3 ms·

>by ignoring computational photography advancements I don't believe it's true. Dedicated cameras never "ignored" the means to get better dynamic range and sign

>by ignoring computational photography advancements I don't believe it's true. Dedicated cameras never "ignored" the means to get better dynamic range and signal/noise ratio with stacking. It's just not needed most of the time, being a very specific trade-off for certain scenes. And RAW processing software is pretty much embracing computational photography (calibrated optics correction, noise reduction, highlight reconstruction etc, including things like auto retouching and tonemapping if you don't want to do it manually). It hugely improved over the years, being able to pull out much better results from the same RAW shot. What they're lacking is miniaturization and speed. Big sensors and many megapixels mean slow reading and writing speeds; my smartphone can record RAW video at 4K@60fps which is out of reach for my camera. Modern super-sharp lenses became huge. Electronic viewfinder and variable aperture mechanism can't be made reasonably small. Thankfully, the need for mechanical shutter seems to slowly go away as recent Nikons show. I hope there will be more super-compact cameras like Sigma fp in the future; combine it with a pancake lens and proper calibration ("computational photography") to compensate for the lack of corrective elements in pancakes, and you got yourself a pocket camera with vastly better capabilities. Smartphones can sometimes get nice results and are pretty good at motion compensation in deep stacks, but they are slow, un-ergonomic and very restricted in what you can shoot well with them. And the people who use dedicated cameras tend to shoot RAW with smartphones for best results as well. (stacking is available in bayer domain so it's not a problem)

Someone1234 5y ago
> Dedicated cameras never "ignored" the means to get better dynamic range and signal/noise ratio with stacking. I respectfully disagree. No dedicated format for stacks (instead just filenames), no first party software to combine the stacks, and absolutely no on-board combination. Heck on a fairly recent camera I own, you cannot even configure the camera to always take a stack cellphone style. You have to bind to button to open a menu, to select stack size, every single shot. > And RAW processing software is pretty much embracing computational photography (calibrated optics correction, noise reduction, highlight reconstruction etc, including things like auto retouching and tonemapping if you don't want to do it manually). True. However, the market has shown time and time again that onboard is more popular than post-progressing. People want to snap and post, they don't want to snap and then post the next day/week after they've post processed it in PhotoLab or Lightroom.
- vladvasiliu 5y ago
  > absolutely no on-board combination. This is false. My 2016 (I think?) Olympus Pen-F does multiple-exposure combination on-board. The results aren't always the best, since you can't control much, and it's JPEG only, but it does do it. The reference manual: https://www.manualslib.com/manual/1081745/Olympus-Pen-F.html?page=95 https://www.manualslib.com/manual/1081745/Olympus-Pen-F.html...
- hatsunearu 5y ago
  The whole stacking argument is, in my opinion, a little silly. Why do we need stacking? Because it allows you to do exposure bracketing to get better dynamic range. Now the question is, in a high end mirrorless like, say an a7r4, you get something like 12-14 stops of dynamic range (expressing this as a number is kind of ambiguous). Do you really need one or two stops of dynamic range with bracketing? I don't think so. If you expose smartly you can absolutely make use of the incredible dynamic range. The only problem I see is on iPhone, it shoots in HEIC which is a great format because it lets you encode much higher dynamic range (and the software support is there to display it correctly). Check out a recent iPhone and take a picture with the sun and something medium-dark and you'll see that the sun is eye searingly bright whereas the medium-dark subject is still visible. I wish there was a flow that lets me produce HEIC images (AFAIK even Photoshop doesn't allow HDR workflows). A lot of the pro imaging workflow is only "aware" of printed end results and low-end digital images like Instagram or JPEG web publishing. Very frustrating in my opinion.
namibj 5y ago
The old (by now) CMOSIS CMV12000 does 4000x3000 pixels at 10bit per pixel and 300 Hz. It reads the 4000 pixel rows in parallel, from the bottom and the top. There's no hard reason why this couldn't be extended to a larger sensor, with the per-row time at-worst linearly increasing with the row count (due to capacitance on the column-lines). But even then, the row count would only have quadratic influence on the readout time, and the column count not at all. Combines to, for fixed aspect ratio, a constant number of pixels-per-second. So going to the 8k equivalent of 8000x6000 pixels (48MP), you'd quarter those 300fps and still have 75fps left. Assuming you take wide-screen video of 8000x4320 (some sacrificial columns that don't matter for speed and double the rows of "4K" (2160p)), the reduced row count would take that up to ~104fps. Oh, that chip also has a pipelined electronic global shutter, so you can take the long exposure of a HDR stack while the short exposure is being read out. Also no jello and shutter down to 1/50000. How big of a sensor were you thinking, that RAW video recording would be inconceivable for the big sensor? And yeah, sure, you'll have to do something with those many gigabits. It's about 36 Gbit/s for the above example. A PCIe4x4 connection would do, but you'd need about 2-and-a-half PCIe4 M.2 SSD's worth of sustained sequential write to handle uncompressed RAW.

Someone1234 5y ago

> Dedicated cameras never "ignored" the means to get better dynamic range and signal/noise ratio with stacking. I respectfully disagree. No dedicated format for stacks (instead just filenames), no first party software to combine the stacks, and absolutely no on-board combination. Heck on a fairly recent camera I own, you cannot even configure the camera to always take a stack cellphone style. You have to bind to button to open a menu, to select stack size, every single shot. > And RAW processing software is pretty much embracing computational photography (calibrated optics correction, noise reduction, highlight reconstruction etc, including things like auto retouching and tonemapping if you don't want to do it manually). True. However, the market has shown time and time again that onboard is more popular than post-progressing. People want to snap and post, they don't want to snap and then post the next day/week after they've post processed it in PhotoLab or Lightroom.

vladvasiliu 5y ago
> absolutely no on-board combination. This is false. My 2016 (I think?) Olympus Pen-F does multiple-exposure combination on-board. The results aren't always the best, since you can't control much, and it's JPEG only, but it does do it. The reference manual: https://www.manualslib.com/manual/1081745/Olympus-Pen-F.html?page=95 https://www.manualslib.com/manual/1081745/Olympus-Pen-F.html...
hatsunearu 5y ago
The whole stacking argument is, in my opinion, a little silly. Why do we need stacking? Because it allows you to do exposure bracketing to get better dynamic range. Now the question is, in a high end mirrorless like, say an a7r4, you get something like 12-14 stops of dynamic range (expressing this as a number is kind of ambiguous). Do you really need one or two stops of dynamic range with bracketing? I don't think so. If you expose smartly you can absolutely make use of the incredible dynamic range. The only problem I see is on iPhone, it shoots in HEIC which is a great format because it lets you encode much higher dynamic range (and the software support is there to display it correctly). Check out a recent iPhone and take a picture with the sun and something medium-dark and you'll see that the sun is eye searingly bright whereas the medium-dark subject is still visible. I wish there was a flow that lets me produce HEIC images (AFAIK even Photoshop doesn't allow HDR workflows). A lot of the pro imaging workflow is only "aware" of printed end results and low-end digital images like Instagram or JPEG web publishing. Very frustrating in my opinion.

namibj 5y ago

The old (by now) CMOSIS CMV12000 does 4000x3000 pixels at 10bit per pixel and 300 Hz. It reads the 4000 pixel rows in parallel, from the bottom and the top. There's no hard reason why this couldn't be extended to a larger sensor, with the per-row time at-worst linearly increasing with the row count (due to capacitance on the column-lines). But even then, the row count would only have quadratic influence on the readout time, and the column count not at all. Combines to, for fixed aspect ratio, a constant number of pixels-per-second. So going to the 8k equivalent of 8000x6000 pixels (48MP), you'd quarter those 300fps and still have 75fps left. Assuming you take wide-screen video of 8000x4320 (some sacrificial columns that don't matter for speed and double the rows of "4K" (2160p)), the reduced row count would take that up to ~104fps. Oh, that chip also has a pipelined electronic global shutter, so you can take the long exposure of a HDR stack while the short exposure is being read out. Also no jello and shutter down to 1/50000. How big of a sensor were you thinking, that RAW video recording would be inconceivable for the big sensor? And yeah, sure, you'll have to do something with those many gigabits. It's about 36 Gbit/s for the above example. A PCIe4x4 connection would do, but you'd need about 2-and-a-half PCIe4 M.2 SSD's worth of sustained sequential write to handle uncompressed RAW.