4 ms·

EDIT: Thanks for the replies, I see now that they're talking about pixel binning. But doesn't the iPhone 13 sensor also have sensor "sub-pixels" that it's gro

EDIT: Thanks for the replies, I see now that they're talking about pixel binning. But doesn't the iPhone 13 sensor also have sensor "sub-pixels" that it's grouping into effective pixels to get a 12 MP camera? i.e. does the iPhone 14 do something different with the data coming from the sensor, or is the sensor different (for example, it allows some sub-pixel control that the iPhone 13 sensor does not)? Apple says the sensor (area I suppose) is 65% larger, but that's not enough to account for this 4x'ing. Thanks! --- I might be totally wrong, but could the 48 MP not be... real megapixels of any one camera or sensor, but a software fusion of the 3 cameras' sensors? Is that possible? There's language on the iPhone page that makes me think this -- "So the new sensor groups four pixels together into one large quad pixel, gathering 4x more light and producing spectacularly better photos at the practical 12MP size" What does "at the practical 12MP size" mean? Which "four pixels"? Where are the pixels captured/coming from? There's also "For the ultimate combination of detail and flexibility, shoot in ProRAW at 48MP resolution. A new machine learning model delivers sharper images and less noise for unprecedented detail." How much is the 48 MP a mix of sensor fusion (which could be purely software, or running on a chip, but still, post-sensor) and AI versus an actual 48 MP sensor? --- I have similar feelings about the crash detection (accelerometer-based) and even the satellite (5G-based?) capability. Are these features just... rebranded existing things // new software features?

quixoticelixer- 4y ago
They are actual pixels all from one sensor. They are put into groups of 4 to get 12 effective megapixels.
- quixoticelixer- 4y ago
  https://images.anandtech.com/doci/17563/36082843_575px.jpg https://images.anandtech.com/doci/17563/36082843_575px.jpg
  - gregsadetsky 4y ago
    Right, that's a Bayer arrangement. But... previous (iPhone 13) sensors also presumably have a similar Bayer arrangements and also group the sensor's pixels (which we could call "sub-pixels") into "effective pixels", right? So what's different here? Are there truly 4 times more sub-pixels? Or is this a new algorithm to interpret the sub-pixels data...? But the sensor is the same as the previous generation?
    - quixoticelixer- 4y ago
      No it's not a bayer arrangement. There are 4 times as many pixels, they are just being over-sampled. The sensor is not the same as the previous generation. Each pixel is grouped together with three others for most applications (taking photos etc) but they can also be used individually (It said it could do 48MP videos)
LegitShady 4y ago
google "pixel binning". It improves signal:noise ratio.
- gregsadetsky 4y ago
  Thanks, that makes sense. I did remember that sensors have "sub-pixels" that are grouped, and that you can access the sub-pixel to get "4 times" more pixels. But... what's the difference between the iPhone 14 and 13 then? The iPhone 13 presumably also has a Bayer sub-pixel arrangement on its sensor? What difference/innovation is happening with the iPhone 14 sensor? That the sub-pixel data can be processed, where in the case of the iPhone 13 it was inaccessible? Thanks!
  - zamadatix 4y ago
    Pixel binning isn't about having 4 subpixels per pixel it's about multiplying the number of supixels per pixel by 4. On a typical Bayer arrangement with 4:1 binning that means 4 red, 4 blue, and 8 green subpixels per pixel (total of 16) instead of 4 pixels each with 1 red, 1 blue, and 2 green subpixels. I've never really seen a good explanation for why this is anything special vs just making the image smaller post capture in a non-naive way (like when they are already running it through the AI offload engine) considering all the hype it gets.
  - LegitShady 4y ago
    > Thanks, that makes sense. I did remember that sensors have "sub-pixels" that are grouped, and that you can access the sub-pixel to get "4 times" more pixels. It's sort of backwards. You can get access to sub pixels to get '4 times' more data, but pixel binning is actually the other direction -> grouping pixels together. A lot of camera performance is related to photosite size - larger photosites for light to hit, better performance/more sensitivity. The problem is that if you have a camera sensor in a DSLR, you have room for big pixels, but when you have the small sensors in phones you don't have the room for large pixels and giant MP count for marketing. So pixel binning treats groups of pixels like one large pixel to try and improve sensor performance while still maintaining large megapixel counts for marketing. In terms of innovation we haven't heard anything. There are already smartphones with pixel binning on the market. This is apple just saying Us Too unless they've released information on a new unique binning technique I haven't seen.

quixoticelixer- 4y ago

They are actual pixels all from one sensor. They are put into groups of 4 to get 12 effective megapixels.

quixoticelixer- 4y ago
https://images.anandtech.com/doci/17563/36082843_575px.jpg https://images.anandtech.com/doci/17563/36082843_575px.jpg
- gregsadetsky 4y ago
  Right, that's a Bayer arrangement. But... previous (iPhone 13) sensors also presumably have a similar Bayer arrangements and also group the sensor's pixels (which we could call "sub-pixels") into "effective pixels", right? So what's different here? Are there truly 4 times more sub-pixels? Or is this a new algorithm to interpret the sub-pixels data...? But the sensor is the same as the previous generation?
  - quixoticelixer- 4y ago
    No it's not a bayer arrangement. There are 4 times as many pixels, they are just being over-sampled. The sensor is not the same as the previous generation. Each pixel is grouped together with three others for most applications (taking photos etc) but they can also be used individually (It said it could do 48MP videos)

LegitShady 4y ago

google "pixel binning". It improves signal:noise ratio.

gregsadetsky 4y ago
Thanks, that makes sense. I did remember that sensors have "sub-pixels" that are grouped, and that you can access the sub-pixel to get "4 times" more pixels. But... what's the difference between the iPhone 14 and 13 then? The iPhone 13 presumably also has a Bayer sub-pixel arrangement on its sensor? What difference/innovation is happening with the iPhone 14 sensor? That the sub-pixel data can be processed, where in the case of the iPhone 13 it was inaccessible? Thanks!
- zamadatix 4y ago
  Pixel binning isn't about having 4 subpixels per pixel it's about multiplying the number of supixels per pixel by 4. On a typical Bayer arrangement with 4:1 binning that means 4 red, 4 blue, and 8 green subpixels per pixel (total of 16) instead of 4 pixels each with 1 red, 1 blue, and 2 green subpixels. I've never really seen a good explanation for why this is anything special vs just making the image smaller post capture in a non-naive way (like when they are already running it through the AI offload engine) considering all the hype it gets.
- LegitShady 4y ago
  > Thanks, that makes sense. I did remember that sensors have "sub-pixels" that are grouped, and that you can access the sub-pixel to get "4 times" more pixels. It's sort of backwards. You can get access to sub pixels to get '4 times' more data, but pixel binning is actually the other direction -> grouping pixels together. A lot of camera performance is related to photosite size - larger photosites for light to hit, better performance/more sensitivity. The problem is that if you have a camera sensor in a DSLR, you have room for big pixels, but when you have the small sensors in phones you don't have the room for large pixels and giant MP count for marketing. So pixel binning treats groups of pixels like one large pixel to try and improve sensor performance while still maintaining large megapixel counts for marketing. In terms of innovation we haven't heard anything. There are already smartphones with pixel binning on the market. This is apple just saying Us Too unless they've released information on a new unique binning technique I haven't seen.