You can still do it manually, but give it 3 years and you'll be able to compose that photo from scratch with an off-the-shelf generative NN photoshop plugin anyway. But still I'd say the market for DSLRs will remain the same because their main function (in my opinion) is to capture an accurate representation of reality with predictable errors in pixel space and not in some latent one.

this was confirmed to be fake news.

Also, as a portrait photographer, I disagree too. Until we can use flashes properly, I'm not getting rid of my camera.

Plus, dynamic range isn't everything.

Minuses are staggering - the amount of money, time invested, annoyance to shooter and subjects, the speed with which you can capture most moments in life.

If I understood you correctly, you're pro/semi-pro. That's not the topic here, its everybody else snapping pics. As long term full frame shooter, for last 2 weeks, its almost dead platform for me. I'll consider to take it for longer vacations, but I expect it to come out of bag very rarely.

You don't disagree with the post you replied to. Niches will always require specialized equipment as it said. It was a comment about general purpose photography (and its ever shrinking market share).

Did you ever build a desktop gui app with php? You actually can, it's just not the most productive tool.

Conversely, you can take good pictures with a smartphone (for some specific cases), it's just not a great tool for the job.

Sure, you have to get better hardware to properly capture the fast moving target of the photo, but it is not an either or question.

But if the other person is telling me "deposit money in this ATM, it's totally secure' then I'd be a bit assertive with holding him to account.

However I want the subject to be tack-sharp, and he's moving and changing shape all the time.

As it is, a DSLR or mirrorless cameras put a lot of money to make good lenses and frames.

I think it is fair to say that the software was usually on desktops, with the likes of Adobe providing all features needed for making a good photo great.

Is it really necessary for the camera companies to cram half baked, low powered software into compact camera cases and increase the cost by another 600 - 1000 USD?

There was a time when everyone bought point and shoot ones to get their pictures. Mobiles have almost fully replaced those ones. Some people bought DSLRs and the like as a step up compared to PAS ones.

Good mobile cameras might make the DSLR ones back to being professionals choice, rather than an enthusiasts choice.

I would rather they build software to interface with phones to make use of their tremendous computational power rather than re-invent the wheel.

How about having a DSLR that can sync its photos with an iPhone and an iPhone app that automates the processing of the photos?

Its not as convenient, but the quality is better.

But, HDR isn't necessary in most cases, specially given that RAW files give you an extra stop in both directions when post processing. OTH, the properties of different lenses (FOV, DOF, magnification, macro) can't be reproduced in a cell phone camera (only approximated in software, and not usually well).

A big DSLR isn't necessary for snapshots, but for serious applications (landscape, portraiture, nature, sports, events, etc...) they don't hold up.

This is not completely true, since a DSLR doesn't have an IMU and LIDAR stream, which are used extensively. Smartphones aren't just sticking independently captured images together.

With modern gear you can even recover 4 or 5 stops of shadow details relatively easily: https://theartofphotography.tv/nikon-z6-dynamic-range/

I asked a very similar question in another comment thread recently and the answer I got was that camera's awareness of the scene isn't nearly good enough when it comes to adjusting its own sensitivity curves, and that our eyes and brain are much better at postprocessing the raw data then the camera is.

Of course, phones aren't necessarily good at this either.

The dynamic range thing is frustrating, indeed. But even so the data is usually still there. Perhaps better scene programs is a solution for traditional DSLR cameras, but most definitely NOT "computational photography"

Photography needs better sensors and better scene programs, not a bunch of algorithms with their poorly-thought-out predefined notions of what is "good"

It is not AI upscaling or whatever “magic” that just guesses at noise, most of it is just clever “hacks” to get out more data than you would ordinarily get, e.g. by using up the small vibration of the photographer’s hand to distribute the adjacent color filter’s input to where it belongs, and the like. Many part of it is done in hardware as well.

There is a clear difference between what the iPhone 13 does computationally and what my DSLR body does computationally. The body doesn't claim to use AI, it doesn't refer to machine learning models, and aside from the physical characteristics of the hardware that affects the images that it takes, what I get out of it is pretty much raw sensor data. Which then gets fed into an image processing postproduction workflow.

I would think that any photographer worth their salt would know that a lot of famous photographs of the stars are false color, as NASA very prominently annotates exactly how it achieved the image that it did on its site. But I would bet they are still hand-tuning the end result. Do these new iPhones come at you with, "hey, it looks like you underexposed the face so I fixed this for you" then ask me to keep or undo the changes?" From the description of it, they don't. It just happens, without your consent, because Apple thinks their phone is smarter than you are.

No it's also hardware. The cellphone has much more powerful CPUs and GPUs than your DSLR that enable this

For day to day photos, where there is no set up, sure, the camera phones are great because you can quickly take that photo. As they say, the best camera is the one you have with you.

As a complete amateur, trying to take a photo of my kid at soccer, I have a Sony A6000 and I bought a used Sigma 100-400mm F5-6.3 DG DN OS. The camera phone is simply useless. Also, good luck trying to take a photo of that bird and showing it to your kids!

That said, I still have to carry my mirrorless into the backcountry because tiny sensors can physically only let in so much light, which makes them unsuitable for astrophotography. I’ve seen the “Astro-modes” on new flagship phones and the result is always “wow, this is good...for a phone”, but nowhere close to what I can get unedited from my camera.

And the worst thing is that it’s unlikely to change. Companies that have the ability to make really good phone cameras have no interest in ILCs, and the ones who make really good ILCs don’t care about software.

And good luck trying to compete in the middle without the hundreds of millions that Apple/Google put into ML research and the decades of research that Canon has done on colour science.

All DSLRs come with an “auto” mode. Why should it be the case that this mode is so rubbish? There is currently a gap in the market for a camera that combines superior hardware with superior computational photography. You can keep your RAW images but why not have a computationally processed output option to go alongside that?

Preach.

Next up: $500,000 cinematic cameras and lenses, because we'll be capturing light volumes and skeletal postures.

The biggest advantage won't be in equipment cost savings, either, but rather substantially increased workflow efficiency.

That said, rarely are HDR shots actually desirable IMO.

Most smartphone pics have a weird feel to it, I thought I was being picky but it seems the article author feels the same. Computing power alter the optics too much.

Until about six months ago, I carried a Fujifilm X-E2. It's an older camera (~2014), and was 16.3 MP. I upgraded to an X-Pro3, which is 26.1MP. Given that the MP stat increases with the square of resolution, that's not a huge change.

My iPhone X has a 12MP camera, the same as the iPhone 13 Pro. The technical resolution _doesn't matter_. It hasn't mattered since the 10MP barrier was breached.

I would prefer the Nikon that I carried in ~2004 to the iPhone 13 Pro for "photography". Note that that is in quotes, because "photography" is very different from "taking pictures". I use my iPhone when I'm with my family most of the time and it's great for that. It's great for documenting activities and "capturing memories". That's not at all the same thing as "photography".

My X-Pro3 can produce excellent images at 12,800 ISO and f/1.4. My iPhone doesn't even tell me what ISO its capable of as far as I know, has a static f/2.4 aperture, and the OS "Camera" app doesn't even let me set shutter speed. It does "does its best".

They aren't comparable tools.

Absolute ease is rarely a requirement for quality, be it in photography, music .. or most things I guess.

DSLRs have their places, but to say that the entire professional camera industry wasn’t caught off-guard and failed to respond in kind to the rise of computational photography is an easy statement to make.

This is film vs digital all over again.

Another important factor is “the best camera is the one you have with you.” This killed the compact camera segment. I don’t think it eats much into the interchangeable lens camera market, which is what you claims are becoming irrelevant.

Just to prove that computation photography is not missing, it’s not new at all. Digital photographer has been using those techniques all the time, including HDR you mentioned. Some Sony camera can stitch panorama automatically so there are examples that in camera processing are also there (some Sony cameras can install apps that process HDR too, IIRC, but they killed that online store.) Again it is the user experience, if it is seamless and doesn’t take you much additional effort, then it becomes a more useful tool (in terms of how often you use it.)

Your criticism of “producing worse result” again is not a matter of computation photography. It is just a matter of FPS, ie how fast the sensor reads out the data. Cell phone cameras especially iPhone (a few years ago Apple hired an indie app developer that optimized the app crazily using hand tuned intrinsics to capture highest FPS that have been seen on that model, IIRC 20fps.) But nowadays even FF sensor can be read out at insane FPS making your point irreverent. This is not a coincidence because a) who do you think develop those sensors used on smart phone and b) it has many other uses other than producing low artifact HDRs.

Also if you want no-artifact HDR, not just low-artifact HDR, then get the biggest sensor you can get and take single shot HDR, not a cell phone.

Lastly, computational photography is a necessity for smart phone, just because of physical limits. Indeed it’s a blessing that you can squeeze that much from such a tiny camera. But make no mistakes, it doesn’t win in the quality department, and people don’t hate it because it is “cheating”, but about the quality.

Just to name one feature considered to be cheating and eventually becomes a standard in lens design: aberration correction. I think Olympus started it first in M43 that the electronic communications also send barrel distortion and CA correction information back to camera and write it in the EXIF data, and correct it both in the fly and in any preview or rendered output (ie this is a mandatory correction.) at first it is seen as an inferior lens design philosophy and eventually it is seen as a better optimization (as you only care the end results and whatever means towards a better end is a better optimization) where the biggest benefits is size (optimized/minimized size given same quality.)

So no, it is not a controversial take, it is just a wrong take.

Phone cameras absolutely unequivocally cannot beat the clarity, sharpness, detail, natural bokeh, shutter speed, and overall quality of even an affordable dedicated camera in the $1K range. The only time a phone camera can somewhat keep up is when you're looking at the photos on your tiny 4" phone screen and throwing some heavy duty filters on it to the point where anything will look "good". Throw that photo up on a computer monitor, or TV screen, or get it printed and you can immediately see the difference just in sharpness and detail alone.

Yes, phone cameras are far more convenient, and they have more filter options, and so on. But there's no comparison whatsoever. It's just that most people don't care enough when they're putting it up as a tiny little square on Instagram or sending it as a Snap that someone will look at for 5 seconds and never see again.

For example, I have seen the Pixel 6 take a better photo of a river before a town at night due to the very hard dynamic range requirements plus the movement of the waves in the dark.

Although a lot of them do just shoot on larger cameras.

11 years ago, I gave a talk about this at the Internet Archive. It seems to have held up. https://youtu.be/UMMogOoWEbI

Of course, you can always take off the training wheels and shoot raw.

Apple even worked with Adobe to create ProRaw, which allows you to selectively turn on or off various parts of their image processing pipeline after you shoot the image.

https://lux.camera/understanding-proraw/

The big picture is that cameras profoundly influence how we see and share ourselves, others, and the world. Fairly recently, cameras (and their developers) started making opinionated decisions for us about things like the color of sky, the texture of skin, and much more. Sure, any one of us can opt out and do things RAW, but the rest of the world will go on taking pictures with these decisions baked in. I don't think the answer is RAW, because the problem is not pixel data -- it is who controls image making in the first place.

That's called "auto white balance" and we've been stuck with it since the advent of digital cameras. Phones don't do much more than that; if they detect the sky it's to adjust denoising to not cause banding artifacts etc.

Is there any company out there doing just HDR, and not making color choices? Or is it baked in at the SOC level now?

I think that's the definition of a "smart" appliance, and you just described what my problem with "smart" appliances is: they are too smart by a half, and end up making bad choices when pushed a little.

That's seems like the opposite of smart—instead, it's dumb yet intrusive. Or as Kurt Vonnegut said: "Beware of the man who works hard to learn something, learns it, and finds himself no wiser than before." These devices do a lot of thinking, but have no wisdom.

But you can't encode a set of heuristics that works every time into a physical device - that is the real problem with "smart". In a small minority of cases, the situation will get so complicated that the rules you embedded will produce suboptimal behaviour. To overcome that, you'd need to embed a human-like intelligence in the device.

Similarly will they fail for the minority of people who have drastically different goals (e.g. aesthetics) for the purpose of the device. Now you end up having to clone the owner wholesale.

Okay, really the problem with "smart" devices is that they typically don't disclose what rules they follow, and typically don't come to the user asking for help choosing when tradeoffs enter the picture (literally in the context of this article).

"Dumb" devices, meanwhile, lay all choices flat.

Granted, cameras have long stopped being comletely dumb - but they often make a good job opening themselves to doubt with things like optional manual color balance, and different priority modes.

I don’t, really. Making lots of good decision, sure. But just because they’re complicated? No.

Aren't hardware manufacturers also making choices for you?

I get that software allows for far broader picture manipulation but there are a decent amount of similar choices on the lens and sensor level, no?

Software makes it more of a black box. The same picture taken days apart could look different due to a software update you weren't even aware of. Hardware is also a forced choice so is more deliberate which again is the opposite of the software situation of changing algos and processing.

The iPhone 13 Pro Max removed wrinkles, sun spots, moles, hair, etc to the point where the results looked like overprocessed, manually edited photos. This wasn't subtle -- the reviewers commented on it as well.

While I don't do video, I was struck by a specific example: the iPhone 13 Pro Max's automatic enhancement for dark scenes decided a barely visible wall was red and recently painted. It looked like someone had drawn a box in the photo and did a red fill.

I suspect it is only a matter of time before Samsung and Pixel's photos look as overprocessed. But, for now, we have some choice.

There’s no such thing. I doubt any reviewer making videos knows what they’re talking about unless it’s from a very few sources like DPReview.

The iPhone camera has never had skin smoothing - that’s something only Asian camera apps do. Deep Fusion is more like the opposite of that.

https://en.wikipedia.org/wiki/IPhone_XS#Excessive_smoothing_...

You were saying?

Actually I really didn’t like the denoising on that camera, it seemed poorly tuned in general…

As for cameras, 10x physical zoom is not something to neglect, Apple doesn't have anything to offer there. In past 2 weeks I have it, I've shot quite a lot of otherwise impossible kids, family, nature, animals etc. pics with this zoom. That was my main motivation for new phone - a camera for pictures I always have with me, unlike dedicated camera (2.5kg being just a detail, I didn't mind carrying it but I didn't have it a lot of casual times where great situation happened).

Another reason that swayed me - with this upgrade I wanted to improve headphones, priority being quality of sounds and battery life to play my collection of flacs. Airpods Pro were just not cutting it, went with Sennheiser momentum 2. You can't connect iphone and Senns via some above-default bluetooth protocol, they support aptX, Apple has their own solution and nothing else.

Rest of cameras are +-comparable, sometimes one wins other times the other. BUT - this over-processing that Apple is famous for is something I don't like, as long term DSLR shooter I 'have it in the eye' how processing should look like. A lot of people got so used to it though, they consider ie full frame pictures with normal level of processing bland and artificial compared to Apple's.

Kinda. I have a 4a because of the camera more or less. It was between that and the iPhone SE 2020.

I did kinda wanna switch up to Android (I've gone back and forth over the years. I don't lock myself into any one company's eco.)

I’m a professional photographer. Hell, she is too - she second shoots at my weddings. We have the best cameras and lenses money can buy…but they aren’t in her pocket when our baby does something cute.

iPhone 13 launch marketing/advertisement was almost entirely about camera.

https://www.youtube.com/watch?v=XKfgdkcIUxw

We did. My wife wanted an Iphone because it takes great pictures, and I got an S9 because I can tinker with the ROM.

I guess I sound like I’ve really drank the iKoolaid but I feel like we’d just constantly be frustrated if we weren’t using the same device family.

Interestingly, it's usually "more processing the better".

Maybe there's some way to monkey with it just right.

I think people are confusing low light noise reduction for something else.

No one is going to go dig for that, if you wish for folks to make such a comparison, it would be better to post a link so that everyone is on the same page.

Even with the s21+, the only way to really get decent shots from the camera in low/medium light before a recent update was to use modded Gcam builds. Even with the update I'd say it maybe compares to an iPhone 8 or x

Overprocessing and oversaturation were so bad before the s10+ that I remember not even bothering with the camera at all because of how frustrating it was to always get mediocre pictures.

You can shoot raw, but I just want the same normal workflow without the run around.

Samsung absolutely does not want someone to switch, and then find out that they "look worse" on their new samsung phone.

I guess its just a matter of whether or not they implement options for post-processing.

I even saw this on a video walkthrough of an apartment – they clearly had placed the lights right outside the window.

Finding a place to live is challenging.

We ended up with the virtual furniture for free. I think they wanted to figure out how it works and our place was small and boxy so an easy starting point.

Not that you could trust the ultrawide angle HDR photos much in the first place, now you have to put up with shoddily inserted virtual furniture wrecking your size/perspective references. It is an Alice in Wonderland trip...

Random example: https://www.redfin.com/CA/Santa-Monica/1319-Harvard-St-90404...

Most real estate photographers use HDR, and the higher end uses off-camera flash (as did I). There’s no nefarious reason behind this - it’s because photos where you can actually see out the windows look better. It’s also more like what your eyes would see rather than a crazy blown out white rectangle. When the house is on a lake you want people to see what the view is like. It was the rooms where I didn’t do that where I was hiding something, like an air conditioning unit being all you could see or whatnot.

The stupidly wide lenses most use? Yeah, that‘s deceptive. I also think it looks bad I tried to ride the line on using the tightest lens I could while still showing the room.

>Random example: https://www.redfin.com/CA/Santa-Monica/1319-Harvard-St-90404...

Honestly it's hard to tell without a reference picture. Looking at the first picture, it seems reasonable that the living room would be well lit because of huge windows. However, the section with the tall houseplants look nearly as bright as the open living room area, which seems doubtful.

They did the HDR so well that you couldn't tell it was HDR. Artificial HDR is a choice, just like natural HDR is in that photo.

In pictures 8 and 9 they didn't HDR, so the windows look like white rectangles and very overexposed. In picture 17 they did the HDR treatment on the horizontal windows, but did not on the skylights. Those skylights should show up as blue vs. white rectangles if they were doing the HDR treatment on the entire photo, etc.

I think they also use off camera flash and bounced it off the ceiling, which is why I think it looks kind of glowy-white with all the white walls and floors in that place.

This is definitely a thing. A lot of real estate photos are manipulated to replace the content of copyrighted photos and paintings, to hide unsightly views through windows, or to show clear skies when the photographs were taken on a rainy day.

A proper HDR photo should blind you when viewed on an HDR display, which are kinda expensive when they're desktop sized.

Overall I actually found it a great experience – I love having an ultra wide angle lens in my pocket and after a bit of time getting used to it, I find the 3x zoom a more useful focal length than the 2x zoom on my previous iPhone X – but I did find the over-processing (the "painterly" look) frustrating, and even more annoying is that it will often use the wide angle lens and upscale rather than using the 3x zoom (even with ProRAW enabled – see [1]).

I understand why the software makes these choices for the "average" user, but it would be nice to enable a "pro" mode which reduces noise reduction and favours the zoom lens in more situations.

I ended up using the stock camera app for "snapshots" (e.g. photos where I didn't care too much about the quality, or where I was just using the wide angle lens), and Lightroom Mobile's camera for shots where I wanted more control, or when I wanted to ensure that the zoom lens was being used (Halide is also good for this, but I found it convenient to have the photos in LR Mobile for processing immediately, even if Lightroom's UX is clunkier).

This actually worked pretty well for me – you can trust the stock camera to take a photo which will look good at mobile screen sizes even in challenging conditions, so it's great for "capturing the moment", while if you are taking a photo where you care about the details, it's usually not an issue to take a few extra seconds to open LR... but it would be nice to be able to do this in the stock camera.

The results with RAWs from Lightroom are actually pretty impressive IMO – there's more noise than the stock camera, but I prefer this to the smudged noise reduction look and I'm sure with some processing I can find a happy medium. Even the ultra wide photos are reasonably sharp.

This was a long way of saying that if you're frustrated by this issue, try a third party camera app and hopefully you'll find that you can get more out of the newer iPhone's great cameras, while still having the default camera there for quick snapshots!

[1] https://lux.camera/iphone-13-pro-camera-app-intelligent-phot...

I have an old Samsung Note 8; I was contemplating Iphone 13 Pro for my wife so we can improve the casual/random photos of our kids, but that'd be a deal breaker :-<

[context - we have Nikon d800, d7200, couple of d90's, V1, etc lying around the house so we do like photos, and being able to zoom or choose a lens is something we take for granted :]

It has always done this to some extent, but as the other link I posted ([1]) describes, this both happens more frequently with the 3x lens on the 13 Pro than the 2x lens on previous generations because the 3x zoom has a smaller aperture than the 2x, and also the effect is more noticeable, because it's blowing a 1x image up to 3x rather than 2x.

I hope Apple will offer some facility to tweak this in future as it seems to me it frequently chooses digital zoom rather than the zoom lens even in "not that challenging" conditions. I do get why they'd do this though – the reality is that if it always used the zoom lens, you'd end up with a lot more noisy/blurred photos due to physics – but it would be nice to say "I'm OK with that"!

One other thing I should mention is that you can't set a third party camera app as the default, so the camera button on the lock screen always opens the stock camera. In practice I don't find this a huge issue, as if I'm grabbing a quick snapshot from the lock screen it might be of some fleeting moment, in which case I'd probably rather get a usable digitally zoomed image, than a blurry optical zoom one.

[1] https://lux.camera/iphone-13-pro-camera-app-intelligent-phot...

It's the auto-macro functionality in the camera settings. The default behavior switches to the ultra-wide below the minimum focus distance of the other two cameras.

It’s just as convenient for quick sharing, and the photos are so much better.

Search for OpenMemories Framework, there's something of an unofficial SDK available.

Phones make this process seamless because the designers realized that the key to a good photo experience is the human experience of sharing.

And ditto ipads: since the first ipad came out I've taken a grand total of one photo with it. The aggressive cluster of lenses is a real downer since I can't lay it flat.

* Yes I know some people chose their phone by the camera and need the physics of a longer lens barrel. They are not wrong, nor am I: I'm just saying that one size doesn't fit all.

It was the hardware designers saying "welp, we could really use that extra few mm of depth for the camera lens, and keeping the rest of the body thinner is a better deal than beefing up the whole thing for those handful of users who don't use a case.

Maybe there's a non-case case out there for you that just sticks on the back and is thick enough to make the whole thing flush without adding any bulk around the sides? Or even just a strip of foamy tape or something that can run across the top of the device adjacent to the camera pop-out so that the device is able to sit flat on a table.

I'm not sure this is good design anymore. Perhaps if the bump was centered so the phone doesn't sit lopsided on a table, or maybe even a tapered design.

[1] https://imgur.com/a/UVt4FyH

My partner has an iPhone 13, though, and my impression was that it wasn't all that different for her, but clearly that's not the case— I should examine it more closely. I see from looking at some phone cases online, many of them do have a raised "frame" around the camera area, so I can see how that would contribute to issues with it rocking when set down on a flat surface.

EDIT: Okay, I realised she uses a popsocket on the back of hers, so this has never come up because it either sits on the table face down, or is face-up but propped up on the socket.

0: https://www.spigen.com/collections/iphone-13-pro-max/product...

isn't that what phone case is for?

Apple could make the camera flush by bulking out the back of the phone with more battery (though these days battery life is adequate for me so this wouldn't be as much of a win as it might have been a few years ago).

2. because they make them slippery as hell

etc.

I tried to use my phone without case to enjoy the finish and smaller dimensions, but the edges were already inconvenient to hold and God forbid I would leave it lying anywhere with vibrations on, it was slowly slipping on any surface which was not perfectly even, even without vibrations.