Unfortunately, camera manufacturers have been very slow to adopt computational photography techniques in professional grade bodies. We're just now starting to see things like image stacking (usually called "digital ND filter" or other such nonsense). I'd like to see in-camera HDR, better processing, etc.

Hopefully the camera industry steps up the quality of it's software before it collapses into obscurity. Right now the only thing keeping it afloat is the fact that we haven't found good ways to take the effects of large apertures and longer focal lengths.

I feel like the moment phone manufacturers are able to convincingly fake something like an 85mm f/1.4, we're just going to be done with full-frame cameras.

  are able to convincingly fake something like an 85mm f/1.4,

If that were the case we have a good long time then, most likely; Really what you are talking about is how you can't beat the physics of large sensors and lenses; for now 'faking it' isn't anywhere close, but also the techniques being pursued aren't likely to get there.

I suspect in-camera processing though is a bit of a dead end for pro and prosumer cameras because your laptop/desktop is going to be able to do a much better job, and your camera just isn't going to have a good workflow for post-processing and organizing photos.

For phone camera use it makes more sense, because people are rarely doing much more than delete these two, crop this one a little and send it to mom.

Don't get me wrong, phone cameras are having a ton of effort put into them and have way more resources than conventional cameras - but the effort is almost entirely being put into making something small and light work pretty well under most circumstances, not into making something great.

What's more important is that most people don't care for most applications. More people use to buy "good" cameras because their other options were poor or nonexistent. A top of the line current phone camera may not be great, but it's good enough for most people.

I don't know what this means. It could well kill off the "prosumer" market and leave professional cameras an expensive niche product. It just likely won't be because phone cameras got as good at creating pictures.

> I suspect in-camera processing though is a bit of a dead end for pro and prosumer cameras because your laptop/desktop is going to be able to do a much better job, and your camera just isn't going to have a good workflow for post-processing and organizing photos.

Something like Night Sight as a post-processing method might not be practical. You need many short exposures instead of one long exposure. So basically, you've got full-resolution raw video to deal with instead of a single shot. More efficient to process that before storing it and transferring it.

Based off of how cameras are currently used (at least from my limited sample size), I don't think that would really matter.

People regularly shoot in various "raw" formats which are significantly larger in size so that they have more flexibility in how to post-process. They capture a whole bunch of extra data, move it around, and generally store it indefinitely knowing they will throw a lot of it away in the final photo so that they can make the decision what to throw away later while they're looking at a 27" colour-corrected screen instead of a small camera screen.

People don't do HDR on-camera, they generally shoot multiple exposures and adjust in software after to get the effect and balance they're looking for which is basically Night Sight to a lesser degree.

A bit of extra disk space and transfer time is small peanuts and already fits the workflow most photographers are used to.

Yea. You can't beat the physics of big optics... but there ARE ways to get around some of the limitations of smaller optics.

You can fake depth of field, use image stabilization and image stacking to gather more light with a smaller aperture, utilize "super resolution" and/or neural network techniques to get or fake more detail, etc.

I'd love to see some of the awesome work being put into phones carry over into full-frame cameras before the whole pro-sumer market and a bunch of the pro market dissolve.

Image stabilization and stacking both started in SLR's before phone cameras were doing it. The best image stabilization is still opto-mechanical, of course.

Nobody fakes depth of field that well, and super resolution (the algorithmic kind, not the real kind) doesn't work that well either. We've been doing research in those areas for 25+ years now, and while some real progress has been made it's not magic.

Phone cameras are doing a lot of heroic stuff to get around the limitations of their optics, and it results in "hey, that's not bad". Nothing wrong with that!

But I don't see why you would want to see that stuff in a digital SLR, say, though - if you take your photography seriously enough to lug that thing around, you're doing post processing on a computer anyway, and you can do all that stuff better there (at least in theory). Now if you were asking why photo processing software on your desktop wasn't absorbing or improving on everything in the phones, that would make more sense to me.

Well first of all, there are things that can be done in camera better than they can be done in post in any reasonable amount of time. Image stacking is one of those, since alignment even with the best software is spotty, and cameras could include fairly robust accelerometers to do a bunch of the work in the background. There's also currently no depth info in DSLR images, so image processing involving depth of field isn't going to work unless you do the work by hand, which is quite laborious. In addition, it would be very nice if HDR pictures could be taken by my full-frame camera. I don't necessarily want or need to deal with bracketing in lightroom just to get an extra couple stops of dynamic range.

You're right that the best image stabilization is opto-mechanical, but it would be very nice to have access to digital image stabilization as well as in-body and in-lens stabilization. Again, a decent accelerometer can go a long way.

Honestly, most of what I'd like to see has involved image stacking in one form or another. Digital ND filters are the same idea.

And don't get me started on ergonomics and usability. I definitely picked the worst camera brand in terms of user interface (sony), but all the DSLRs and mirrorless cameras I've touched are noticably lacking in that department. I don't necessarily need a giant phone screen like my S9+, but I would expect a device costing $2000+ to at least have a fairly high resolution OLED panel on the back of it with solid touch controls. Meanwhile touch controls are distinctly lacking in any camera I've tried.

Software is important too!

> Well first of all, there are things that can be done in camera better than they can be done in post in any reasonable amount of time. Image stacking is one of those, since alignment even with the best software is spotty, and cameras could include fairly robust accelerometers to do a bunch of the work in the background. There's also currently no depth info in DSLR images, so image processing involving depth of field isn't going to work unless you do the work by hand, which is quite laborious. In addition, it would be very nice if HDR pictures could be taken by my full-frame camera. I don't necessarily want or need to deal with bracketing in lightroom just to get an extra couple stops of dynamic range.

> You're right that the best image stabilization is opto-mechanical, but it would be very nice to have access to digital image stabilization as well as in-body and in-lens stabilization. Again, a decent accelerometer can go a long way.

> Honestly, most of what I'd like to see has involved image stacking in one form or another. Digital ND filters are the same idea.

> And don't get me started on ergonomics and usability. I definitely picked the worst camera brand in terms of user interface (sony), but all the DSLRs and mirrorless cameras I've touched are noticably lacking in that department. I don't necessarily need a giant phone screen like my S9+, but I would expect a device costing $2000+ to at least have a fairly high resolution OLED panel on the back of it with solid touch controls. Meanwhile touch controls are distinctly lacking in any camera I've tried.

> Software is important too!

I would like to see a panorama mode also vs firing up stitcher applications. All these should be basic firmware upgrades tbh even on the older models. Features available on a CoolPix entry level camera should not be missing from prosumer DSLRs.

Panoramas are one of the things I wish I hadn't left out of that comment. It blows my mind that a phone can do this so much better than a camera unless you invest in special mounts and such.

I got a Canon Rebel SL3 recently (APS-C, not full-frame, but still...) and it can do HDR on-camera. I wish it would save each individual exposure too instead of just the composite though. It also has touch controls on the display and a REST API for remote control via wifi. And this is not that expensive a camera.

Sounds like another direction to expand RAW formats. Accelerometer data is small, burst shots wouldn’t be that hard to take diff and store along. Then you could hypothetically feed it to Photoshop 2022 and run a 3D reconstruction.

To be fair, Sony cameras have high resolution OLED panels, in the viewfinder.

I don't think the convincing fake will ever happen anytime soon (no ML will give you that pixel-perfect sharpness gradient along each strand of air on a 30MP image).

Generally speaking I also haven't seen a phone picture yet, which beats my cams at sharpness (which is simple physics...). It's pixelpeeping ofc but I can happily look at my pictures right from the raw converter w/o any sharpening or AI-magic applied and they are sharp and crisp. Do that with you phone-raw or jpeg. You'll spot a difference even on a phone screen...

Also, I don't think that these things should be put in camera software as a feature (making them subject to obsolescense as phones...) but instead they should just add a very basic scripting API (=tool) so you can store all the things you need for postprocessing and control capture (like the lua-scripts on CDHK).

I think it's important to bear in mind that the phone doesn't have to beat the full-frame camera at this game. It only needs to be "good enough". The fact that the phone is much more convenient to carry and use already gives it a huge leg up. What's the saying? "The best camera is the one you have with you"

If someone can make a phone that gives convincing fake bokeh and covers wide, normal, and tele focal lengths while producing images that are even remotely comparable to a "real" camera in an 8x10 print... that person will demolish what's left of the camera industry.

  "The best camera is the one you have with you"

Right, which is why the low end consumer camera market has basically died. For average use, a phone is already "good enough".

So another plausible path is that the damage has already been done, and what is left of the prosumer and pro level stuff will remain stable.

Note, I'm not saying that I think that is what will happen, I haven't thought about it much. But there is nothing in current tech development that suggests the phones will bridge that gap (to prosumer level image quality) any time soon...

I think you might be right. A lot of the damage already has been done... but the pro and prosumer camera markets are still shrinking pretty quickly.

It's a concerning trend, which makes me hope that camera manufacturers get ahead of the trend by introducing features that I feel are lacking in "real" cameras.

Well, if you look at Nikon DSLRs: if you are shooting landscapes or portraits and are ok with the ergonomics of your gear/not printing _really_ big posters: your D750 is still state of the art (6 years old), so why buy new. What is a little bit disconcerting is, that everyone is jumping the full-frame wagon and not going the niche/mini-series way, they should probably go (this might lead to a disruptive change with a lot of knowledge lost in bankruptcies – which would be bad).

basically, except for the family pictures application, documentation (traffic accident) or "art"/hobby, noone needs a camera anymore, because today, everything there is, is already documented a thousand times, available in the mediocre quality for free, anytime, with an internet connection.

The first two categories are well served by any smartphone camera nowadays, which leaves the third. And here I think that this is still served best by a dedicated device because you are basically really interested in taking pictures of a certain quality. And no, the current state of the art in fake-bokeh will not really hit this group (it's a very small group though and the instagram-people with their nice portraits will probably go for the fake too. Though they probably could just do a 3D-scan of the person and then do a render (no more bad hair. no more bad skin. pure perfection). Doesn't necessarily have a lot to do with photography though).

It’s not just the look of the larger aperture and longer focal lengths, The sharpness is abysmal. I can’t stand that watercolor look that every smartphone photo has.

Also dynamic range, but that’s much easier to fake.

It's easier to fake, but it's still not the same. By that, I mean that the final image might have that look, but you can't do anything with it after the fact. With a RAW image with true dynamic range, you can do whatever you want with it and change it up as you wish.

The compression on the images is also another thing contributing to the watercolor look and lack of sharpness. Cell phone cameras are worried about space constraints on the device itself and across the network.

Most phones I've dealt with actually allow you to output raw image files.

I'm not saying they have the same dynamic range as my A7iii, but they're not compressed or color-value limited like JPEG.

True, but there are going to be ways to make smartphone images sharper. You can already do some work in Lightroom to mostly bridge the gap.

IMO professional camera manufacturers need to get ahead of that by putting computational photography techniques in their full frame bodies.

Yes, but ultimately no matter how much computational photography and odd filter arrays you throw it, the diffraction limit will kill you with a 2mm wide smartphone camera.

IMHO, they should mostly optimize the bits of the pipeline that are needed to capture the kind of data needed (e.g. recording short high speed bursts etc), and maybe make ad-hoc data transfer of the camera nicer. More flexibility to do it on another device and update the software there quickly, while keeping the camera platform stable, and the battery life jump downwards from DSLR to mirrorless is already painful enough.

An 85 1.4 vs an equivalent smartphone lens is literally in the 100s of times more light. It's quite crazy, I don't think it'll ever be doable.

Anyways, smartphone cameras are and will be for a long time stuck because of their diffraction limited sensors due to low lens diameters.

I've swithched to Olympus about a year ago and was surprised to find out they are doing some nifty things. Alas, with all that FF hype they are not getting much attention, despite having some smazing things to offer.

You'd think there'd be an opportunity for a disruptive newcomer here, but I guess there are enormous barriers to entry on both sides. Perhaps the watershed will be when one of the major camera manufacturers offers a device with good optics that is also capable of running user-installed software, whether that device is a conventional form-factor camera, or perhaps a sensor/optics-only box that works in conjunction with a phone.

EDIT: Nevermind, it looks like this has happened— Sony released some lens-only cameras in 2013 and it doesn't seem like too much has come of it: https://lifehacker.com/how-to-use-your-smartphone-to-get-bet...

Fundamentally, phones just give bad workflow for photo processing and organizing. The more you care about this, the harder it is to justify this approach as your primary workflow.

Sure, but no one's doing processing or organizing on the back of a DSLR either. Maybe the real problem here is that 13 years after the original iPhone, iCloud and Google Photos are still the least-bad solutions for archiving a family photo stream from multiple devices— there's a missing middle option between those and the prosumer solutions like Lightroom and Aperture (which in turn have their own issues, like the fact that AFAIK they mostly tackle organizing a local library; you're largely still on your own when it comes to things like collaboration and managing backups).

Anyway, I would have thought the dealbreaking issues with an augmented phone/camera would more be around the dearth of tactile controls, but that can be at least partially addressed by moving some of them to the strap-on accessory.

I think we agree.

I don't see a gap in camera-processing for either phones or DSLRs, because the experience sucks on both. But you are right there is a gap between "consumer" software and "pro" software that could be addressed.

I don't know if offloading a bit of processing to your phone from a camera would fix much really. Ergonomics of cameras is much better for taking photos, and there isn't much in there you don't need.

I guess you could get rid of the screen entirely, and some of the user controls. That could be nice but would be difficult from a product point of view, as you would be dependent on the whims of a larger market. See all the difficulties getting standardized on things for car, where there is much more incentive...

I agree with both of you.

I'm hoping to hit a chunk of that "middle option" you spoke of with PhotoStructure (details in my profile). I'm building sharing next, so hopefully that aspect will be addressed as well. Hosting the software yourself addresses privacy concerns. The beta is being offered for free in exchange for feedback.