Home · Register · Join Upload & Sell

Moderated by: Fred Miranda
Username  

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Sony Forum | Join Upload & Sell

1
       2       end
  

Speculating on the Sony a7RIV photographic dynamic range

  
 
Fred Miranda
Offline
Admin
Upload & Sell: On
p.1 #1 · p.1 #1 · Speculating on the Sony a7RIV photographic dynamic range


Speculating on the Sony a7RIV photographic dynamic range
@JimKasson, at blog.kasson.com, just posted an interesting article speculating on the A7R IV's PDR (photographic dynamic range)

From Jim:
"The wording of the announcement has some people saying that the new camera will have far greater photographic dynamic range than the previous version.
Time will tell, but I donít think so. The reason is that Sony has introduced a line of sensors with the same 3.76 micrometer (um) pixel pitch in three different sizes, and the pixel designs of the two versions that are currently shipping (or maybe, in the case of the GFX 100, dribbling would be a better word) appear to be very similar, if not identical.
...So, unless Sony has a trick up their corporate sleeve, the photographic dynamic range of the a7RIV is going to be about the same as the a7RIII."

Read the entire article


___________________

Pre-orders now available for the new Sony A7R IV:

B&H Photo | Adorama | Amazon ($3,498)



Aug 01, 2019 at 09:06 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
p.1 #2 · p.1 #2 · Speculating on the Sony a7RIV photographic dynamic range


When the A7R III was announced, Sony also claimed it delivered up to 15-stops of DR.
The only difference from the announcements is that the A7R IV's 15-stop DR is "enhanced". (??)

I'm not expecting the A7R IV to have higher DR and Jim's article is pretty convincing but we never know. Just matching the A7R III's dynamic range is already an accomplishment at 61MP.





From Sony's A7R III specs




Aug 01, 2019 at 10:31 PM
ytwong
Offline
• • • •
Upload & Sell: Off
p.1 #3 · p.1 #3 · Speculating on the Sony a7RIV photographic dynamic range


So, the 15 stops DR in A7R3 is crippled ?


Aug 01, 2019 at 10:44 PM
DavidBM
Offline
• • • • •
Upload & Sell: On
p.1 #4 · p.1 #4 · Speculating on the Sony a7RIV photographic dynamic range


Fred Miranda wrote:
When the A7R III was announced, Sony also claimed it delivered up to 15-stops of DR.
The only difference from the announcements is that the A7R IV's 15-stop DR is "enhanced". (??)

I'm not expecting the A7R IV to have higher DR and Jim's article is pretty convincing but we never know. Just matching the A7R III's dynamic range is already an accomplishment at 61MP.


Yes; I commented on Jim's post that the only claim about DR which suggested an improvement is "enhanced"

Enhanced may just mean that the tech has improved so that we can have more pixels in the same area for the same dynamic range, despite the wiring and other costs that you would expect to reduce DR.

So, if I was in a fanboy mood, I could accept that keeping the DR the same at the image level and having almost 50% more pixels counts as enhanced...



Aug 01, 2019 at 10:45 PM
ytwong
Offline
• • • •
Upload & Sell: Off
p.1 #5 · p.1 #5 · Speculating on the Sony a7RIV photographic dynamic range


Since I'm still using A7R2, I think it should be an improvement whether it is enhanced or not.

They are talking about BIONZ & LSI, Maybe the enhancement applies to jpg only??

Edited on Aug 01, 2019 at 11:22 PM · View previous versions



Aug 01, 2019 at 11:12 PM
dclark
Online
• • • • •
Upload & Sell: On
p.1 #6 · p.1 #6 · Speculating on the Sony a7RIV photographic dynamic range


We need to be a bit cautious since DR and PDR are not the same thing. PDR was defined by Claff and involves some pixel averaging. DR is not so well defined and can be applied to a number of measures of sensor and pixel performance.

If Sony is doing some image processing to enhance "dynamic range", they may have some other definition of what they mean by the term. Since they specify "stills", they may be referring to the 16 frame multi-frame mode of the camera, although why they would highlight the BIONZ for that is not clear since processing of the frames is done in post. I hope Sony is more clear about this soon.

Kasson is careful to use Claff's definition of PDR.

Dave



Aug 01, 2019 at 11:17 PM
pdmphoto
Offline
• • • •
Upload & Sell: Off
p.1 #7 · p.1 #7 · Speculating on the Sony a7RIV photographic dynamic range


Just more Sony marketing hype. They have done it with every A7x that they have announced. Weather resistant, water resistant, 15 stops DR, ... you name it...

They never qualify their claims and are often contradictory depending on the Sony source.



Edited on Aug 01, 2019 at 11:32 PM · View previous versions



Aug 01, 2019 at 11:30 PM
AmbientMike
Offline
• • • • •
Upload & Sell: Off
p.1 #8 · p.1 #8 · Speculating on the Sony a7RIV photographic dynamic range




Fred Miranda wrote:
From Jim:
"The wording of the announcement has some people saying that the new camera will have far greater photographic dynamic range than the previous version.
Time will tell, but I donít think so. The reason is that Sony has introduced a line of sensors with the same 3.76 micrometer (um) pixel pitch in three different sizes, and the pixel designs of the two versions that are currently shipping (or maybe, in the case of the GFX 100, dribbling would be a better word) appear to be very similar, if not identical.
...So, unless Sony has a trick up their corporate sleeve,
...Show more

Interesting, this is pretty much what I was thinking. Seems unlikely that Sony was able to sell a sensor that had inferior pixels to Fuji to put in their $10K camera. With similar if not identical pixel density, and FF so close to 33x44 to begin with.



Aug 01, 2019 at 11:30 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
p.1 #9 · p.1 #9 · Speculating on the Sony a7RIV photographic dynamic range


dclark wrote:
We need to be a bit cautious since DR and PDR are not the same thing. PDR was defined by Claff and involves some pixel averaging. DR is not so well defined and can be applied to a number of measures of sensor and pixel performance.

If Sony is doing some image processing to enhance "dynamic range", they may have some other definition of what they mean by the term. Since they specify "stills", they may be referring to the 16 frame multi-frame mode of the camera, although why they would highlight the BIONZ for that is not clear since processing
...Show more

I could be wrong but I think that averaging images (with or without pixel shift) always improve DR and SNR.
At least I see it in my images.



Aug 01, 2019 at 11:47 PM
dclark
Online
• • • • •
Upload & Sell: On
p.1 #10 · p.1 #10 · Speculating on the Sony a7RIV photographic dynamic range


Fred Miranda wrote:
I could be wrong but I think that averaging images (with or without pixel shift) always improve DR and SNR.
At least I see it in my images.


Yes, I agree.

I don't know if that is the source of enhanced dynamic range that Sony is claiming. I doubt it, but since they have provided no specific information on what they mean by enhanced dynamic range and how it is achieved, we are left to speculate.



Aug 02, 2019 at 12:03 AM
 


Search in Used Dept. 

snapsy
Offline
• • • • • •
Upload & Sell: On
p.1 #11 · p.1 #11 · Speculating on the Sony a7RIV photographic dynamic range


Fred Miranda wrote:
When the A7R III was announced, Sony also claimed it delivered up to 15-stops of DR.
The only difference from the announcements is that the A7R IV's 15-stop DR is "enhanced". (??)

I'm not expecting the A7R IV to have higher DR and Jim's article is pretty convincing but we never know. Just matching the A7R III's dynamic range is already an accomplishment at 61MP.


Sony's 15-stop DR claim for the A7rIV is based on an 8MP downsampled image, per Dpreview's Richard Bulter. This would be just 0.3EV higher than the A7rIII, which has 14.7EV of DR downsampled @ 8MP.



Aug 02, 2019 at 12:17 AM
JimKasson
Offline
• • •
Upload & Sell: On
p.1 #12 · p.1 #12 · Speculating on the Sony a7RIV photographic dynamic range


dclark wrote:
Kasson is careful to use Claff's definition of PDR.


The black horizontal line is Bill's definition. But you can draw another horizontal line wherever you wish if you prefer another definition. Move it up 1 unit, and you'll have the threshold for a 16-inch-high print viewed at the same distance.




Aug 02, 2019 at 09:25 AM
Steve Spencer
Offline
• • • • • •
Upload & Sell: On
p.1 #13 · p.1 #13 · Speculating on the Sony a7RIV photographic dynamic range


AmbientMike wrote:
Interesting, this is pretty much what I was thinking. Seems unlikely that Sony was able to sell a sensor that had inferior pixels to Fuji to put in their $10K camera. With similar if not identical pixel density, and FF so close to 33x44 to begin with.


I think Jim proposed the exact opposite in this article. Why would you think the pixels are worse on the 100MP 44 X 33 sensor? And FF sensors are not all that close to 44 X 33 to begin with. If you crop to 4 X 3 or squarer 44 X 33 sensors are almost twice as big. Even if you crop to 3 X 2 or a skinnier rectangle, the 44 X 33 sensor is 50% bigger. Saying they are almost the same is like saying an f/1.4 lens has almost the same aperture as an f/2 lens (if you do a 4 X 3 crop or squarer) or that an f/1.4 lens has almost the same aperture as an f/1.7 lens (if you crop to 3 X 2 or a skinnier rectangle).



Aug 02, 2019 at 09:52 AM
AmbientMike
Offline
• • • • •
Upload & Sell: Off
p.1 #14 · p.1 #14 · Speculating on the Sony a7RIV photographic dynamic range




Steve Spencer wrote:
I think Jim proposed the exact opposite in this article. Why would you think the pixels are worse on the 100MP 44 X 33 sensor? And FF sensors are not all that close to 44 X 33 to begin with. If you crop to 4 X 3 or squarer 44 X 33 sensors are almost twice as big. Even if you crop to 3 X 2 or a skinnier rectangle, the 44 X 33 sensor is 50% bigger. Saying they are almost the same is like saying an f/1.4 lens has almost the same aperture as an f/2 lens
...Show more

Interesting post, since I was agreeing with the article. I think I tend to get suspicious of people who talk about sensor area. It looks more sensational that way.



Aug 02, 2019 at 12:27 PM
nhsonyshooter
Offline
• • •
Upload & Sell: On
p.1 #15 · p.1 #15 · Speculating on the Sony a7RIV photographic dynamic range


pdmphoto wrote:
Just more Sony marketing hype. They have done it with every A7x that they have announced. Weather resistant, water resistant, 15 stops DR, ... you name it...

They never qualify their claims and are often contradictory depending on the Sony source.



Which makes them just like everyone else



Aug 02, 2019 at 12:36 PM
dclark
Online
• • • • •
Upload & Sell: On
p.1 #16 · p.1 #16 · Speculating on the Sony a7RIV photographic dynamic range


JimKasson wrote:
The black horizontal line is Bill's definition. But you can draw another horizontal line wherever you wish if you prefer another definition. Move it up 1 unit, and you'll have the threshold for a 16-inch-high print viewed at the same distance.



Thanks for responding. That is the same answer you gave to my post on your site.

I have not done a careful review of all your information or the information Bill Claff has provided, but I believe that Claff's original definition of PDR involved "normalizing" the "photon transfer curve" by using the COC for an 8" tall print held at arms length. This is essentially the same as averaging the number of sensor pixels that fit within the COC so that sensor and pixel area are normalized so that the dynamic range of cameras with different sensors could be compared. He adopted a threshold SNR of 20 for the definition of PDR, and then adjusted that based on the COC to draw the "black horizontal line" at a value less that SNR=20. When I look at the D300 example he gave, he used a COC of 22 microns. That can be translated to about 2MP image sampling. I believe somewhere along the line that may have changed to 8" tall at 18" viewing and a COC of ~10-11 microns for full frame sensors, which is ~8MP image sampling, and the threshold SNR = 10, which is what you are using. I also note that you normalize the vertical axis rather than shift the threshold SNR.

I admit I have not done full due diligence so maybe this is covered in some of your posts or Claff's posts. If so, maybe you can direct me to them.

The point of my original post was to point out that PDR can change depending on how you define the image area over which there is pixel averaging, and that PDR is distinct from other uses of the term dynamic range.

BTW, it's also worth noting that the pixel averaging is why the PDR can exceed the 14 bits to which the pixel data is quantized. There have been some posts stating that Sony is clearly wrong in their claims since it is not possible to exceed 14 bits with data that is quantized to 14 bits.

Dave



Aug 02, 2019 at 01:04 PM
Steve Spencer
Offline
• • • • • •
Upload & Sell: On
p.1 #17 · p.1 #17 · Speculating on the Sony a7RIV photographic dynamic range


AmbientMike wrote:
Interesting post, since I was agreeing with the article. I think I tend to get suspicious of people who talk about sensor area. It looks more sensational that way.



Yeah, I think I was misunderstanding your post as I did once before. Sorry about that. I think we agree the pixels are very similar on the Sony A7r IV and the GFX 100 and may just disagree whether the FF sensor is close in size to the 33 X 44 sensor.


Edited on Aug 02, 2019 at 02:16 PM · View previous versions



Aug 02, 2019 at 02:14 PM
cgarcia
Offline
• •
Upload & Sell: Off
p.1 #18 · p.1 #18 · Speculating on the Sony a7RIV photographic dynamic range


dclark wrote:
Thanks for responding. That is the same answer you gave to my post on your site.

I have not done a careful review of all your information or the information Bill Claff has provided, but I believe that Claff's original definition of PDR involved "normalizing" the "photon transfer curve" by using the COC for an 8" tall print held at arms length. This is essentially the same as averaging the number of sensor pixels that fit within the COC so that sensor and pixel area are normalized so that the dynamic range of cameras with different sensors could be compared. He
...Show more

The way the "photographic" dynamic range is calculated causes it to exactly match the "engineering" dynamic range (screen level) for any sensor of 384 megapixels (exactly). Sensors with less than 384 MP may have a photographic dynamic range bellow the engineering, and sensors beyond that amount of pixels will have a "photographic" dynamic range above the engineering one. All of this without changing the definition of engineering DR.

My personal take is that maybe we all are getting a bit silly about this (trying to measure with exact numbers subjective things). I think that the dynamic range (engineering, without any further "cooking") is by itself the useful measure to predict the shadows recovery ability, once normalized to the same amount of pixels (e.g. 8MP, as DXO has "standarized") because we are interested in the full picture. That can be calculated by simply measuring the read noise (for short exposures all other sources are mostly irrelevant) with a single shot with the lens cap on.

The other essential sensor parameter is the signal-to-noise ratio (SNR) which will measure the "midtones performance" (the signal depends on the sensor efficiency, and most of the noise there comes from the physics laws). This is why a FF sensor is better than an APS-C: for the same final output size, it has 2.3x the area (with the same technology it gets 2.3 times more light and much better SNR). This is not so easy to be calculated, but can still be done by yourself with relatively low effort (yes, we can measure at home, with good accuracy, how many photons a pixel from our sensor may capture before getting overexposed).

The "photographic" dynamic range tries to combine both measures by moving the "the acceptable noise" threshold depending on the SNR at low signal levels. Theoretically is a more "real" DR. It is still subjective, and it is not too easy to be computed. I have some clue but don't know the exact maths behind. And since there is mostly only one Internet site (excelent, but a single one) doing the maths and the measures, we have no easy way to apply criticism to its alleged values. Personally I don't care about how real is the DR measure. For me (the "engineering" DR normalized) it is a mere shadows recovery indicator comparable across different sensors, which everybody can replicate. Maybe if trying to compare digital DR to film DR I should be more picky...



Aug 02, 2019 at 02:34 PM
JimKasson
Offline
• • •
Upload & Sell: On
p.1 #19 · p.1 #19 · Speculating on the Sony a7RIV photographic dynamic range


dclark wrote:
Thanks for responding. That is the same answer you gave to my post on your site.

I have not done a careful review of all your information or the information Bill Claff has provided, but I believe that Claff's original definition of PDR involved "normalizing" the "photon transfer curve" by using the COC for an 8" tall print held at arms length. This is essentially the same as averaging the number of sensor pixels that fit within the COC so that sensor and pixel area are normalized so that the dynamic range of cameras with different sensors could be compared. He
...Show more

I use the same normalization that Bill does. The normalization factor for the horizontal line on my chart is for a threshold SNR of 10 for 1600-pixel high sensor. To get the threshold SNR for a sensor of a give height in pixels -- call that h --, the formula is 16000/h.

dclark wrote:
I admit I have not done full due diligence so maybe this is covered in some of your posts or Claff's posts. If so, maybe you can direct me to them.


This is old, but I think it's still accurate:

https://blog.kasson.com/the-last-word/isolessness-comparisons-across-resolutions/

https://blog.kasson.com/the-last-word/photographic-dynamic-range-comparisons/



dclark wrote:
The point of my original post was to point out that PDR can change depending on how you define the image area over which there is pixel averaging, and that PDR is distinct from other uses of the term dynamic range.

BTW, it's also worth noting that the pixel averaging is why the PDR can exceed the 14 bits to which the pixel data is quantized. There have been some posts stating that Sony is clearly wrong in their claims since it is not possible to exceed 14 bits with data that is quantized to 14 bits.

Dave


All true, although my objection to normalizing DR at an SNR of 0 remains.

Jim




Aug 02, 2019 at 02:58 PM
JimKasson
Offline
• • •
Upload & Sell: On
p.1 #20 · p.1 #20 · Speculating on the Sony a7RIV photographic dynamic range


cgarcia wrote:
The way the "photographic" dynamic range is calculated causes it to exactly match the "engineering" dynamic range (screen level) for any sensor of 384 megapixels (exactly). Sensors with less than 384 MP may have a photographic dynamic range bellow the engineering, and sensors beyond that amount of pixels will have a "photographic" dynamic range above the engineering one. All of this without changing the definition of engineering DR.


That is true if you use a threshold SNR for EDR of unity. Not everyone does. I don't, for example. I compute EDR as full scale over read noise, which is more optimistic.

cgarcia wrote:
My personal take is that maybe we all are getting a bit silly about this (trying to measure with exact numbers subjective things). I think that the dynamic range (engineering, without any further "cooking") is by itself the useful measure to predict the shadows recovery ability, once normalized to the same amount of pixels (e.g. 8MP, as DXO has "standarized") because we are interested in the full picture. That can be calculated by simply measuring the read noise (for short exposures all other sources are mostly irrelevant) with a single shot with the lens cap on.

The other essential sensor parameter
...Show more

My normalized shadow SNR curves allow you to pick the SNR that you consider valid for your application.





Aug 02, 2019 at 03:03 PM
1
       2       end






FM Forums | Sony Forum | Join Upload & Sell

1
       2       end
    
 

You are not logged in. Login or Register

Username      Reset password