The reason someone carries a DSLR today is because of the optics, the speed and the control mechanisms. Having shutter speed, aperture, white balance etc. at your fingertips when shooting. Having the ability to switch from a 105mm portrait lens to a 20mm fisheye.
I can see that in 5 years smartphone speed, coupled with advanced phase detection AF, will make smartphone as fast as today's pro DSLR bodies (the likes of the D5, Canon 1D line etc..)
The need to quickly adjust settings(f/stop, shutter, etc..) can be greatly diminished by having even more advanced software. Like a bokeh/portrait mode, where the lens is kept wide open and the phone maintains all other settings, include fake bokeh via depth map.
Different lens focal lengths is a big differentiator. However, if the sensor has sufficiently high resolution, the software can crop the sensor up an extend to largely simulate telephoto. Granted, compression won't be achieved, but it's largely there...
if the sensor has sufficiently high resolution, the software can crop the sensor up an extend to largely simulate telephoto
The software will in this case be inventing an image, not capturing one. Don't get me wrong - I'm delighted with the improvement in quality that pocket cameras, embodied in smartphones, have achieved in a mere decade. But physics is still physics, and a 6x5mm sensor can only capture so much light.
6x5mm of your eyes retina, coupled with our brain does a pretty good job, so there is no reason we couldn't achieve similar results.
I love using my old film cameras, and as an amateur, I actually only use an iPhone and film cameras.
I don't really see a point using a bulky and short lived dslr to produce nor quick nor great pictures.
As a professional, I guess the issues are different though.
I'm anything but a professional at photography. But the thing about the way our brains interpret visual input is that it's lossy as hell. I don't need my camera to be the same.
but traditional physical limitations of an image sensor's light gathering ability is always in context of the sensitivity of the sensor. If the sensor is made even more sensitive, there is more than enough light to create images.
More noise, too, which scales as 1/(pixel size * sensitivity) - that is, decreasing pixel size and increasing sensitivity both raise the noise floor.
This can be filtered to some extent - indeed modern phone cameras already do so quite heavily, to overcome the limitations of their already small pixel size and already high sensitivity. But doing so costs the same detail you'd need to try to invent a "simulated telephoto" image, which would give your fake telephoto process even less to work with than otherwise - guessing you have a neural net in mind here, and while I'm not about to argue that a sufficiently well-trained net won't produce some kind of result given an input of sufficient similarity to its training set, I see no reason to expect that result to bear any particularly photographic similarity to the original input.
I mean, don't get me wrong - what you seem to be suggesting isn't all that dissimilar from how we currently understand the human brain's own optical system to work. But I don't think it is especially likely that many such brains will happily accept another neural network's best-effort guesses as the output of a process that we've all learned to expect will give us representations as precise as is within the capabilities of the devices we use to make them.
Comments
I can see that in 5 years smartphone speed, coupled with advanced phase detection AF, will make smartphone as fast as today's pro DSLR bodies (the likes of the D5, Canon 1D line etc..)
The need to quickly adjust settings(f/stop, shutter, etc..) can be greatly diminished by having even more advanced software. Like a bokeh/portrait mode, where the lens is kept wide open and the phone maintains all other settings, include fake bokeh via depth map.
Different lens focal lengths is a big differentiator. However, if the sensor has sufficiently high resolution, the software can crop the sensor up an extend to largely simulate telephoto. Granted, compression won't be achieved, but it's largely there...
5 years is a long time.
The software will in this case be inventing an image, not capturing one. Don't get me wrong - I'm delighted with the improvement in quality that pocket cameras, embodied in smartphones, have achieved in a mere decade. But physics is still physics, and a 6x5mm sensor can only capture so much light.
6x5mm of your eyes retina, coupled with our brain does a pretty good job, so there is no reason we couldn't achieve similar results.
I love using my old film cameras, and as an amateur, I actually only use an iPhone and film cameras. I don't really see a point using a bulky and short lived dslr to produce nor quick nor great pictures.
As a professional, I guess the issues are different though.
I'm anything but a professional at photography. But the thing about the way our brains interpret visual input is that it's lossy as hell. I don't need my camera to be the same.
but traditional physical limitations of an image sensor's light gathering ability is always in context of the sensitivity of the sensor. If the sensor is made even more sensitive, there is more than enough light to create images.
More noise, too, which scales as 1/(pixel size * sensitivity) - that is, decreasing pixel size and increasing sensitivity both raise the noise floor.
This can be filtered to some extent - indeed modern phone cameras already do so quite heavily, to overcome the limitations of their already small pixel size and already high sensitivity. But doing so costs the same detail you'd need to try to invent a "simulated telephoto" image, which would give your fake telephoto process even less to work with than otherwise - guessing you have a neural net in mind here, and while I'm not about to argue that a sufficiently well-trained net won't produce some kind of result given an input of sufficient similarity to its training set, I see no reason to expect that result to bear any particularly photographic similarity to the original input.
I mean, don't get me wrong - what you seem to be suggesting isn't all that dissimilar from how we currently understand the human brain's own optical system to work. But I don't think it is especially likely that many such brains will happily accept another neural network's best-effort guesses as the output of a process that we've all learned to expect will give us representations as precise as is within the capabilities of the devices we use to make them.