Hacker News new | ask | show | jobs
by arrrg 3923 days ago
There is nowhere to go, not with the available physical space.

I expect noise to get lower and lower and low light performance to increase, leading to sharper (due to less noise reduction) and less noisier images in the coming years, though, due to the small available physical space and small sensor probably always a step behind larger sensors (like the ones one would find in larger cameras).

Maybe it will be possible to close that gap? For many situations that gap is already quite small when, e.g. comparing many typical image viewing situations with images made under good lighting conditions. On close examination images from cameras with larger sensors would typically still be sharper, though, mostly due to them being able to have a higher resolution with lower noise (though, as I said, in many typical viewing applications those differences are, always depending on the photos and in which context it was made, sometimes imperceptible).

What will not be possible are zoom lenses, interchangeable lenses or, really, any kind of freedom with the lenses at all. Probably also unrealistic is a variable aperture and a sensor size that would allow one to play with depth of field in the first place. That is a fundamental difference that cannot be overcome within the physical envelope of a smartphone that is still recognizable a smartphone and not something else, some weird smartphone camera hybrid.

All that DSLR wankery is kind of pointless, though. The goal here is to make the very goddamn best camera that the physical envelope allows for and smartphones have done an astonishingly great job at that. Those are awesome cameras, no ifs and buts about it. Given what they have to work with they are unabashedly awesome.

DSLRs and other large-sensor cameras are fundamentally different beasts. Comparing them or even holding them up as some sort of great goal for smartphone cameras seems kinda … dumb? … I don’t know, pointless? … short-sighted? to me.

8 comments

> There is nowhere to go, not with the available physical space.

While true, also false.

Take something like the Sony Nex series; mirrorless digital system camera. Not much larger than a (big) smart phone. Sure, bigger than an iPhone. But it might not be that difficult to fit the "rest" of a smart phone into one of those. In fact, it probably have all the parts: battery, (touch?)screen, microphone, wireless radio (not cellular, but that could be changed), speaker (maybe needs to be added).

But a better way to go would probably be some kind of light-field technology. I think that holds more promise for better pictures in a similar form factor, and the possibility of ease of use.

Look at the lens of those devices … I mean, really, it’s a basic physics problem.

The device itself can obviously be very, very small. Obviously. No one is disputing that. But the lens is the issue if the sensor gets larger and a smartphone with Sony Nex sensor size (APS-C) is impossible.

There is nowhere to go with the available space. Not if you want to keep the smartphone actually smartphone sized.

I was thinking about something like:

http://www.kurtmunger.com/sony_nex_20mm_f_2_8id345.html

Sure, we're definitely talking something bigger than a standard iphone. So maybe not possible in a general market smart phone (I'd love to have the possibility of changing lenses on my phone, even if it wouldn't quite fit in the smallest pockets any more...).

The only other option would probably be some kind of lightfield tech, say a grid of 8x10 VGA resolution cameras that feed into a single software or ASIC processing unit...

I saw someone with a really neat camera. It was an Olympus lens + sensor in a package that was hardly bigger than a camera lens, which wirelessly talked to a smartphone for image display and controls. A much smaller add-on than an entire mirrorless camera, acknowledging that most people are already carrying an LCD screen and flash storage.
This. I own a Sony NEX because I want something portable that can take good pictures. I don't care that much about a mediocre telephone, web browser, and GameBoy, unless it's nearly as good at photography.
Ah, but the iPhones actually already use Sony sensors. One key difference between its sensor and the NEX/Mirrorless Alpha cameras is sensor size and the lens optics, which is why one might say there isn't much difference to make up (that can be made up) between the phone and cameras.

Also worth noting is that Sony cameras already run a customized and locked-down version of Android, though you wouldn't know from using it.

> There is nowhere to go, not with the available physical space.

I think that taking a classical digital still image on a tiny sensor is probably soon going to hit an "ISO-wall", but so far we haven't started doing much trickery with the images apart from simple noise/sharpness/toning treatments.

I think smartphones will lead the development of "cheats" to get better images. Imagine taking a portrait of 3 people in low-light by exposing a 1 second film (maybe including a couple of flashes). The processor then works (very hard) to build together single frame in which everyone's eyes are open, the moving subjects are frozen and sharp, and the shadow detail is pulled from the dark background by averaging out the noise from the thousands of frames.

I think the accelerometer should be used to trace the movement during this exposure, which can be used in combination with the anti-shake blur removal algorithms which are being introduced into various programs now.

Add to this some other technologies like deep learning that can be used to "guess" what's in a region of an image where we have noisy, obscured or OOF data.

I'm optimist, I think we will see not better classical images, but just better images. There will be lots of crazy "three-eyes"-artifacts produced by this, just like we see when programs try to stitch panoramas. Also, it will upset photography purists to no end. It will question the entire notion of what a photograph is.

Probably downvote bait, but all I can think of here is the age old saying, the best camera you have, is the camera you have with you.
Certainly true with traditional optics. I think computational optics will change things though - there's been research into using camera arrays for simulating a large aperture, for example. I don't actually know if this is physically feasible, but maybe in lieu of a single large sensor we simulate one with a ton of smaller ones (cool gifs: http://www.mit.edu/~sysun/MAS531_lightfield.html)
Well, it depends how you use the available space. With a simple stack of sensor and lens pointing out of the phone, there's very little height available. But if you were to arrange the optics down the length/width of the phone, you'd grant yourself rather more space. This patent talks of external add-ons, but the same principle would be applicable internally:

http://appleinsider.com/articles/14/01/28/apple-granted-two-...

And then there's the option of having multiple cameras taking the same shot, with different focal lengths (whether fixed or zoom) on each. You're still limited in dynamic range by the photosite size, but as we're seeing now, that's often not so huge a problem in many scenarios. One startup working on such a principle is Light:

http://www.imaging-resource.com/news/2015/04/18/photo-startu...

> Comparing them or even holding them up as some sort of great goal for smartphone cameras seems kinda … dumb? … I don’t know, pointless? … short-sighted?

It seems like the opposite of short-sighted to me; to look beyond the limitations of today's technology and find ways to make the impossible incrementally more achievable over time.

There is nowhere to go, not with the available physical space.

Where we are now is the "nowhere to go" of not long ago.

I had a Kodak DC-1. The notion of "a digital camera the size of a sugar cube" was laughable. Now we've got something even smaller taking pictures most people can't distinguish from a DSLR.

I expect the next breakthrough will be using the display itself as a platform for a phased-array "light as electromagnetic waves" sensor, allowing a software-defined virtual lens for a large-format imager (and oh so much more).

Nowhere to go? We've done amazing things with that nothing for the last 30 years. We're not gonna stop.

Yeah, but a lot of the limits we're hitting are tied to physics. But hey, we're used to trade offs and will go for slightly lower quality in exchange for ease of use and convenience
There is somewhere to go, the way technology is going. Phones have done lot's of catching up even if they have long ways to go... Many people that haven't ditched traditional cameras yet could possibly do it now. But either way it's always nice to compare good cameras with convenient cameras.