Pages: 1 2 [3]   Go Down

Author Topic: Sony A9 - SonyAlphaRumors (SR5) information  (Read 12117 times)

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2794
Re: Sony A9 - SonyAlphaRumors (SR5) information
« Reply #40 on: April 12, 2016, 05:03:36 am »

If the stills camera is able to better focus its 6MP and better freeze motion by using a 1/1000 s shutter, then I'd suggest that it is still potentially valuable as an action camera.

That said, MP seems to be cheap these days. Manufacturers can offers lots of them in a nice package with low noise, sufficient battery life, reasonable frame rates. So why strive for less?

-h

If the shutter speed's the same, then the amount of motion blur will be the same. Just that the 33MP sensor will show the blur smeared over more pixels. The amount of blurring in the whole image will still be the same, though; the high-resolution sensor will never be worse than the low-resolution one.

There's no reason you can't turn the shutter speed on the 33MP sensor to 1/1000 and get the same motion-freezing capability. The output at that shitter speed wouldn't be ideal for video, but, if you're aiming at capturing stills, there's no technical reason you couldn't do that.

The unlimited RAW burst capability has other implications for video, though. One of the difficulties with video, when shooting in a dark environment, is the ability to shoot at slower shutter speeds. Even after turning up the ISO to the limits of acceptable quality, it mightn't be enough. At 25fps, the slowest possible shutter speed is 1/25. But what if you need 1/10, 1/5 or even slower? With unlimited RAW burst, you can do that by using a buffer - still shoot at 25fps, but store the frames in a buffer, so that the frames can be added together (effectively giving a slower shutter speed) before being saved to storage. The merged frames would overlap - frame 1 might consist of frames 1-5 added together, frame 2 2-6, frame 3 3-7 and so on - but you'd get your 25fps while still being able to shoot at a shutter speed slower than 1/25.
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Sony A9 - SonyAlphaRumors (SR5) information
« Reply #41 on: April 12, 2016, 05:10:32 am »

If the shutter speed's the same, then the amount of motion blur will be the same.
Of course.

I was merely pointing out that there is more to good images than resolution, and a video camera with more spatial resolution than a stills camera might not be able to surpass that stills camera due to operational limits and/or sensor/electronics/encoding limits.

-h
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2794
Re: Sony A9 - SonyAlphaRumors (SR5) information
« Reply #42 on: April 12, 2016, 05:17:01 am »

What are the highest sustained read bandwidths of cutting edge memory cards? Say that Sony release an "A9" featuring the on-sensor processing and memory of the RX-100M4. They would still (eventually) have to dump the buffer to memory card.

https://en.wikipedia.org/wiki/CompactFlash#CFast
CFast seemed tp allow 600MB/s in 2011. That is:
600*1e6*8/(12e6*10) = 40fps@12MP 10bpp
600*1e6*8/(24e6*14) = 14.3fps@24MP 14bpp
600*1e6*8/(72e6*14) = 4.8fps@72MP 14bpp

There is of course the possibility of using a video codec for 50:1 or more lossy compression. Or using 2 or more memory cards interleaved. Or using a proprietary flash standard (Sony seems to be fond of that). If the camera included, say, 32 GB of built-in memory that could be used for this purpose, I guess that many would be satisfied.

-h

This uses XQD cards. These currently go to 350MBps write speed (read speed is 400MBps, but that's irrelevant for burst speed); the next series is supposed to go to 1000MBps.

An uncompressed, 14-bit RAW file will be 35MB in size at 20MP, 70MB at 40MP and 140MP at 80MP at minimum (assuming the two spare bits in each two-byte block aren't just 'wasted'). This equates to 10fps, 5ps and 2.5fps at 350MBps, or 28fps, 14fps and 7fps at 1000MBps. Most likely, there will be some sort of compression, bringing these speeds up slightly (for lossless compression) or significantly (for lossy compression). Likely, both compression modes will be available - one for quality, another for speed. Perhaps continuous-shooting will only be available with lossy compression, or with pixel binning to give a smaller output resolution; full-sized, losslessly-compressed RAWs would still have a limited burst, although the limit could be extremely large/functionally unlimited if the camera contained a large, fast buffer. But, either way, 8K video will likely require the faster, next-gen series of cards.
Logged

Dan Wells

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1040
Re: Sony A9 - SonyAlphaRumors (SR5) information
« Reply #43 on: April 13, 2016, 01:54:52 pm »

At some point, this begins to compete mostly with medium format cameras (which have no burst mode to speak of - the Pentax will go 3 fps for 10 frames, and that's by far the best on the MF market)... Of course, the D3x didn't have a meaningful burst mode at full quality (it was 5 fps, but went down to 1.5 when you turned on 14 bit raw!), either, and that's probably the best analog for this highly specialized camera.

Watch Fuji jump in to this space (oddly, as usual) with a camera that is no heavier than the Sony, but uses a 33x44 mm sensor, either the one in the Pentax 645z or more likely a successor...

There will soon be a blurring of the lines between frame sizes - full frame was more meaningful when anything EXCEPT full frame meant that lenses were cropped (and bigger than they really needed to be). While that's still largely true for Canon and Nikon (most DX and EF-S lenses are cheap kit zooms), other companies are making matched lens lineups for a variety of sensor sizes. Fuji has a superb lineup of APS-C lenses, and I fully expect to see them do the same thing at 33x44mm (they will be a little bigger than FF lenses, but they won't be anywhere near as big as 645 lenses). Just as they are competing with 24 MP full frame with the APS-C X-Pro 2 and its great lenses, I expect them to go after the highest end of FF and the lower end of medium format with a 33x44 mm camera (Photokina?)
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2794
Re: Sony A9 - SonyAlphaRumors (SR5) information
« Reply #44 on: April 14, 2016, 03:28:47 am »

At some point, this begins to compete mostly with medium format cameras (which have no burst mode to speak of - the Pentax will go 3 fps for 10 frames, and that's by far the best on the MF market)... Of course, the D3x didn't have a meaningful burst mode at full quality (it was 5 fps, but went down to 1.5 when you turned on 14 bit raw!), either, and that's probably the best analog for this highly specialized camera.

Watch Fuji jump in to this space (oddly, as usual) with a camera that is no heavier than the Sony, but uses a 33x44 mm sensor, either the one in the Pentax 645z or more likely a successor...

There will soon be a blurring of the lines between frame sizes - full frame was more meaningful when anything EXCEPT full frame meant that lenses were cropped (and bigger than they really needed to be). While that's still largely true for Canon and Nikon (most DX and EF-S lenses are cheap kit zooms), other companies are making matched lens lineups for a variety of sensor sizes. Fuji has a superb lineup of APS-C lenses, and I fully expect to see them do the same thing at 33x44mm (they will be a little bigger than FF lenses, but they won't be anywhere near as big as 645 lenses). Just as they are competing with 24 MP full frame with the APS-C X-Pro 2 and its great lenses, I expect them to go after the highest end of FF and the lower end of medium format with a 33x44 mm camera (Photokina?)

Definitely agree here.

Pro-level cameras tend to lean towards the high-speed action camp (D4s, 1Dx) or the low-speed-but-incredible-detail camp (MF, 5Ds, D3x, 1Ds3), with a few (5D3, D810) closer to the middle ground, but still leaning one way or the other. Sony is unlikely to match the high-speed action cameras this time (chiefly due to EVF speed), so is likely to target the hi-res camp.

Thing is, they could potentially do very, very well at it - even against medium format. Sure, it's a smaller sensor, but that just means the optics need to be more precise, which is easier to do for a smaller lens - Zeiss Otus lenses on full-frame are just as sharp as the best Rodenstock or Schneider lenses on MF. Light collecting area is smaller, but CMOS has better quantum efficiency than CCD, so the total number of photons collected is similar at the same ISO (obviously, the newer CMOS-based MF sensors are better in this regard), giving the same noise performance.

The key would be to increase photon well capacity and decrease minimum ISO to the same as that achievable with MF or lower, in order to collect a similar number of photons, minimising shot noise and maximising DR.
Logged
Pages: 1 2 [3]   Go Up