Photographic Dynamic Range (PDR) is a More Useful Measure than the DxOMark Landscape Score

Started Apr 17, 2015 | Discussions
OP bclaff Veteran Member • Posts: 8,735
Re: Chart doesn't load

Jack Hogan wrote:

bclaff wrote:

Luke Kaven wrote:

On my machine (Vista/Firefox 37.0.1), the chart doesn't load at all now. I just get a blank canvas in the chart region. Passing it along in case it alerts you to something you didn't already know.

Sent you email to see if we can figure it out offline.
If anyone else has chart difficulties please Private Message or email (preferred).

Try erasing the 's' in https. That is in the address bar, instead of

https://home.comcast.net/~NikonD70/Charts/DXOPDR.htm

try this

http://home.comcast.net/~NikonD70/Charts/DXOPDR.htm

If I do that it works for me.

Jack

Arg ! I usually make that change when I post a link and apologize if I forgot on that one.

Thanks Jack!

-- hide signature --
OP bclaff Veteran Member • Posts: 8,735
Revised Link to DxOMark Photographic Dynamic Range Chart

The link in my original post was for https protocol rather than http

Here's a better link:

DxOMark Photographic Dynamic Range Chart

-- hide signature --
Horshack Veteran Member • Posts: 7,993
Re: Regarding PDR of Sony A7S at ISO 409600
3

bclaff wrote:

Horshack wrote:

Horshack wrote:

bclaff wrote:

Horshack wrote:

bclaff wrote:

However, normalized read noise is not photographically relevant; and the DxOMark Landscape Use Case score is supposed to be a photographic use case.

Read noise is not relevant to whose photography specifically?

I think I could have been more explicit.
I mean that a measure that is solely read noise and has no photon noise component is not photographically relevant. (It has engineering relevance.)
If we disagree on that, I guess we should just let it go!

The problem with randomly selecting an SNR cutoff for DR and calling it relevant is that it's only relevant to some and not others.

I don't believe the PDR cutoff is random at all.
It is based on the acuity of the human eye and established values for image quality.

Regards,

I've got to run out but this is an interesting discussion to continue later. Until then here are two A7s images for thought - at their full native resolution - one at ISO 100 and the other at 409600 - at the same relative exposure. According to DxO the A7s has 5.76DR @ ISO 409,600 (normalized to 8MP) vs your PDR measurement of 0.93. Looking at the ISO 409,600 image do you believe we cannot discern > 1EV of DR in it?

A7s ISO 100 4 seconds

A7s ISO 409,600 1/1000

Here's perhaps a better ISO 409,600 example, with the scene DR more visible. raw is temporarily available here.

I assume you're trying to show that the Sony A7S has more DR than the 0.93 that PDR predicts at 409600.
But I think you're forgetting that there is an acceptable quality factor built in.
When I put your image on my screen at 8"x12" and view it at a reasonable distance I don't find it acceptable (usable perhaps, but not acceptable).
PDR predicts that the brightest 0.93 stop in such an image if ETTRed would be acceptable. (Maybe up by those stairs)
I have some other input on this matter but since it's really not directly on the topic of this thread I'll start another when I get my act together.

Regards,

You proposed two foundations for your 20:1 SNR cutoff - human sensory limitation and acceptable quality. I believe this image demonstrates we can perceive more than the 1EV in this image. As for what is acceptable quality that'll always be up for debate.

Luke Kaven Veteran Member • Posts: 5,715
Re: Chart doesn't load

bclaff wrote:

Jack Hogan wrote:

bclaff wrote:

Luke Kaven wrote:

On my machine (Vista/Firefox 37.0.1), the chart doesn't load at all now. I just get a blank canvas in the chart region. Passing it along in case it alerts you to something you didn't already know.

Sent you email to see if we can figure it out offline.
If anyone else has chart difficulties please Private Message or email (preferred).

Try erasing the 's' in https. That is in the address bar, instead of

https://home.comcast.net/~NikonD70/Charts/DXOPDR.htm

try this

http://home.comcast.net/~NikonD70/Charts/DXOPDR.htm

If I do that it works for me.

Jack

Arg ! I usually make that change when I post a link and apologize if I forgot on that one.

Thanks Jack!

That fixed it for me.  Now we're cooking.

OP bclaff Veteran Member • Posts: 8,735
Re: Regarding PDR of Sony A7S at ISO 409600
1

Horshack wrote:

bclaff wrote:

I assume you're trying to show that the Sony A7S has more DR than the 0.93 that PDR predicts at 409600.

But I think you're forgetting that there is an acceptable quality factor built in.
When I put your image on my screen at 8"x12" and view it at a reasonable distance I don't find it acceptable (usable perhaps, but not acceptable).
PDR predicts that the brightest 0.93 stop in such an image if ETTRed would be acceptable. (Maybe up by those stairs)
I have some other input on this matter but since it's really not directly on the topic of this thread I'll start another when I get my act together.

You proposed two foundations for your 20:1 SNR cutoff - human sensory limitation and acceptable quality. I believe this image demonstrates we can perceive more than the 1EV in this image. As for what is acceptable quality that'll always be up for debate.

No disagreement.

Just to recap.

I chose SNR = 20 which ISO says is "good" (They also say SNR = 40 is "excellent" )
ISO also says SNR = 10 is "acceptable"
As I understand it, SNR = 5 is considered the limit of usable feature detection.

So certainly, depending on what you are after, you could add stops to the PDR value to suit your particular purpose.

Regards,

-- hide signature --
Detail Man
Detail Man Forum Pro • Posts: 17,078
Re: Regarding PDR of Sony A7S at ISO 409600

bclaff wrote:

Horshack wrote:

bclaff wrote:

I assume you're trying to show that the Sony A7S has more DR than the 0.93 that PDR predicts at 409600.

But I think you're forgetting that there is an acceptable quality factor built in.
When I put your image on my screen at 8"x12" and view it at a reasonable distance I don't find it acceptable (usable perhaps, but not acceptable).
PDR predicts that the brightest 0.93 stop in such an image if ETTRed would be acceptable. (Maybe up by those stairs)
I have some other input on this matter but since it's really not directly on the topic of this thread I'll start another when I get my act together.

You proposed two foundations for your 20:1 SNR cutoff - human sensory limitation and acceptable quality. I believe this image demonstrates we can perceive more than the 1EV in this image. As for what is acceptable quality that'll always be up for debate.

No disagreement.

Just to recap.

I chose SNR = 20 which ISO says is "good" (They also say SNR = 40 is "excellent" )
ISO also says SNR = 10 is "acceptable"
As I understand it, SNR = 5 is considered the limit of usable feature detection.

Just out of curiosity, are you saying that the "International Standards Organization" ("ISO") has deemed this or that image SNR as "acceptable" ?

So certainly, depending on what you are after, you could add stops to the PDR value to suit your particular purpose.

The image is indeed "a mess". So much digital scaling is taking place that non-zero valued RawDigger histogram bins represent only 25% of the total (way down in the "boondocks" below)":

That said, landscape photography (shot in relatively low, indirect natural lighting) can indeed include shadow areas approaching/ including Read/Dark noise levels. I don't really "get" the point of a debate as to whether or why such occurances would or would not be "acceptable". They just "are" ... and in processing one may want/need to deal with them.

The "debate" seems to be whether some DR/SNR "metric" should "bother" with such "depths".

DM

OP bclaff Veteran Member • Posts: 8,735
Re: Regarding PDR of Sony A7S at ISO 409600

Detail Man wrote:

bclaff wrote:


Just to recap.

I chose SNR = 20 which ISO says is "good" (They also say SNR = 40 is "excellent" )
ISO also says SNR = 10 is "acceptable"
As I understand it, SNR = 5 is considered the limit of usable feature detection.

Just out of curiosity, are you saying that the "International Standards Organization" ("ISO") has deemed this or that image SNR as "acceptable" ?

Yes, ISO 12232

Regards,

-- hide signature --
Detail Man
Detail Man Forum Pro • Posts: 17,078
Re: Deemed (by committee) "Acceptable" Image SNRs ...
1

bclaff wrote:

Detail Man wrote:

bclaff wrote:

Just to recap.

I chose SNR = 20 which ISO says is "good" (They also say SNR = 40 is "excellent" )
ISO also says SNR = 10 is "acceptable"
As I understand it, SNR = 5 is considered the limit of usable feature detection.

Just out of curiosity, are you saying that the "International Standards Organization" ("ISO") has deemed this or that image SNR as "acceptable" ?

Yes, ISO 12232.

OK. Found some of the relevant quotes from that standard. So, you picked a (linear) SNR=20. It may make sense to separate analysis of results that are dominantly influenced by Photon Shot Noise (as opposed to Read/Dark Noise), at least at base-ISO - from the dark-frame Read/Dark Noise analysis (as you do). What you are doing is a valuable thing. Keep up the interesting, thought provoking work !

DM ...

Jack Hogan Veteran Member • Posts: 7,026
Black Level Without Optical Black Pixels

Jack Hogan wrote: Bill, I get very similar numbers to yours, within a few hundreds of a stop. The D7100 Dxomark Landscape score looks off, it should be much closer to the D7200's, which looks right. Say within 1/3 of a stop.

Here are the results I get from the files that you kindly provided. One question that comes to mind has to do with how to accurately determine Black Level when optical black pixels are not present. Those were quite useful because they allowed to compensate for capture environment biases such as temperature, dark current etc. How do raw converters estimate Black Level now? It can make quite a difference to read noise (not of course to fwc or gain). Here is data from your D750 captures for instance, assuming that shutter speed is fairly accurate it seems that it's 1.36 DN off from the standard 600 used by Raw Digger for instance. That's a lot:

The D750's green channels are well behaved as one can see. The same cannot be said for the B and R ones at base ISO, the subject of an older Jim Kasson thread . If one uses a Black Level of 600.0 instead of 601.36 read noise drops to 3.84e- from 4.55e-. I think the latter makes more sense. Here is the D610 which does have OB pixels and subtracts them before writing data to the raw file.

The D610's data was not of the highest quality overall, but the Black Level here is within about 1/8th of a DN, good enough for most applications I think. Here is the complete ISO 100 set, using average Black Level estimated from the two green channels only:

So how do we (and raw converters) deal with black level when there are no OB pixels?

Jack

OP bclaff Veteran Member • Posts: 8,735
Re: Black Level Without Optical Black Pixels

That's great work Jack.
It's always good to get another set of eyes on something.
And any raw files I have or can produce are always available to people such as yourself.

Black Level can be a problem. It certainly is helpful when Optiocal Black (OB) is available.

For my PDR measurements I only use whatever values are in the metadata (Exif).
Since I operate further away from read noise than DxOMark I suspect PDR is less affected by variations between the metadata and true levels.

(I use entirely different data to estimate read noise as opposed to PDR.)

Regards,

-- hide signature --
Victor Engel Forum Pro • Posts: 18,764
Re: Revised Link to DxOMark Photographic Dynamic Range Chart

bclaff wrote:

The link in my original post was for https protocol rather than http

Here's a better link:

DxOMark Photographic Dynamic Range Chart

Has the link changed again?

-- hide signature --

Victor Engel

 Victor Engel's gear list:Victor Engel's gear list
Canon EOS 5D Canon EOS 5D Mark II Canon EOS 600D Canon EOS 5DS Canon EF 50mm F1.8 II +12 more
OP bclaff Veteran Member • Posts: 8,735
Re: Revised Link to DxOMark Photographic Dynamic Range Chart
2

Victor Engel wrote:

bclaff wrote:

The link in my original post was for https protocol rather than http

Here's a better link:

DxOMark Photographic Dynamic Range Chart

Has the link changed again?

Yes, quite right, moved everything to PhotonsToPhotos later in 2015

DxoMark Photographic Dynamic Range

-- hide signature --

Bill ( Your trusted source for independent sensor data at http://www.PhotonsToPhotos.net )

PaulSnowcat
PaulSnowcat Senior Member • Posts: 1,020
I have a question...

I'VE seen that some sources have different PDR measured for the SAME camra when it's working in FF and in APS-C modes. For example, PDR or Sony A7RII in FF mode is bigger then it's PDR in crop mode.

How this can be true? Dynamic Range is a difference between the most light and the most dark pixel on the sensor that still have color information in them. How this parameter can change depending on which part of the same sensor you are talking about? It should be exaclty the same... Please explain this thing to me...

-- hide signature --
 PaulSnowcat's gear list:PaulSnowcat's gear list
Nikon D7500 Tokina AT-X Pro 12-24mm f/4 DX II Sigma 17-70mm F2.8-4 DC Macro OS HSM | C Tamron 16-300mm F/3.5-6.3 Di II VC PZD Macro Nikon AF-S DX Nikkor 16-80mm F2.8-4E ED VR +1 more
OP bclaff Veteran Member • Posts: 8,735
Re: I have a question...
3

PaulSnowcat wrote:

I'VE seen that some sources have different PDR measured for the SAME camra when it's working in FF and in APS-C modes. For example, PDR or Sony A7RII in FF mode is bigger then it's PDR in crop mode.

How this can be true? Dynamic Range is a difference between the most light and the most dark pixel on the sensor that still have color information in them. How this parameter can change depending on which part of the same sensor you are talking about? It should be exaclty the same... Please explain this thing to me...

This is a common statement. Your confusion results from not understanding what Photographic Dynamic Range (PDR) is.

PDR is normalized to a constant viewing condition.

The APS-C data will have to be enlarged more than the FF and that will make noise more apparent.
Therefore the useful photographic dynamic range, which is limited by noise in the shadows becoming objectionable, will be lower.

(The same applies to DxOMark Landscape dynamic range although they never publish separate values like PhotonsToPhotos.)
--

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

PaulSnowcat
PaulSnowcat Senior Member • Posts: 1,020
Re: I have a question...

Ok, then the next question is - to WHAT viewing conditions you normalize the DR data? What do you consider to be that needed vieweing conditions?

And then, different viewing conditions must affect PDR? So PDR is not only a sensor's property but also a property of how you USE the picture, rigth?

-- hide signature --
 PaulSnowcat's gear list:PaulSnowcat's gear list
Nikon D7500 Tokina AT-X Pro 12-24mm f/4 DX II Sigma 17-70mm F2.8-4 DC Macro OS HSM | C Tamron 16-300mm F/3.5-6.3 Di II VC PZD Macro Nikon AF-S DX Nikkor 16-80mm F2.8-4E ED VR +1 more
OP bclaff Veteran Member • Posts: 8,735
Re: I have a question...

PaulSnowcat wrote:

Ok, then the next question is - to WHAT viewing conditions you normalize the DR data? What do you consider to be that needed viewing conditions?

And then, different viewing conditions must affect PDR? So PDR is not only a sensor's property but also a property of how you USE the picture, rigth?

This is all covered at PhotonsToPhotos.
See the links below the Photographic Dynamic Range (PDR) chart :

You might want to start with the one I highlighted.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

Keyboard shortcuts:
FForum MMy threads