Dynamic Range Loss | ISO Settings

Can anyone give me a brief understanding of how much Dynamic Range is lost as we raise our ISO? As an example: At what ISO would I have had to shoot the enclosed picture with my Z9 to see a "noticeable difference" if I'm delivering in basic sRGB.
Why are you concerned about "dynamic range" ? Is it a term you heard and now think is important, or have you already had an issue with something that is clearly a DR issue and want to understand or enumerate it?

I don't think that most forum conversation that mentions "DR" is really about DR at all; often people just use it as a proxy for noise, and that is done in error. DR at high ISOs is not directly relevant to noise except for one thing, a really low DR for an ISO compared to other cameras most likely also means more noise, but there is no direct correlation between DR and noise, and a high DR at a given high ISO is not an indication of low noise; it may be an indication of more headroom; headroom that you would have had anyway had you chosen to "under-expose" at a lower ISO setting instead, with the same Av and Tv values.

If you're shooting a spot-lit stage and plan to leave the darkest areas very dark, "DR" probably has little to do with your concerns, unless there are also very bright highlights outside the normal range that you wish to capture well by using less exposure than normal for an ISO setting. If you're just worried about noise in "normal" tonal ranges, then DR is not a valuable metric, and you should just be concentrating on exposure, by using the largest aperture that gives the look you want, and the slowest shutter speed that you can get away with, and forget about "DR".

In any event, I think that DR is best understood by breaking it up into two parts; headroom, and footroom, both relative to the standard exposure of a grey card for an ISO setting. "Footroom" is how much aesthetically usable range you get below middle grey, and "headroom" is how much extra there is for highlights that can be captured without clipping, and any camera that scores high for DR can be doing that by giving ample headroom and poor footroom, and footroom is the only one that has anything to do with exposure-referred noise.
Hi John, How are you?

My issues are blownout faces and dark feet in mixed lighting. Comparing my edits to well lit even runways hung well the post process is pure hell. A guy next to me with a cheap EOS R did a bit better and I saw he had more latitude in a DR chart.
Could you maybe post a link to these charts?
Ive shot Hasselblad with 15 stops and also the GFX 100s. I had a lot more latitude in my blacks, shadows and highlights in post. IMHO, the R1 will be my answer to deal with really difficult lighting. I hear it will be 15 or16 stops of DR at 85MP at 30fps. I shot my last show wrong with the Z9. MY FAULT.
I don't think it's the camera. It's about your settings as you mention below. By the way, do you shoot in JPG or RAW?
My ISO at one point was at 8000. For full page fashion magazine from runway shots I need to be closer to base ISO vs away from it. Color, contrast, detail, noise.... I'm only happy when I light my own room or outdoors. People do a shi* job hanging rooms nowadays. A simple scrim at the front of a runway would do wonders to stop blown our faces and hair and shoulders. But that's too simple.;)
 
Can anyone give me a brief understanding of how much Dynamic Range is lost as we raise our ISO? As an example: At what ISO would I have had to shoot the enclosed picture with my Z9 to see a "noticeable difference" if I'm delivering in basic sRGB.
Why are you concerned about "dynamic range" ? Is it a term you heard and now think is important, or have you already had an issue with something that is clearly a DR issue and want to understand or enumerate it?

I don't think that most forum conversation that mentions "DR" is really about DR at all; often people just use it as a proxy for noise, and that is done in error. DR at high ISOs is not directly relevant to noise except for one thing, a really low DR for an ISO compared to other cameras most likely also means more noise, but there is no direct correlation between DR and noise, and a high DR at a given high ISO is not an indication of low noise; it may be an indication of more headroom; headroom that you would have had anyway had you chosen to "under-expose" at a lower ISO setting instead, with the same Av and Tv values.

If you're shooting a spot-lit stage and plan to leave the darkest areas very dark, "DR" probably has little to do with your concerns, unless there are also very bright highlights outside the normal range that you wish to capture well by using less exposure than normal for an ISO setting. If you're just worried about noise in "normal" tonal ranges, then DR is not a valuable metric, and you should just be concentrating on exposure, by using the largest aperture that gives the look you want, and the slowest shutter speed that you can get away with, and forget about "DR".

In any event, I think that DR is best understood by breaking it up into two parts; headroom, and footroom, both relative to the standard exposure of a grey card for an ISO setting. "Footroom" is how much aesthetically usable range you get below middle grey, and "headroom" is how much extra there is for highlights that can be captured without clipping, and any camera that scores high for DR can be doing that by giving ample headroom and poor footroom, and footroom is the only one that has anything to do with exposure-referred noise.
Hi John, How are you?

My issues are blownout faces and dark feet in mixed lighting. Comparing my edits to well lit even runways hung well the post process is pure hell. A guy next to me with a cheap EOS R did a bit better and I saw he had more latitude in a DR chart.
Could you maybe post a link to these charts?

Here’s our selection of the currently available cameras with the best dynamic range (we’ve linked to the Shotkit reviews where available):
  • Nikon D850 – 14.8 stops at ISO 64
  • Leica Q2 – 14.5 stops at ISO 400
  • Canon EOS R – 14.1 stops at ISO 800.
  • Leica CL – 13.6 stops at ISO 100.
  • Canon EOS M50 – 13.4 stops at ISO 100.
  • Olympus OM-D E-M10 Mark III – 12.3 stops at ISO 800.
  • Canon 5D Mark IV – 13.6 stops at ISO 100.
  • Sony α7R IV – 14.7 stops at ISO 50
  • Sony a7 III – 14.7 stops at ISO 50
  • Nikon Z7 – 14.6 stops at ISO 64
  • Panasonic Lumix DC-GH5 – 13 stops at ISO 200
Ive shot Hasselblad with 15 stops and also the GFX 100s. I had a lot more latitude in my blacks, shadows and highlights in post. IMHO, the R1 will be my answer to deal with really difficult lighting. I hear it will be 15 or16 stops of DR at 85MP at 30fps. I shot my last show wrong with the Z9. MY FAULT.
I don't think it's the camera. It's about your settings as you mention below. By the way, do you shoot in JPG or RAW?
RAW with small JPEGS for my catalog thumbnails
My ISO at one point was at 8000. For full page fashion magazine from runway shots I need to be closer to base ISO vs away from it. Color, contrast, detail, noise.... I'm only happy when I light my own room or outdoors. People do a shi* job hanging rooms nowadays. A simple scrim at the front of a runway would do wonders to stop blown our faces and hair and shoulders. But that's too simple.;)
 
Can anyone give me a brief understanding of how much Dynamic Range is lost as we raise our ISO? As an example: At what ISO would I have had to shoot the enclosed picture with my Z9 to see a "noticeable difference" if I'm delivering in basic sRGB.
Why are you concerned about "dynamic range" ? Is it a term you heard and now think is important, or have you already had an issue with something that is clearly a DR issue and want to understand or enumerate it?

I don't think that most forum conversation that mentions "DR" is really about DR at all; often people just use it as a proxy for noise, and that is done in error. DR at high ISOs is not directly relevant to noise except for one thing, a really low DR for an ISO compared to other cameras most likely also means more noise, but there is no direct correlation between DR and noise, and a high DR at a given high ISO is not an indication of low noise; it may be an indication of more headroom; headroom that you would have had anyway had you chosen to "under-expose" at a lower ISO setting instead, with the same Av and Tv values.

If you're shooting a spot-lit stage and plan to leave the darkest areas very dark, "DR" probably has little to do with your concerns, unless there are also very bright highlights outside the normal range that you wish to capture well by using less exposure than normal for an ISO setting. If you're just worried about noise in "normal" tonal ranges, then DR is not a valuable metric, and you should just be concentrating on exposure, by using the largest aperture that gives the look you want, and the slowest shutter speed that you can get away with, and forget about "DR".

In any event, I think that DR is best understood by breaking it up into two parts; headroom, and footroom, both relative to the standard exposure of a grey card for an ISO setting. "Footroom" is how much aesthetically usable range you get below middle grey, and "headroom" is how much extra there is for highlights that can be captured without clipping, and any camera that scores high for DR can be doing that by giving ample headroom and poor footroom, and footroom is the only one that has anything to do with exposure-referred noise.
Hi John, How are you?

My issues are blownout faces and dark feet in mixed lighting. Comparing my edits to well lit even runways hung well the post process is pure hell. A guy next to me with a cheap EOS R did a bit better and I saw he had more latitude in a DR chart.
Could you maybe post a link to these charts?

Here’s our selection of the currently available cameras with the best dynamic range (we’ve linked to the Shotkit reviews where available):
  • Nikon D850 – 14.8 stops at ISO 64
  • Leica Q2 – 14.5 stops at ISO 400
  • Canon EOS R – 14.1 stops at ISO 800.
  • Leica CL – 13.6 stops at ISO 100.
  • Canon EOS M50 – 13.4 stops at ISO 100.
  • Olympus OM-D E-M10 Mark III – 12.3 stops at ISO 800.
  • Canon 5D Mark IV – 13.6 stops at ISO 100.
  • Sony α7R IV – 14.7 stops at ISO 50
  • Sony a7 III – 14.7 stops at ISO 50
  • Nikon Z7 – 14.6 stops at ISO 64
  • Panasonic Lumix DC-GH5 – 13 stops at ISO 200
As I and also others mentioned, these numbers are not important for the work that you are doing (at least based on what you presented here). Do not confuse correlation with causation.
Ive shot Hasselblad with 15 stops and also the GFX 100s. I had a lot more latitude in my blacks, shadows and highlights in post. IMHO, the R1 will be my answer to deal with really difficult lighting. I hear it will be 15 or16 stops of DR at 85MP at 30fps. I shot my last show wrong with the Z9. MY FAULT.
I don't think it's the camera. It's about your settings as you mention below. By the way, do you shoot in JPG or RAW?
RAW with small JPEGS for my catalog thumbnails
I was thinking if you don't see a difference in JPG processing between difference cameras. If you shoot in RAW, then the difference is in the amount of light that gets on the sensor. Focus on selecting appropriate DOF and exposure time. Switching from Z9 to another camera won't help you.
My ISO at one point was at 8000. For full page fashion magazine from runway shots I need to be closer to base ISO vs away from it. Color, contrast, detail, noise.... I'm only happy when I light my own room or outdoors. People do a shi* job hanging rooms nowadays. A simple scrim at the front of a runway would do wonders to stop blown our faces and hair and shoulders. But that's too simple.;)
 
Typically, each doubling of ISO will halve your range, with nuances that vary by make and model. See camera data here:

https://photonstophotos.net/Charts/PDR.htm

“Dynamic range” is an unusual measurement because it combines an objective measure with a subjective judgement. So when determining dynamic range, *you* have to determine your tolerance of noise.
WOW! GREAT chart. Now how to keep my ISO below 200 for fashion runway??? I'll need to shoot at 2.8 and 1/250. Almost impossible, no runway in the world is lit that good and 1/250 won't freeze the models. Perhaps 800 ISO if I cut off chicken heads and light candles!! Thanks again Mark, I appreciate the reference. Tony
The second gain that kicks in at ISO 500 and above is less noisy than ISO settings lower than 500.

1ed8172be33a4873b6dc4ab2543db3cf.jpg.png
 
Typically, each doubling of ISO will halve your range, with nuances that vary by make and model. See camera data here:

https://photonstophotos.net/Charts/PDR.htm

“Dynamic range” is an unusual measurement because it combines an objective measure with a subjective judgement. So when determining dynamic range, *you* have to determine your tolerance of noise.
WOW! GREAT chart. Now how to keep my ISO below 200 for fashion runway??? I'll need to shoot at 2.8 and 1/250. Almost impossible, no runway in the world is lit that good and 1/250 won't freeze the models. Perhaps 800 ISO if I cut off chicken heads and light candles!! Thanks again Mark, I appreciate the reference. Tony
The second gain that kicks in at ISO 500 and above is less noisy than ISO settings lower than 500.

1ed8172be33a4873b6dc4ab2543db3cf.jpg.png
This is very interesting 🤔..... Id like to try working around this with lighting to what I need to achieve when lighting a room with fixed lighting. Perhaps we need to hang a room to be able to achive 1/500 at 500 ISO, keeping 2.8 static? The even wash would require scrims.

--




Fashion, Fashion BTS, Fashion Editorial, Sports, Athlete Portrature, Editorial and Creative Portraits are my world. Shoot Canon, Nikon & Fuji. https://www.kissmykite.com/nyfw-photographers-tony-filson
 
Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks :) 😀
That doesn't sound like it has that much to do with DR, to be honest.

A critical thing to recognize is that the added noise isn't coming from increasing the ISO, per se. The noise is likely to primarily stem from the light you're capturing (specifically how little of it you're capturing).

The biggest improvements will come from either opening up your aperture or increasing your shutter speed, as these get more light onto the sensor. (It's the greater exposure, not the lower ISO value that's the cause of low ISO images to look cleaner).
In my book, decreasing the shutter speed increases exposure.

Edit: since shutter speed is constant, it would be better to say that increasing exposure time increases exposure.
The DR chart helps point out that the lower ISO setting has wider DR. You could exploit this by:

Using your current exposure settings and reduce your ISO. This will give an underexposed JPEG but would also give you Raw files with roughly comparable levels of noise to your current shots, and with up to a stop of recoverable highlights.
+1
But you're likely to get a much bigger improvement by pushing to see whether you have any scope for increasing your exposure (and using a lower ISO). This will improve the image noise across all of your image, not just the deep shadows that DR numbers describe.

Richard - DPreview.com
 
Last edited:
Typically, each doubling of ISO will halve your range, with nuances that vary by make and model. See camera data here:

https://photonstophotos.net/Charts/PDR.htm

“Dynamic range” is an unusual measurement because it combines an objective measure with a subjective judgement. So when determining dynamic range, *you* have to determine your tolerance of noise.
Amazing the Canon 1DX is equal to the Nikon Z9 in DR starting at 400 ISO. Granted it's 21MP vs. 45MP but the 1DX is still a very relevant camera. Thanks again for the chart
A couple of key things to recognise:

DR is not image quality: it's a measure of one very specific aspect of image quality (the point at which the impact of the sum of all noise sources exceeds a specific threshold). You can make some assumptions about what happens above that threshold, but they won't always hold.

Two cameras can have the same DR figure and very different image quality.

As a measurement, DR becomes increasingly irrelevant as you raise ISO. Once you've minimized the role of any noise that occurs after the amplification step, then you just decrease DR by a stop for every stop you increase ISO (once you reach that point, there's little benefit to raising ISO any further, but that's another story).

Most sensors will deliver DR comparable to others with the same sensor size once you reach this point. Where DR differences (the ability to pull additional usable information out of the shadows) make most difference is base ISO. Beyond that point you're intentionally sacrificing it in the hope of improving the tones within your image (before trying to incorporate a wider DR), so it becomes a less useful metric.

Richard
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
 
Typically, each doubling of ISO will halve your range, with nuances that vary by make and model. See camera data here:

https://photonstophotos.net/Charts/PDR.htm

“Dynamic range” is an unusual measurement because it combines an objective measure with a subjective judgement. So when determining dynamic range, *you* have to determine your tolerance of noise.
Amazing the Canon 1DX is equal to the Nikon Z9 in DR starting at 400 ISO. Granted it's 21MP vs. 45MP but the 1DX is still a very relevant camera. Thanks again for the chart
A couple of key things to recognise:

DR is not image quality: it's a measure of one very specific aspect of image quality (the point at which the impact of the sum of all noise sources exceeds a specific threshold). You can make some assumptions about what happens above that threshold, but they won't always hold.

Two cameras can have the same DR figure and very different image quality.

As a measurement, DR becomes increasingly irrelevant as you raise ISO. Once you've minimized the role of any noise that occurs after the amplification step, then you just decrease DR by a stop for every stop you increase ISO (once you reach that point, there's little benefit to raising ISO any further, but that's another story).

Most sensors will deliver DR comparable to others with the same sensor size once you reach this point. Where DR differences (the ability to pull additional usable information out of the shadows) make most difference is base ISO. Beyond that point you're intentionally sacrificing it in the hope of improving the tones within your image (before trying to incorporate a wider DR), so it becomes a less useful metric.

Richard
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
The thing is that modern cameras with the same sensor size only rarely exhibit one stop difference in DR at high ISO. So when you see such difference, it likely comes from a comparison of cameras with different sensor sizes and then you also have different SNR and so on (at the same ISO). And then the question is whether it's really DR that causing the observed difference in IQ.

Did you have some particular comparison of cameras in your mind?
 
Typically, each doubling of ISO will halve your range, with nuances that vary by make and model. See camera data here:

https://photonstophotos.net/Charts/PDR.htm

“Dynamic range” is an unusual measurement because it combines an objective measure with a subjective judgement. So when determining dynamic range, *you* have to determine your tolerance of noise.
Amazing the Canon 1DX is equal to the Nikon Z9 in DR starting at 400 ISO. Granted it's 21MP vs. 45MP but the 1DX is still a very relevant camera. Thanks again for the chart
A couple of key things to recognise:

DR is not image quality: it's a measure of one very specific aspect of image quality (the point at which the impact of the sum of all noise sources exceeds a specific threshold). You can make some assumptions about what happens above that threshold, but they won't always hold.

Two cameras can have the same DR figure and very different image quality.

As a measurement, DR becomes increasingly irrelevant as you raise ISO. Once you've minimized the role of any noise that occurs after the amplification step, then you just decrease DR by a stop for every stop you increase ISO (once you reach that point, there's little benefit to raising ISO any further, but that's another story).

Most sensors will deliver DR comparable to others with the same sensor size once you reach this point. Where DR differences (the ability to pull additional usable information out of the shadows) make most difference is base ISO. Beyond that point you're intentionally sacrificing it in the hope of improving the tones within your image (before trying to incorporate a wider DR), so it becomes a less useful metric.

Richard
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
The thing is that modern cameras with the same sensor size only rarely exhibit one stop difference in DR at high ISO. So when you see such difference, it likely comes from a comparison of cameras with different sensor sizes and then you also have different SNR and so on (at the same ISO). And then the question is whether it's really DR that causing the observed difference in IQ.

Did you have some particular comparison of cameras in your mind?

DR is computed from SNR, they are related.

I was mainly thinking of cameras with different sensor sizes, but Leica SL2 has less PDR than Z7, and it matters mostly at higher ISOs
 
Typically, each doubling of ISO will halve your range, with nuances that vary by make and model. See camera data here:

https://photonstophotos.net/Charts/PDR.htm

“Dynamic range” is an unusual measurement because it combines an objective measure with a subjective judgement. So when determining dynamic range, *you* have to determine your tolerance of noise.
Amazing the Canon 1DX is equal to the Nikon Z9 in DR starting at 400 ISO. Granted it's 21MP vs. 45MP but the 1DX is still a very relevant camera. Thanks again for the chart
A couple of key things to recognise:

DR is not image quality: it's a measure of one very specific aspect of image quality (the point at which the impact of the sum of all noise sources exceeds a specific threshold). You can make some assumptions about what happens above that threshold, but they won't always hold.

Two cameras can have the same DR figure and very different image quality.

As a measurement, DR becomes increasingly irrelevant as you raise ISO. Once you've minimized the role of any noise that occurs after the amplification step, then you just decrease DR by a stop for every stop you increase ISO (once you reach that point, there's little benefit to raising ISO any further, but that's another story).

Most sensors will deliver DR comparable to others with the same sensor size once you reach this point. Where DR differences (the ability to pull additional usable information out of the shadows) make most difference is base ISO. Beyond that point you're intentionally sacrificing it in the hope of improving the tones within your image (before trying to incorporate a wider DR), so it becomes a less useful metric.

Richard
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
The thing is that modern cameras with the same sensor size only rarely exhibit one stop difference in DR at high ISO. So when you see such difference, it likely comes from a comparison of cameras with different sensor sizes and then you also have different SNR and so on (at the same ISO). And then the question is whether it's really DR that causing the observed difference in IQ.

Did you have some particular comparison of cameras in your mind?
DR is computed from SNR, they are related.

I was mainly thinking of cameras with different sensor sizes, but Leica SL2 has less PDR than Z7, and it matters mostly at higher ISOs
SNR corresponds to noise that you see at given tone, typically some neutral gray when it is used for camera comparison. DR corresponds to a range of tones between saturation and some SNR. SNR threshold of 1 is typically used in engineering but such low SNR is not very useful for general photography. So Bill Claff uses SNR threshold of 10, if I remember correctly, and calls this variation PDR.

The fact that DR and SNR are highly correlated in modern camera sensors comes from the physics of light noise and the fact that read noise is quite low these days. But that doesn't mean that these terms can be used interchangeably.
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization. Either could yield a higher "DR" or "PDR" when actually having MORE exposure-referred noise. Neither of those metrics, for example, is applied to Canon's "HTP", but if they were measured for HTP, they would give up to a stop more "DR" or "PDR", but actually have more read noise.

"DR" or "PDR" are both poor proxies for exposure-referred noise, that just happen to have some correlation, in some cases (same raw headroom).
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:


Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
Either could yield a higher "DR" or "PDR" when actually having MORE exposure-referred noise. Neither of those metrics, for example, is applied to Canon's "HTP", but if they were measured for HTP, they would give up to a stop more "DR" or "PDR", but actually have more read noise.

"DR" or "PDR" are both poor proxies for exposure-referred noise, that just happen to have some correlation, in some cases (same raw headroom).
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Either could yield a higher "DR" or "PDR" when actually having MORE exposure-referred noise. Neither of those metrics, for example, is applied to Canon's "HTP", but if they were measured for HTP, they would give up to a stop more "DR" or "PDR", but actually have more read noise.

"DR" or "PDR" are both poor proxies for exposure-referred noise, that just happen to have some correlation, in some cases (same raw headroom).
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
Midtones.
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
Midtones.
And what about the shutter speed, is it the same as for Z7 (assuming the same aperture and ISO setting are used)?
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
Midtones.
And what about the shutter speed, is it the same as for Z7 (assuming the same aperture and ISO setting are used)?
IIRC, I compared same exposures (shutter speed and aperture).
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
Midtones.
And what about the shutter speed, is it the same as for Z7 (assuming the same aperture and ISO setting are used)?
IIRC, I compared same exposures (shutter speed and aperture).
That would mean that SL2 sensor has lower efficiency. Like if you would put an ND filter in front of the Z7 sensor and compensated with gain to provide the same image lightness. Have you any idea what Leica did with SL2 sensor?
 
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
Midtones.
And what about the shutter speed, is it the same as for Z7 (assuming the same aperture and ISO setting are used)?
IIRC, I compared same exposures (shutter speed and aperture).
That would mean that SL2 sensor has lower efficiency. Like if you would put an ND filter in front of the Z7 sensor and compensated with gain to provide the same image lightness. Have you any idea what Leica did with SL2 sensor?
I do not know the details or manufacturer of SL2 sensor. SL2-S (24MP with a newer sensor), on the other hand, has excellent high ISO behavior and especially in deep shadows.

Edit: Jim Kasson ran some testes with Q2M sensor which is the same as in SL2 but without CFA.
 
Last edited:
My impression, when using different cameras, is that DR difference becomes increasingly relevant as I increase the ISO. For example, one stop of max DR difference at base ISO is less relevant than one stop of max DR difference at high ISO.
That is only strictly true when you normalize for saturation, such as when looking at DxOMark's "DR", but looking at the trendlines, rather than the data points. The data points, like Photons2Photos' "PDR" have no exposure reference or normalization.
I believe this is the reason why Leica SL2 shows "lower" PDR than other FF cameras, see this graph:

https://www.photonstophotos.net/Cha...6II,Nikon Z 7II,Sony ILCE-7RM5,Sony ILCE-7SM3

Notice than line for SL2 shows a one stop improvement between ISO 50 and 100 while most sensors reach highest PDR at ISO 100 (Nikon Z7 is an exception). This is a strong indication that ISO values of SL2 are about one stop lower than for the other cameras. If one shifts SL2 curve by one stop to the right, it would nicely match other FF cameras and there is no more PDR deficit.
No, SL2’s and Q2’s curves shouldn’t be shifted. There is some weirdness around ISO50 (nonlinearity ?) and it is best avoided. At ISO 800 and higher, the visible noise is clearly higher than in a Z7.
Are you talking about noise in midtones or very deep shadows?
Midtones.
And what about the shutter speed, is it the same as for Z7 (assuming the same aperture and ISO setting are used)?
IIRC, I compared same exposures (shutter speed and aperture).
That would mean that SL2 sensor has lower efficiency. Like if you would put an ND filter in front of the Z7 sensor and compensated with gain to provide the same image lightness. Have you any idea what Leica did with SL2 sensor?
I do not know the details or manufacturer of SL2 sensor. SL2-S (24MP with a newer sensor), on the other hand, has excellent high ISO behavior and especially in deep shadows.

Edit: Jim Kasson ran some testes with Q2M sensor which is the same as in SL2 but without CFA.
It seems that Leica is doing something unusual, it's also mentioned in DPR review of Q2. I checked the test scene and Q2 is noisier than Z7 but the difference is rather small. Maybe 0.33 stop? But the PDR is 1 stop lower. Bill Claff's measurement indicate that read noise of Q2 is significantly larger than one of Z7.

All of this shows that these metrics tell you something different about the sensor performance and they are not interchangeable.
 

Keyboard shortcuts

Back
Top