Pages: 1 ... 3 4 [5] 6   Go Down

Author Topic: A7r4 announcement today  (Read 21997 times)

D Fuller

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
    • AirStream Pictures
Re: A7r4 announcement today
« Reply #80 on: July 21, 2019, 08:15:09 am »

Mentioned sporadically on FredMiranda. Also seems to be mentioned a lot on Chinese-language websites (such as this article), accessible using Google Translate. Type in "A7r4" or "A7r IV" and "IMX551" into Google and you'll get a few hits from FredMiranda and other English websites, but a ton of articles in Chinese about it.

Also, it's the only 61MP sensor that Sony itself has announced manages close to 15 stops of DR (14.9 stops). The other 61MP sensor only manages 14.4 stops and also has a base ISO of 64, not 100 like the A7r4.

Does a sensor have a (non-theoretical) dynamic range in isolation from its surrounding electronics? Or is DR the product of the electronic system as a whole?
Logged
business website: www.airstream.pictures
blog: thirtynineframes.com/blog

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: A7r4 announcement today
« Reply #81 on: July 21, 2019, 11:19:31 am »

Does a sensor have a (non-theoretical) dynamic range in isolation from its surrounding electronics?

Since the sensor has column ADCs, it does indeed. That assumes perfect power supplies and grounding.

Jim

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: A7r4 announcement today
« Reply #82 on: July 21, 2019, 12:34:14 pm »

Regarding the dynamic range claim, maybe this news from 7 months ago reveals a bit more:

https://www.sonyalpharumors.com/sr5-more-specs-from-the-five-new-sony-full-frame-sensors/

The A7r4 uses the IMX 551 sensor, with a 14.9 stop DR. This sensor was not made available to other companies (at least not yet).

The less-capable IMX 455 was made available to third parties, and is the sensor Nikon has purchased for its own cameras. This will likely show up in the Z8. Note that the IMX 455 has less DR (14.4 stops), but uses a base ISO of 64 (with another base at 400) - the same setup as has appeared in recent Nikon bodies.

Hi,

I could be somewhat skeptical about that DR figure. They could of course talk about normalized DxO-mark style figures. But, handling 14.9 bits of DR would need a 15 bit data format. It would also mean slower read out. So, color me skeptical about that.

Best regards
Erik
Logged
Erik Kaffehr
 

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
14 bits!
« Reply #83 on: July 21, 2019, 12:44:53 pm »

I could be somewhat skeptical about that DR figure. They could of course talk about normalized DxO-mark style figures. But, handling 14.9 bits of DR would need a 15 bit data format. It would also mean slower read out. So, color me skeptical about that.

Sony says it's 14 bits:

Dan Wells

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1044
Re: A7r4 announcement today
« Reply #84 on: July 21, 2019, 02:04:41 pm »

The SonyAlphaRumors specs on the IMX455 mention that it has a 16-bit readout capability. Since the links on SAR are screwed up and the article purporting to lead to the IMX551 specs also goes to the 455 (same article), I can't see whether the 551 has the (probably slowed down) 16-bit mode.

One of the very first cameras to have a true 14 bit readout was the Nikon D3x, which was very slow in 14-bit mode. The (MUCH less expensive) Sony DSLR using a similar sensor released at almost exactly the same time (Alpha 900? It was a long time ago) did not enable the slow 14-bit mode, and ignored several other tricks Nikon used to improve image quality.

Could Nikon pull the same trick again? A camera with a very, very high quality mode that is also disproportionately slow in high quality mode? The D3x 14-bit mode was only a little over 1 fps, and I believe it may have disabled continuous shooting entirely (or could it have automatically reverted to 12-bit when you moved the drive mode dial off of S)?

Would make for an interesting Z8, D900 or both - a camera that does most or all of what the A7r IV can do from a sensor perspective, but also has a mode that offers extra image quality at the cost of speed. They can't charge like a D3x for it, because the GFX 100 is sitting not that far above that price point...

Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: A7r4 announcement today
« Reply #85 on: July 21, 2019, 02:54:20 pm »

The SonyAlphaRumors specs on the IMX455 mention that it has a 16-bit readout capability. Since the links on SAR are screwed up and the article purporting to lead to the IMX551 specs also goes to the 455 (same article), I can't see whether the 551 has the (probably slowed down) 16-bit mode.

One of the very first cameras to have a true 14 bit readout was the Nikon D3x, which was very slow in 14-bit mode. The (MUCH less expensive) Sony DSLR using a similar sensor released at almost exactly the same time (Alpha 900? It was a long time ago) did not enable the slow 14-bit mode, and ignored several other tricks Nikon used to improve image quality.

Could Nikon pull the same trick again? A camera with a very, very high quality mode that is also disproportionately slow in high quality mode? The D3x 14-bit mode was only a little over 1 fps, and I believe it may have disabled continuous shooting entirely (or could it have automatically reverted to 12-bit when you moved the drive mode dial off of S)?

Would make for an interesting Z8, D900 or both - a camera that does most or all of what the A7r IV can do from a sensor perspective, but also has a mode that offers extra image quality at the cost of speed. They can't charge like a D3x for it, because the GFX 100 is sitting not that far above that price point...

If the pixel design is the same as in the 3.76 um IMX 461 and IMX 411, 16-bit precision won't offer much advantage. The read noise is sufficient to provide adequate dither to a 14-bit ADC. At least that's what it looks like from Bill Claff's numbers, although there are a few inconsistencies. It's pretty close, though.

Jim


Dinarius

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1225
Re: A7r4 announcement today
« Reply #86 on: July 23, 2019, 05:32:57 am »

Unless it’s reinventing the wheel in terms of image quality, doesn’t 16 step pixel shift shooting strike anyone as more than a little unwieldy and impractical in 2019?

Hasselblad had this years ago, and it was outstanding, if you had the time and shooting environment to do it.

But, if you really need it now, you’re probably working for an institution that might stretch to the six-shot Hasselblad.

I just don’t see the point in this camera.

I’m still using my old 39Mp multi-shot (4 shot) Hasselblad. In effect, all that it does is remove the effect of the filter. (And having only 39Mp spread over such a huge sensor produces a file of wonderful DR and elasticity) The difference between single and 4 shot is marked.

I’d certainly be interested in seeing how the 4-shot option in the Sony compares with the added bonus of 20+ extra MPs.

But, 16 shot strikes me cumbersome and retrograde.

D.
« Last Edit: July 23, 2019, 07:03:24 am by Dinarius »
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: A7r4 announcement today
« Reply #87 on: July 23, 2019, 08:32:43 am »

Unless it’s reinventing the wheel in terms of image quality, doesn’t 16 step pixel shift shooting strike anyone as more than a little unwieldy and impractical in 2019?

Hasselblad had this years ago, and it was outstanding, if you had the time and shooting environment to do it.

But, if you really need it now, you’re probably working for an institution that might stretch to the six-shot Hasselblad.

I just don’t see the point in this camera.

I’m still using my old 39Mp multi-shot (4 shot) Hasselblad. In effect, all that it does is remove the effect of the filter. (And having only 39Mp spread over such a huge sensor produces a file of wonderful DR and elasticity) The difference between single and 4 shot is marked.

I’d certainly be interested in seeing how the 4-shot option in the Sony compares with the added bonus of 20+ extra MPs.

But, 16 shot strikes me cumbersome and retrograde.

D.

If you have a lens that's sharp enough, it would be great for product photography and art reproduction.

This mode will give you 240 million pixels with true RGB values for each - provided nothing in you scene moves at all. At present, no other camera will do this - you either get 4x the pixel count but still rely on Bayer interpolation, or you get true RGB values for each pixel, but no overall increase in spatial resolution.

But the requirement for a totally static scene pretty much restricts it to product photography or reproduction. Almost everything else has some sort of moving element in it that would kill much of the resolution gain and/or introduce artifacts. It's a niche application, but an application nonetheless.
Logged

Dinarius

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1225
Re: A7r4 announcement today
« Reply #88 on: July 23, 2019, 10:27:02 am »

Repro/copy work is what I use my 4-shot Hasselblad for and it’s superb - DR, colour, etc.

Hasselblad’s new(ish) 6-shot offering looks fantastic, if you trust the PR.

My point still stands; in any kind of pressured environment, where time is money, a 16-shot offering looks like regression to me. I await the (unbiased) reviews.

D.

Ps. On the point of good glass for repro/copy; the Sony 90mm macro has been recommended, albeit a bit on the long side. Any other suggestions?
Logged

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1527
    • Martin Kristiansen
Re: A7r4 announcement today
« Reply #89 on: July 23, 2019, 12:18:39 pm »

If you have a lens that's sharp enough, it would be great for product photography and art reproduction.

This mode will give you 240 million pixels with true RGB values for each - provided nothing in you scene moves at all. At present, no other camera will do this - you either get 4x the pixel count but still rely on Bayer interpolation, or you get true RGB values for each pixel, but no overall increase in spatial resolution.

But the requirement for a totally static scene pretty much restricts it to product photography or reproduction. Almost everything else has some sort of moving element in it that would kill much of the resolution gain and/or introduce artifacts. It's a niche application, but an application nonetheless.

Who on earth needs that many pixels for product photography? I shoot products day in and day out and 99% ends up on websites. Very few products are so well made than they can stand to show that much detail.
Logged
Commercial photography is 10% inspiration and 90% moving furniture around.

Dinarius

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1225
Re: A7r4 announcement today
« Reply #90 on: July 23, 2019, 02:00:25 pm »

Who on earth needs that many pixels for product photography? I shoot products day in and day out and 99% ends up on websites. Very few products are so well made than they can stand to show that much detail.

This is the kind of use to which shed loads of pixels can be put. (Though I think that this is a composite image.)

http://www.insidebruegel.net/#p/v=udroom&lan=en&a=1016&x=s:3_l:1_v1:1016,vis

But, I’d rather get there via a 6-shot Hasselblad than a 16-shot anything else.

D.
Logged

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1527
    • Martin Kristiansen
Re: A7r4 announcement today
« Reply #91 on: July 23, 2019, 02:25:34 pm »

This is the kind of use to which shed loads of pixels can be put. (Though I think that this is a composite image.)

http://www.insidebruegel.net/#p/v=udroom&lan=en&a=1016&x=s:3_l:1_v1:1016,vis

But, I’d rather get there via a 6-shot Hasselblad than a 16-shot anything else.

D.

I totally get the need for as many pixels as you can lay your hands on with that type of stuff. I used to shoot art work for Sotheby’s on 8x10 and even 11x14 on occasion. Wouldn’t call it product photography though.

When I got into digital there was Kodak instant Capture and Phase and Agfa scanning backs pretty much. I made the decision I wasn’t doing scanning backs and later made the decision I wasn’t doing multishot anything.

If I really needed the ultimate in pixels as your link demonstrates I would buy the Phase 150MP monster. Sixteen frames on a Sony? I don’t think so.
Logged
Commercial photography is 10% inspiration and 90% moving furniture around.

DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: A7r4 announcement today
« Reply #92 on: July 23, 2019, 08:19:26 pm »

And having only 39Mp spread over such a huge sensor produces a file of wonderful DR and elasticity
with consumer level non liquid nitrogen or something cooled CCD sensor ? may be ~2 stops less DR than consumer level CMOS in modern 50mp MF or 40-something FF  ;D ... but you can dream of course about that 16-bit color tales from those backs so popular on this very forum a while back
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13980
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: A7r4 announcement today
« Reply #93 on: July 24, 2019, 12:29:44 am »

Why bother bringing actual photographic needs in this discussion? :D

Wouldn’t it be faster to just admit that the a7rIV is superior in every possible way for every possible application?

Cheers,
Bernard

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1527
    • Martin Kristiansen
Re: A7r4 announcement today
« Reply #94 on: July 24, 2019, 01:41:05 am »

Why bother bringing actual photographic needs in this discussion? :D

Wouldn’t it be faster to just admit that the a7rIV is superior in every possible way for every possible application?

Cheers,
Bernard

Finally you are starting to make sense Bernard  :)

Speaking of needs. If the the wireless tethering is fast enough the iv will be close to perfect for my admittedly modest studio requirements. High enough res when I need it and APSC giving me a useful reach and perfect MP count for most of my work. As an example today I am photographing 50 power sockets. Tomorrow about 40 Polo bags. Unfortunately that’s not all I do. But it is true that cameras are becoming more versatile. I have 6 short videos to produce and will use the same camera I used for all the products to shoot that.
Logged
Commercial photography is 10% inspiration and 90% moving furniture around.

Shiftworker

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 71
Re: A7r4 announcement today
« Reply #95 on: July 24, 2019, 03:35:12 am »

61mp on a 36x24mm sensor means that the pixel size is roughly the same as an M43 20mp sensor. On my GH5 I can see diffraction softening setting in when I stop down past 5.6 so for critical work so on the IV you will need to keep below F8 to get the best out of this sensor. Given that many lenses used in applications like landscape where edge and corner sharpness as well as DOF are important you have to ask is everyone going to benefit from 61mp? My main commercial tool is an A7r3 used in conjunction with Canon Shift lenses ( 24mm TS-EII and 17mm) for architecture. I find that in order to optimise resolution across the frame I need to stop down past the diffraction limit to F11 so I have serious doubts as to how useful 61mp will be unless Canon updates these now quite old lenses (or Sony brings out a set of shift lenses). Having said that I've just pre ordered a IV.....
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: A7r4 announcement today
« Reply #97 on: July 24, 2019, 07:25:54 am »

Repro/copy work is what I use my 4-shot Hasselblad for and it’s superb - DR, colour, etc.

Hasselblad’s new(ish) 6-shot offering looks fantastic, if you trust the PR.

My point still stands; in any kind of pressured environment, where time is money, a 16-shot offering looks like regression to me. I await the (unbiased) reviews.

99% of the time spent shooting anything that would be amenable to any sort of pixel shift technique consists of setting up the shot - subject, backdrops, lights, cameras, tripods. Actually taking the shot takes almost no time - and, if multiple exposures are exposed, the camera does it all for you once the remote trigger is pressed. Whether it's a single exposure, four exposures or sixteen exposures, the time difference is minimal - seconds at most - compared to the many minutes spent setting up the shot. And, once the shots are taken, the computer does all the work combining them, with minimal human input.

A sixteen-shot technique has a significant advantage over current four-shot techniques. By getting a true RGB reading for every pixel, without relying on interpolation, colour artifacts are almost completely eliminated. At 240MP - likely outresolving the lens - moire is also almost completely eliminated. With a four-shot technique that uses full pixel shifts, you eliminate the colour artifacts but don't get the resolution increase. With a four-shot technique that uses half pixel shifts, you get the increased resolution, but still get the colour artifacts. In most subjects, the artifacts will be subtle, but, if you're going to the effort of pixel shifting anyway, you're probably after every little bit you can get.

I could see this technique being taken to the extreme for certain macro subjects (e.g. dead bugs or flowers), in combination with focus stacking and averaged exposure (HDR being unnecessary with controlled lighting). Sixteen exposures, repeated eight times and averaged for an effective ISO of 12.5, focus stacked perhaps 10 times to get the whole object in focus, for a total of 1280 frames (or more, if more stacked layers are required). Probably overkill, but someone's bound to try it, if only to get a tack-sharp, flawless, everything-in-focus, billboard-sized print of an ant.
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: A7r4 announcement today
« Reply #98 on: July 24, 2019, 10:06:05 am »

99% of the time spent shooting anything that would be amenable to any sort of pixel shift technique consists of setting up the shot - subject, backdrops, lights, cameras, tripods. Actually taking the shot takes almost no time - and, if multiple exposures are exposed, the camera does it all for you once the remote trigger is pressed. Whether it's a single exposure, four exposures or sixteen exposures, the time difference is minimal - seconds at most - compared to the many minutes spent setting up the shot. And, once the shots are taken, the computer does all the work combining them, with minimal human input.

A sixteen-shot technique has a significant advantage over current four-shot techniques. By getting a true RGB reading for every pixel, without relying on interpolation, colour artifacts are almost completely eliminated. At 240MP - likely outresolving the lens - moire is also almost completely eliminated. With a four-shot technique that uses full pixel shifts, you eliminate the colour artifacts but don't get the resolution increase. With a four-shot technique that uses half pixel shifts, you get the increased resolution, but still get the colour artifacts. In most subjects, the artifacts will be subtle, but, if you're going to the effort of pixel shifting anyway, you're probably after every little bit you can get.

I could see this technique being taken to the extreme for certain macro subjects (e.g. dead bugs or flowers), in combination with focus stacking and averaged exposure (HDR being unnecessary with controlled lighting). Sixteen exposures, repeated eight times and averaged for an effective ISO of 12.5, focus stacked perhaps 10 times to get the whole object in focus, for a total of 1280 frames (or more, if more stacked layers are required). Probably overkill, but someone's bound to try it, if only to get a tack-sharp, flawless, everything-in-focus, billboard-sized print of an ant.

True enough. The a7RIV seems to lack focus bracketing, which seems like an oversight in this context.

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2013
    • http://www.guillermoluijk.com
Re: A7r4 announcement today
« Reply #99 on: July 24, 2019, 12:20:01 pm »

https://m.dpreview.com/news/4477765388/sony-shows-off-specs-of-six-full-frame-sensors-including-a-quad-bayer-sensor

Plenty of opportunities.

Cheers,
Bernard

Interesting Quad-Bayer structure sensor:


Photodetectors can be read:
  • With different exposure time (HDR): could produce artifacts with moving subjects
  • With different conversion gains (ISO) at the same exposure time: no moving artifacts but SNR enhancement will not be that high with ISO-invariant cells
  • Regular Bayer readout by binning each 4-photodetectors in a single pixel
Reminds me of the Fuji's Super CCD philosophy.

Regards

Pages: 1 ... 3 4 [5] 6   Go Up