4 ms·

A lot of post processing that modern phones are doing isn't really practical though in post. For super low light shots for example, the iPhone will take hundre

A lot of post processing that modern phones are doing isn't really practical though in post. For super low light shots for example, the iPhone will take hundreds of exposures and merge them. If you were going to save this all as RAW, it's multiple gigabytes of data per frame.

badlucklottery 4y ago
> For super low light shots for example, the iPhone will take hundreds of exposures and merge them. If you were going to save this all as RAW, it's multiple gigabytes of data per frame. A bit of a digression but the reason why phones need so many exposures to get a decent low light image is due to their sensors/optics being relatively poor at low light. You would only need a small fraction of the exposures from a modern pro-grade camera to generate a comparable final image. This also lets you capture objects in motion in low light far, far better since you can get your required image data in a smaller temporal window.
formerly_proven 4y ago
"How would you do {night, portrait} mode on a real camera" has been brought up a few times in this post and the answer is: you don't, because it's not needed. I can prop an iPhone on the windowsill and let it do a 30 second night mode capture, and I can also prop my Nikon Z7 on the same windowsill and let it do a 2-5 second exposure. The latter is so much higher quality it's no comparison at all. (And 2s can be hand-held at 24 mm, unlike the 30s night mode) Though of course deep sky photographers do actually just this - many exposures composited together - but then we are talking about subjects so dim you need effective exposures of 30 minutes or so.
mirsadm 4y ago
It does not take hundreds of exposures. That is impractical and above all not a good idea. In low light the shutter speed is likely slow (1/10th of a second for a example). Even 16 frames is more than a seconds worth of frames. At that point the differences start to become too high for stacking to be worthwhile, especially because it is hand held. Secondly, the reduction of noise by merging is at best the sqrt (number of frames). Past even 12 frames the benefits start to be questionable.
- Xcelerate 4y ago
  > Secondly, the reduction of noise by merging is at best the sqrt (number of frames) This assumes the (typically Gaussian) noise is applied to a static image. Arguably, one could exploit the slight shakiness in a handheld shot to create an image with even less noise. As a thought experiment, consider thousands of shots of a perfectly static scene made with an idealized, noiseless camera that is moved a very tiny amount for each shot. You could continue improving the resolution of a generated, composite image quite a bit until warping due to camera displacement became noticeable. Recent techniques like this are actually being used for cryo-electron microscopy to create extremely high-resolution imaging of proteins.
  - mirsadm 4y ago
    Yes what you are referring to is implemented already by Google and probably others. The Google camera uses it to skip demosaicking by shifting the image by a pixel using OIS. Still the original post was under the impression that hundreds of exposures are merged which is just not the case

badlucklottery 4y ago

> For super low light shots for example, the iPhone will take hundreds of exposures and merge them. If you were going to save this all as RAW, it's multiple gigabytes of data per frame. A bit of a digression but the reason why phones need so many exposures to get a decent low light image is due to their sensors/optics being relatively poor at low light. You would only need a small fraction of the exposures from a modern pro-grade camera to generate a comparable final image. This also lets you capture objects in motion in low light far, far better since you can get your required image data in a smaller temporal window.

formerly_proven 4y ago

"How would you do {night, portrait} mode on a real camera" has been brought up a few times in this post and the answer is: you don't, because it's not needed. I can prop an iPhone on the windowsill and let it do a 30 second night mode capture, and I can also prop my Nikon Z7 on the same windowsill and let it do a 2-5 second exposure. The latter is so much higher quality it's no comparison at all. (And 2s can be hand-held at 24 mm, unlike the 30s night mode) Though of course deep sky photographers do actually just this - many exposures composited together - but then we are talking about subjects so dim you need effective exposures of 30 minutes or so.

mirsadm 4y ago

It does not take hundreds of exposures. That is impractical and above all not a good idea. In low light the shutter speed is likely slow (1/10th of a second for a example). Even 16 frames is more than a seconds worth of frames. At that point the differences start to become too high for stacking to be worthwhile, especially because it is hand held. Secondly, the reduction of noise by merging is at best the sqrt (number of frames). Past even 12 frames the benefits start to be questionable.

Xcelerate 4y ago
> Secondly, the reduction of noise by merging is at best the sqrt (number of frames) This assumes the (typically Gaussian) noise is applied to a static image. Arguably, one could exploit the slight shakiness in a handheld shot to create an image with even less noise. As a thought experiment, consider thousands of shots of a perfectly static scene made with an idealized, noiseless camera that is moved a very tiny amount for each shot. You could continue improving the resolution of a generated, composite image quite a bit until warping due to camera displacement became noticeable. Recent techniques like this are actually being used for cryo-electron microscopy to create extremely high-resolution imaging of proteins.
- mirsadm 4y ago
  Yes what you are referring to is implemented already by Google and probably others. The Google camera uses it to skip demosaicking by shifting the image by a pixel using OIS. Still the original post was under the impression that hundreds of exposures are merged which is just not the case