Utility of dynamic range beyond a certain point
p.10 #1Utility of dynamic range beyond a certain point
jtra wrote:
Is my thinking wrong?Cameras usually have lower (or same) read noise (compared to captured data) when they are set for higher for high ISOs, not for lower ISOs.
So relative lower read noise is achieved by using higher ISO (second base ISO, e.g. 640 over using base ISO 100).
You can see read noise increases in this Read noise in DN chart: https://www.photonstophotos.net/Charts/RN_ADU.htm
But relative to captured data it decreases in this Input-referred Read Noise chart: https://www.photonstophotos.net/Charts/RN_e.htm
It sounds as if you refer to the conventional Dual Conversion Gain architecture, that indeed works to reduce read noise at the "second base ISO."
My question to Horshack was about the Dual Gain Output.
ruthenium wrote
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows". My understanding is that the latter should work to reduce BOTH the read noise and whatever signal that happens to be in those "shadows." It is this latter scenario that I referred to when speculated that the read noise "is suppressed together with the shadow signal". Thus, I expected that the shadow signal-to-read noise ratio shouldn't change in the A7RVI vs A7RV.Is my thinking wrong?
So relative lower read noise is achieved by using higher ISO (second base ISO, e.g. 640 over using base ISO 100).
You can see read noise increases in this Read noise in DN chart: https://www.photonstophotos.net/Charts/RN_ADU.htm
But relative to captured data it decreases in this Input-referred Read Noise chart: https://www.photonstophotos.net/Charts/RN_e.htm
It sounds as if you refer to the conventional Dual Conversion Gain architecture, that indeed works to reduce read noise at the "second base ISO."
My question to Horshack was about the Dual Gain Output.
Jun 21, 2026 at 07:02 AM
p.10 #2Utility of dynamic range beyond a certain point
ruthenium wrote:
The a7rVI achieves it's higher DR with the mechanical shutter via a lower-noise readout path - there's no suppression of signal.
Deep shadow noise is composed primarily of read noise (electronics), not shot noise (light), thus even if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor. Here's a visualization of a sample Photon Transfer Curve which demonstrates this:
The horizontal axis is the sensor saturation, left is 0% at noise floor, right is 100% at sensor saturation. The vertical axis is the % of noise each color-coded noise component represents in the total noise for that saturation level.Can you expand on the "lower-noise readout path"?
My thinking is that the A7RVI uses the Dual Gain Output (one optimized for highlights and one optimized for shadows).
Perhaps you mean the same, only use different terminology.
I fully appreciate that "if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor"
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows"....Show more →
The a7rVI achieves its lower low-iso read noise via DGO, which can apply both low (LCG) and high (HCG) conversion gain to the same signal, merging the two readouts into a single pixel value. The LCG-converted value represents the highlights, midtones, and upper shadows; the HCG-converted value the deep shadows. HCG achieves the lower read noise for its component of the signal via more precise electron counting, as it can apply more divisions of signal over a given voltage range. You can see this described in detail in the original Aptina whitepaper (which applies to DCG sensors in general):
https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf
snapsy wrote
ruthenium wrote
4) I suspect that the extended dynamic range of the A7RVI (and that of A7V) is misleading. Obviously, the A7RVI at ISO 100 captures the same (or practically the same) amount of light as the A7RV at ISO 100. Thus, the shadow signal-to-read noise ratio should be practically the same in both. The way how the A7RVI achieves the "better" dynamic range is by suppressing the shadow signal. The problem is that the read noise is not suppressed selectively. It is suppressed together with the shadow signal. I speculate here. To prove me right or wrong, one should compare the shadows in the same scene captured with the A7RVI and A7RV side-by-side.Deep shadow noise is composed primarily of read noise (electronics), not shot noise (light), thus even if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor. Here's a visualization of a sample Photon Transfer Curve which demonstrates this:
The horizontal axis is the sensor saturation, left is 0% at noise floor, right is 100% at sensor saturation. The vertical axis is the % of noise each color-coded noise component represents in the total noise for that saturation level.
My thinking is that the A7RVI uses the Dual Gain Output (one optimized for highlights and one optimized for shadows).
Perhaps you mean the same, only use different terminology.
I fully appreciate that "if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor"
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows"....Show more →
The a7rVI achieves its lower low-iso read noise via DGO, which can apply both low (LCG) and high (HCG) conversion gain to the same signal, merging the two readouts into a single pixel value. The LCG-converted value represents the highlights, midtones, and upper shadows; the HCG-converted value the deep shadows. HCG achieves the lower read noise for its component of the signal via more precise electron counting, as it can apply more divisions of signal over a given voltage range. You can see this described in detail in the original Aptina whitepaper (which applies to DCG sensors in general):
https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf
Jun 21, 2026 at 07:02 AM
p.10 #3Utility of dynamic range beyond a certain point
snapsy wrote:
The a7rVI achieves it's higher DR with the mechanical shutter via a lower-noise readout path - there's no suppression of signal.
Deep shadow noise is composed primarily of read noise (electronics), not shot noise (light), thus even if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor. Here's a visualization of a sample Photon Transfer Curve which demonstrates this:
The horizontal axis is the sensor saturation, left is 0% at noise floor, right is 100% at sensor saturation. The vertical axis is the % of noise each color-coded noise component represents in the total noise for that saturation level.Can you expand on the "lower-noise readout path"?
My thinking is that the A7RVI uses the Dual Gain Output (one optimized for highlights and one optimized for shadows).
Perhaps you mean the same, only use different terminology.
I fully appreciate that "if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor"
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows". My understanding is that the latter should work to reduce BOTH the read noise and whatever signal that happens to be in those "shadows." It is this latter scenario that I referred to when speculated that the read noise "is suppressed together with the shadow signal". Thus, I expected that the shadow signal-to-read noise ratio shouldn't change in the A7RVI vs A7RV.
Is my thinking wrong?The a7rVI achieves its lower low-iso read noise via DGO, which can apply both low (LCG) and high (HCG) conversion gain to the same signal, merging the two readouts into a single pixel value. The LCG-converted value represents the highlights, midtones, and upper shadows; the HCG-converted value the deep shadows. HCG achieves the lower read noise for its component of the signal via more precise electron counting, as it can apply more divisions of signal over a given voltage range. You can see this described in detail in the original Aptina whitepaper (which applies to DCG sensors in general):
https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf
I am not so much interested in the deep technical details of how the DGO is implemented, but more in the question on whether the deep shadow signal - to - read noise ratio is similar in the A7RVI and A7RV.
Now, you state specifically "the HCG-converted value the deep shadows" - that implies both the read noise and any signal in the deep shadows.
Thus, my expectation is that only the "upper shadows" should benefit from the DR increased as a result of the DGO.
Correct?
ruthenium wrote
snapsy wrote
ruthenium wrote
4) I suspect that the extended dynamic range of the A7RVI (and that of A7V) is misleading. Obviously, the A7RVI at ISO 100 captures the same (or practically the same) amount of light as the A7RV at ISO 100. Thus, the shadow signal-to-read noise ratio should be practically the same in both. The way how the A7RVI achieves the "better" dynamic range is by suppressing the shadow signal. The problem is that the read noise is not suppressed selectively. It is suppressed together with the shadow signal. I speculate here. To prove me right or wrong, one should compare the shadows in the same scene captured with the A7RVI and A7RV side-by-side.Deep shadow noise is composed primarily of read noise (electronics), not shot noise (light), thus even if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor. Here's a visualization of a sample Photon Transfer Curve which demonstrates this:
The horizontal axis is the sensor saturation, left is 0% at noise floor, right is 100% at sensor saturation. The vertical axis is the % of noise each color-coded noise component represents in the total noise for that saturation level.
My thinking is that the A7RVI uses the Dual Gain Output (one optimized for highlights and one optimized for shadows).
Perhaps you mean the same, only use different terminology.
I fully appreciate that "if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor"
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows". My understanding is that the latter should work to reduce BOTH the read noise and whatever signal that happens to be in those "shadows." It is this latter scenario that I referred to when speculated that the read noise "is suppressed together with the shadow signal". Thus, I expected that the shadow signal-to-read noise ratio shouldn't change in the A7RVI vs A7RV.
Is my thinking wrong?
https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf
I am not so much interested in the deep technical details of how the DGO is implemented, but more in the question on whether the deep shadow signal - to - read noise ratio is similar in the A7RVI and A7RV.
Now, you state specifically "the HCG-converted value the deep shadows" - that implies both the read noise and any signal in the deep shadows.
Thus, my expectation is that only the "upper shadows" should benefit from the DR increased as a result of the DGO.
Correct?
Jun 21, 2026 at 07:10 AM
p.10 #4Utility of dynamic range beyond a certain point
ruthenium wrote:
The a7rVI achieves it's higher DR with the mechanical shutter via a lower-noise readout path - there's no suppression of signal.
Deep shadow noise is composed primarily of read noise (electronics), not shot noise (light), thus even if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor. Here's a visualization of a sample Photon Transfer Curve which demonstrates this:
The horizontal axis is the sensor saturation, left is 0% at noise floor, right is 100% at sensor saturation. The vertical axis is the % of noise each color-coded noise component represents in the total noise for that saturation level.Can you expand on the "lower-noise readout path"?
My thinking is that the A7RVI uses the Dual Gain Output (one optimized for highlights and one optimized for shadows).
Perhaps you mean the same, only use different terminology.
I fully appreciate that "if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor"
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows". My understanding is that the latter should work to reduce BOTH the read noise and whatever signal that happens to be in those "shadows." It is this latter scenario that I referred to when speculated that the read noise "is suppressed together with the shadow signal". Thus, I expected that the shadow signal-to-read noise ratio shouldn't change in the A7RVI vs A7RV.
Is my thinking wrong?The a7rVI achieves its lower low-iso read noise via DGO, which can apply both low (LCG) and high (HCG) conversion gain to the same signal, merging the two readouts into a single pixel value. The LCG-converted value represents the highlights, midtones, and upper shadows; the HCG-converted value the deep shadows. HCG achieves the lower read noise for its component of the signal via more precise electron counting, as it can apply more divisions of signal over a given voltage range. You can see this described in detail in the original Aptina whitepaper (which applies to DCG sensors in general):
https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdfI am not so much interested in the deep technical details of how the DGO is implemented, but more in the question on whether the deep shadow signal - to - read noise ratio is similar in the A7RVI and A7RV.
Now, you state specifically "the HCG-converted value the deep shadows" - that implies both the read noise and any signal in the deep shadows.
Thus, my expectation is that only the "upper shadows" should benefit from the DR increased as a result of the DGO.
Correct?
DR is the full tonal range of detectable, non-overflowed signal above a minimum SNR threshold, which means DR can be increased by either reducing the noise near the shadow floor (which is what's SNR-limited) or by increasing the electron holding capacity near saturation (which has plenty of SNR but is "overflow limited"). For the a7rVI it's the former, which means the same signal levels near the shadow floor are read out with lower noise.
snapsy wrote
ruthenium wrote
snapsy wrote
ruthenium wrote
4) I suspect that the extended dynamic range of the A7RVI (and that of A7V) is misleading. Obviously, the A7RVI at ISO 100 captures the same (or practically the same) amount of light as the A7RV at ISO 100. Thus, the shadow signal-to-read noise ratio should be practically the same in both. The way how the A7RVI achieves the "better" dynamic range is by suppressing the shadow signal. The problem is that the read noise is not suppressed selectively. It is suppressed together with the shadow signal. I speculate here. To prove me right or wrong, one should compare the shadows in the same scene captured with the A7RVI and A7RV side-by-side.Deep shadow noise is composed primarily of read noise (electronics), not shot noise (light), thus even if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor. Here's a visualization of a sample Photon Transfer Curve which demonstrates this:
The horizontal axis is the sensor saturation, left is 0% at noise floor, right is 100% at sensor saturation. The vertical axis is the % of noise each color-coded noise component represents in the total noise for that saturation level.
My thinking is that the A7RVI uses the Dual Gain Output (one optimized for highlights and one optimized for shadows).
Perhaps you mean the same, only use different terminology.
I fully appreciate that "if the two sensors capture the same amount of light the sensor with the lower read noise will have a higher SNR in the deep shadows near the noise floor"
The key question is whether the A7RVI has an intrinsically "lower read noise" when compared to A7RV, or if this "lower read noise" is achieved by applying a lower gain "optimized for shadows". My understanding is that the latter should work to reduce BOTH the read noise and whatever signal that happens to be in those "shadows." It is this latter scenario that I referred to when speculated that the read noise "is suppressed together with the shadow signal". Thus, I expected that the shadow signal-to-read noise ratio shouldn't change in the A7RVI vs A7RV.
Is my thinking wrong?
https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf
Now, you state specifically "the HCG-converted value the deep shadows" - that implies both the read noise and any signal in the deep shadows.
Thus, my expectation is that only the "upper shadows" should benefit from the DR increased as a result of the DGO.
Correct?
DR is the full tonal range of detectable, non-overflowed signal above a minimum SNR threshold, which means DR can be increased by either reducing the noise near the shadow floor (which is what's SNR-limited) or by increasing the electron holding capacity near saturation (which has plenty of SNR but is "overflow limited"). For the a7rVI it's the former, which means the same signal levels near the shadow floor are read out with lower noise.
Jun 21, 2026 at 07:25 AM
p.10 #5Utility of dynamic range beyond a certain point
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands 
Jun 21, 2026 at 07:46 AM
p.10 #6Utility of dynamic range beyond a certain point
fjablo wrote:
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands
https://www.fredmiranda.com/forum/topic/1945915/5#17057193
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands
https://www.fredmiranda.com/forum/topic/1945915/5#17057193
Jun 21, 2026 at 07:50 AM
p.10 #7Utility of dynamic range beyond a certain point
tctmp wrote:
https://live.staticflickr.com/65535/52708021815_3d8b1c9d4f_b.jpgCranes, Sunrise, Winter Sky
This is, obviously, a single exposure. Exposure settings were chosen to a) deal with moving birds, b) avoid blowing out the sun entirely, c) keep barely enough shadow detail to get the effect I wanted in post. And, yes, this photograph required some sophisticated work in post, as I knew it would when I made the exposure. (Single exposure using a Canon 5DsR.)While I understand it's a personal choice, this picture looks unnatural to me immediately. The upper band of yellow that's as bright as the sun, I don't think it needs to be a photographer who's heavily engaged in post processing to notice this. People who have seen open space sunsets a few times will likely catch that already.
It is “unnatural,” of course. as no human eye and no camera can actually look directly into the sun that way and “see” everything that is in this photograph. The “natural” experience of looking that direction might be blindness. And the birds would be moving. And the difference in brightness betwee the direct sun and the dark areas at the bottom would be beyond the ability of eyes to perceive.
Maybe I’ll hunt up the original raw later on and show you what the “reality” (from the camera’s perspective) actually looked like. :-)
There’s this idea that the best photographs are “real” and just the result of getting all the settings right and punching the shutter release. A perfect copy of “reality” will emerge from the camera.
But that’s how photography works.
As we’ve discussed in this thread, the camera does not “see” the way that human vision sees, and it is really a lost cause to chase that impossible goal. Cameras stop action. Cameras allow action to blur the subject. Cameras capture image elements that cannot be seen. (For example, color in extremely low light.)
I also agree that those of us who understand post-processing through experience are likely to look at a photograph like this
one, realize that it could not possibly be the results of a simple capture, and speculate about how it was realized. (Do that with Adam’s “Moonrise, Hernandez” sometime…)
As to your aesthetic response to the photograph, I’m fine with that. I’ve learned that there is nothing creative that any of us can do in this world that will be loved by everyone, and I expect that no photograph I create will speak to every viewer. :-)
- - -
ruthenium wrote:
1) Preventing or minimizing highlight clipping is achieved by adjusting exposure, not by extending the dynamic range.
2) An extended dynamic range can allow capturing more light. The utility of this is in the shadows, not in the highlights. ..
Conceptually this is right on. In situations where the subject contains a wide dynamic range, strive for a curve where the brightest tones are as close to the right edge as possible (to generalize) without blowing them out and then plan to recover/lighten the darkest tones in post as necessary.
Another way to say it is that the goal is to capture the maximum range of usable image data in the file so that you can realize the final image in post.
- - -
Daran wrote:
That strikes me as the logical equivalent of saying that some people make espresso poorly, therefore espresso is bad. ;-)It is not. It may have eluded you, but I did not write that masks make bad images. Not sure why what I meant wasn't obvious to you, but maybe you just wanted to use me as a segway? If so, I'm sorry I responded.
Here’s what you actually wrote:
Daran wrote:
One mans flaw is another mans feature. By using masks you introduce discontinuities into your processing. Which if overdone (common) is something that makes the result look artificial. A safer and more natural approach (assuming the required DR range isn't too wide) is to limit yourself to monotone continuous transfer curves, where what is brighter consistently remains brighter, just not by as much. The C1 HDR adjustments do just this and tend to work very well for typical scenes with a bright sky. But even there, if you overdo it, the result can look cooked.
Also, I suspect you meant “segue,” not “segway?” ;-)
- - -
fjablo wrote:
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands
That’s a fair point, for sure! (I can’t help you since I have neither the A7r6 nor A7r5, etc.)
I’ll speculate, based on doing a lot of photography with cameras that have various levels of DR capability that…
… the differences between most of the cameras that people want to compare here will be pretty darned trivial, bordering on insignificant in real world photography.
That doesn’t mean that continuing to expand camera dynamic range isn’t a worthwhile endeavor. It just means that each incremental step along the way will often be quite small. The improvement from one model to its successor is rarely earthshaking. The sum improvement since my first digital camera (in the mid-1990s) until today is utterly remarkable.
- - -
I’ll end this post (finally!) and get back to the initial “utility of dynamic range beyond a certain point” question by reprising a context that helps me frame the DR stuff. Relative to DR differences among cameras there are three situations you might encounter:
1. The scene/subject presents a dynamic range no larger than what can be handled by essentially any good camera. This is by far the most common situation.
2. the subject presents a dynamic range that is so large that no current camera (that any of us has access to) can fully handle it in a single exposure. This is a small percentage of shots, but it does happen.
3. The subject presents a dynamic range that is just barely within the range of the best DR camera and just slightly out of range of some alternative camera with incrementally smaller DR and the difference is visible in the final image. This is extremely rare.
Edited on Jun 21, 2026 at 09:23 PM · View previous versions
gdanmitchell wrote
Since I’m here, a contribution to the sun-in-the-frame-with-older-camera thread:https://live.staticflickr.com/65535/52708021815_3d8b1c9d4f_b.jpgCranes, Sunrise, Winter Sky
This is, obviously, a single exposure. Exposure settings were chosen to a) deal with moving birds, b) avoid blowing out the sun entirely, c) keep barely enough shadow detail to get the effect I wanted in post. And, yes, this photograph required some sophisticated work in post, as I knew it would when I made the exposure. (Single exposure using a Canon 5DsR.)
It is “unnatural,” of course. as no human eye and no camera can actually look directly into the sun that way and “see” everything that is in this photograph. The “natural” experience of looking that direction might be blindness. And the birds would be moving. And the difference in brightness betwee the direct sun and the dark areas at the bottom would be beyond the ability of eyes to perceive.
Maybe I’ll hunt up the original raw later on and show you what the “reality” (from the camera’s perspective) actually looked like. :-)
There’s this idea that the best photographs are “real” and just the result of getting all the settings right and punching the shutter release. A perfect copy of “reality” will emerge from the camera.
But that’s how photography works.
As we’ve discussed in this thread, the camera does not “see” the way that human vision sees, and it is really a lost cause to chase that impossible goal. Cameras stop action. Cameras allow action to blur the subject. Cameras capture image elements that cannot be seen. (For example, color in extremely low light.)
I also agree that those of us who understand post-processing through experience are likely to look at a photograph like this
one, realize that it could not possibly be the results of a simple capture, and speculate about how it was realized. (Do that with Adam’s “Moonrise, Hernandez” sometime…)
As to your aesthetic response to the photograph, I’m fine with that. I’ve learned that there is nothing creative that any of us can do in this world that will be loved by everyone, and I expect that no photograph I create will speak to every viewer. :-)
- - -
ruthenium wrote:
1) Preventing or minimizing highlight clipping is achieved by adjusting exposure, not by extending the dynamic range.
2) An extended dynamic range can allow capturing more light. The utility of this is in the shadows, not in the highlights. ..
Conceptually this is right on. In situations where the subject contains a wide dynamic range, strive for a curve where the brightest tones are as close to the right edge as possible (to generalize) without blowing them out and then plan to recover/lighten the darkest tones in post as necessary.
Another way to say it is that the goal is to capture the maximum range of usable image data in the file so that you can realize the final image in post.
- - -
Daran wrote:
gdanmitchell wrote
The words “which if overdone” are pulling a whole lot of weight in that post!That strikes me as the logical equivalent of saying that some people make espresso poorly, therefore espresso is bad. ;-)
Here’s what you actually wrote:
Daran wrote:
One mans flaw is another mans feature. By using masks you introduce discontinuities into your processing. Which if overdone (common) is something that makes the result look artificial. A safer and more natural approach (assuming the required DR range isn't too wide) is to limit yourself to monotone continuous transfer curves, where what is brighter consistently remains brighter, just not by as much. The C1 HDR adjustments do just this and tend to work very well for typical scenes with a bright sky. But even there, if you overdo it, the result can look cooked.
Also, I suspect you meant “segue,” not “segway?” ;-)
- - -
fjablo wrote:
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands
That’s a fair point, for sure! (I can’t help you since I have neither the A7r6 nor A7r5, etc.)
I’ll speculate, based on doing a lot of photography with cameras that have various levels of DR capability that…
… the differences between most of the cameras that people want to compare here will be pretty darned trivial, bordering on insignificant in real world photography.
That doesn’t mean that continuing to expand camera dynamic range isn’t a worthwhile endeavor. It just means that each incremental step along the way will often be quite small. The improvement from one model to its successor is rarely earthshaking. The sum improvement since my first digital camera (in the mid-1990s) until today is utterly remarkable.
- - -
I’ll end this post (finally!) and get back to the initial “utility of dynamic range beyond a certain point” question by reprising a context that helps me frame the DR stuff. Relative to DR differences among cameras there are three situations you might encounter:
1. The scene/subject presents a dynamic range no larger than what can be handled by essentially any good camera. This is by far the most common situation.
2. the subject presents a dynamic range that is so large that no current camera (that any of us has access to) can fully handle it in a single exposure. This is a small percentage of shots, but it does happen.
3. The subject presents a dynamic range that is just barely within the range of the best DR camera and just slightly out of range of some alternative camera with incrementally smaller DR and the difference is visible in the final image. This is extremely rare.
Edited on Jun 21, 2026 at 09:23 PM · View previous versions
Jun 21, 2026 at 09:08 AM
p.10 #8Utility of dynamic range beyond a certain point
snapsy wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193
Except that particular image is essentially irrelevant to showing an actual photographic need. I believe the point was directed to one extra stop of DR making a good image better, and here I have to agree that nobody’s shown us an example where it does, only the contrary. A lot of theory and measurbating attempting to support the claim, but no images where it actually makes a notable difference.
In theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsExcept that particular image is essentially irrelevant to showing an actual photographic need. I believe the point was directed to one extra stop of DR making a good image better, and here I have to agree that nobody’s shown us an example where it does, only the contrary. A lot of theory and measurbating attempting to support the claim, but no images where it actually makes a notable difference.
In theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
Jun 21, 2026 at 09:14 AM
p.10 #9Utility of dynamic range beyond a certain point
Jack Flesher wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193Except that particular image is essentially irrelevant to showing an actual photographic need. I believe the point was directed to one extra stop of DR making a good image better, and here I have to agree that nobody’s shown us an example where it does, only the contrary. A lot of theory and measurbating attempting to support the claim, but no images where it actually makes a notable difference.
In theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
"Good" images are a matter of opinion. The increase in DR in the example provided is a demonstrated fact.
snapsy wrote
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsIn theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
"Good" images are a matter of opinion. The increase in DR in the example provided is a demonstrated fact.
Jun 21, 2026 at 09:31 AM
p.10 #10Utility of dynamic range beyond a certain point
snapsy wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193
Not a real-world example, rather an immediate delete (no offense, I know you just constructed a test )
Imo this image is incorrectly exposed - approx 2 stops under - which makes your analysis moot. The edit is also unnatural, the window should be much brighter (ideally not completely blown, but much brighter than the interior). Also your later attempt did not get the lighting ratio right.
Here are two real-world examples of interiors with windows:
Kyoto by Felix, auf Flickr
Sant'Andrea della Valle III by Felix, auf Flickr
And to answer my own question: I f*cked up the exposure on the Kyoto image, which is why it has more noise in the shadows than it needs to have. Guess that new sensor tech would have helped here (or me not messing up)
Edited on Jun 21, 2026 at 10:50 AM · View previous versions
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsNot a real-world example, rather an immediate delete (no offense, I know you just constructed a test
)Imo this image is incorrectly exposed - approx 2 stops under - which makes your analysis moot. The edit is also unnatural, the window should be much brighter (ideally not completely blown, but much brighter than the interior). Also your later attempt did not get the lighting ratio right.
Here are two real-world examples of interiors with windows:
Kyoto by Felix, auf FlickrSant'Andrea della Valle III by Felix, auf FlickrAnd to answer my own question: I f*cked up the exposure on the Kyoto image, which is why it has more noise in the shadows than it needs to have. Guess that new sensor tech would have helped here (or me not messing up)
Edited on Jun 21, 2026 at 10:50 AM · View previous versions
Jun 21, 2026 at 10:48 AM
|
p.10 #11Utility of dynamic range beyond a certain point
snapsy wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193Except that particular image is essentially irrelevant to showing an actual photographic need. I believe the point was directed to one extra stop of DR making a good image better, and here I have to agree that nobody’s shown us an example where it does, only the contrary. A lot of theory and measurbating attempting to support the claim, but no images where it actually makes a notable difference.
In theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12."Good" images are a matter of opinion. The increase in DR in the example provided is a demonstrated fact.
On this we disagree. To my eyes in this example, the “increase” shown was irrelevant.
Jack Flesher wrote
snapsy wrote
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsIn theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
On this we disagree. To my eyes in this example, the “increase” shown was irrelevant.
Jun 21, 2026 at 10:49 AM
p.10 #12Utility of dynamic range beyond a certain point
fjablo wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193Not a real-world example, rather an immediate delete (no offense, I know you just constructed a test )
Imo this image is incorrectly exposed - approx 2 stops under - which makes your analysis moot. The edit is also unnatural, the window should be much brighter (ideally not completely blown, but much brighter than the interior). Also your later attempt did not get the lighting ratio right.
Here are two real-world examples of interiors with windows:
https://live.staticflickr.com/65535/55187429095_81a324dec5_b.jpgKyoto by Felix, auf Flickr
https://live.staticflickr.com/65535/53602072430_0c454a8040_b.jpgSant'Andrea della Valle III by Felix, auf Flickr
And to answer my own question: I f*cked up the exposure on the...Show more →
No offense taken - it was a real photo, taken several years ago for a quick proof to a realtor. I also bracketed the exposure for the final deliverable.
snapsy wrote
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands )Imo this image is incorrectly exposed - approx 2 stops under - which makes your analysis moot. The edit is also unnatural, the window should be much brighter (ideally not completely blown, but much brighter than the interior). Also your later attempt did not get the lighting ratio right.
Here are two real-world examples of interiors with windows:
https://live.staticflickr.com/65535/55187429095_81a324dec5_b.jpgKyoto by Felix, auf Flickr
https://live.staticflickr.com/65535/53602072430_0c454a8040_b.jpgSant'Andrea della Valle III by Felix, auf Flickr
And to answer my own question: I f*cked up the exposure on the...Show more →
No offense taken - it was a real photo, taken several years ago for a quick proof to a realtor. I also bracketed the exposure for the final deliverable.
Jun 21, 2026 at 11:04 AM
p.10 #13Utility of dynamic range beyond a certain point
Jack Flesher wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193Except that particular image is essentially irrelevant to showing an actual photographic need. I believe the point was directed to one extra stop of DR making a good image better, and here I have to agree that nobody’s shown us an example where it does, only the contrary. A lot of theory and measurbating attempting to support the claim, but no images where it actually makes a notable difference.
In theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
I think it is more akin to choosing a car for city driving between one that has room for 10 gallons in the tank, and another that can hold 20 gallons (1 stop difference 😊
. As it is the case with the dynamic range, one can reasonably argue that 10 gallons is enough for most practical reasons while driving in a city.
Similarly, it is arguably a valid point that the amount of noise in a full-frame image taken at the base ISO is already low and may not be visible for most practical reasons. Having said this, I admit that I strongly like the general idea of reducing noise (relative to signal).
I do believe that a good photographer should be able to produce good photography with a micro-four-thirds camera. My simple argument for (lately) using the GFX100S II with its extended DR and a range of excellent GF lenses is "why deny yourself some extra capacity" (and I am less interested in pursuing ever smaller and more compact camera systems)? Similarly, many A1 owners replaced this camera by the A1 II (perhaps because of the pre-capture) even if one can argue that the original A1 is already a perfectly capable camera.
The important thing is to be well-informed on the technical significance of the new features, like what the dynamic range means, and what it does in practice. The purchasing decisions are going to be different for different photographers but it helps to have these based on knowledge, rather than on advertisings or reviews from influencers.
snapsy wrote
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsIn theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
I think it is more akin to choosing a car for city driving between one that has room for 10 gallons in the tank, and another that can hold 20 gallons (1 stop difference 😊
. As it is the case with the dynamic range, one can reasonably argue that 10 gallons is enough for most practical reasons while driving in a city.Similarly, it is arguably a valid point that the amount of noise in a full-frame image taken at the base ISO is already low and may not be visible for most practical reasons. Having said this, I admit that I strongly like the general idea of reducing noise (relative to signal).
I do believe that a good photographer should be able to produce good photography with a micro-four-thirds camera. My simple argument for (lately) using the GFX100S II with its extended DR and a range of excellent GF lenses is "why deny yourself some extra capacity" (and I am less interested in pursuing ever smaller and more compact camera systems)? Similarly, many A1 owners replaced this camera by the A1 II (perhaps because of the pre-capture) even if one can argue that the original A1 is already a perfectly capable camera.
The important thing is to be well-informed on the technical significance of the new features, like what the dynamic range means, and what it does in practice. The purchasing decisions are going to be different for different photographers but it helps to have these based on knowledge, rather than on advertisings or reviews from influencers.
Jun 21, 2026 at 08:35 PM
p.10 #14Utility of dynamic range beyond a certain point
ruthenium wrote:
https://www.fredmiranda.com/forum/topic/1945915/5#17057193Except that particular image is essentially irrelevant to showing an actual photographic need. I believe the point was directed to one extra stop of DR making a good image better, and here I have to agree that nobody’s shown us an example where it does, only the contrary. A lot of theory and measurbating attempting to support the claim, but no images where it actually makes a notable difference.
In theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.I think it is more akin to choosing a car for city driving between one that has room for 10 gallons in the tank, and another that can hold 20 gallons (1 stop difference 😊. As it is the case with the dynamic range, one can reasonably argue that 10 gallons is enough for most practical reasons while driving in a city.
Similarly, it is arguably a valid point that the amount of noise in a full-frame image taken at the base ISO is already low and may not be visible for most practical reasons. Having said this, I admit that...Show more →
I shoot extreme macro 10-15x using FF no crops for magnifacation, if i shoot a 400 image stack using the A1 the A7r5 and the A7iv which camera do you think has the least noise and the most DR and the cleanest detail. 🤔
Jack Flesher wrote
snapsy wrote
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsIn theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
. As it is the case with the dynamic range, one can reasonably argue that 10 gallons is enough for most practical reasons while driving in a city.Similarly, it is arguably a valid point that the amount of noise in a full-frame image taken at the base ISO is already low and may not be visible for most practical reasons. Having said this, I admit that...Show more →
I shoot extreme macro 10-15x using FF no crops for magnifacation, if i shoot a 400 image stack using the A1 the A7r5 and the A7iv which camera do you think has the least noise and the most DR and the cleanest detail. 🤔
Jun 21, 2026 at 11:11 PM
p.10 #15Utility of dynamic range beyond a certain point
A74me wrote:
ruthenium wrote
Jack Flesher wrote
snapsy wrote
fjablo wrote
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still standsIn theory, more is better, but at a certain point it’s akin to giving you 15 gallons of gasoline when you only have room for 12.
. As it is the case with the dynamic range, one can reasonably argue that 10 gallons is enough for most practical reasons while driving in a city.Similarly, it is arguably a valid point that the amount of noise in a full-frame image taken at the base ISO is already low and may not be visible for most practical reasons. Having said this, I admit that I strongly like the general idea of reducing noise (relative to signal).
I do believe that a good photographer should be able to produce good photography with a micro-four-thirds camera. My simple argument for (lately) using the GFX100S II with its extended DR and a range of excellent GF lenses is "why deny yourself some extra capacity" (and I am less interested in pursuing ever smaller and more compact camera systems)? Similarly, many A1 owners replaced this camera by the A1 II (perhaps because of the pre-capture) even if one can argue that the original A1 is already a perfectly capable camera.
The important thing is to be well-informed on the technical significance of the new features, like what the dynamic range means, and what it does in practice. The purchasing decisions are going to be different for different photographers but it helps to have these based on knowledge, rather than on advertisings or reviews from influencers.
At what ISO (assuming the same on all three)?
I understand that your case (400 image stack, with 10-15 magnification without cropping) is rather special; nevertheless, a question of DR must come with the information about the ISO on the cameras.
Jun 21, 2026 at 11:25 PM
p.10 #16Utility of dynamic range beyond a certain point
fjablo wrote:
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands
Suppose you expose a scene so that no additional exposure adjustment is required in post-processing. In practice, this usually means placing the midtones around 18% grey.
If the highlights are clipping (indicated by 100% zebras), I would personally start using exposure bracketing so that multiple exposures can be stacked later to increase the effective dynamic range.
More generally, whenever an image can benefit from multi-image stacking, a sensor with greater dynamic range is also beneficial because it may eliminate the need for stacking altogether.
In situations where the image can benefit from multi-image stacking, but stacking may not be feasible due to subject movement in the foreground, higher sensor dynamic range becomes even more valuable.
Personally, I would like to see higher dynamic range at high ISOs.
At low ISOs, subjects are often amenable to multi-image stacking. At high ISOs, however, subjects are frequently moving, making these techniques impractical. Since dynamic range is already reduced at high ISOs, this only exacerbates the problem.
Lots of tech talk (and it is impressive technology), yet we still haven't seen a real-world example where this increased DR would come in handy, have we? I guess the original question still stands
Suppose you expose a scene so that no additional exposure adjustment is required in post-processing. In practice, this usually means placing the midtones around 18% grey.
If the highlights are clipping (indicated by 100% zebras), I would personally start using exposure bracketing so that multiple exposures can be stacked later to increase the effective dynamic range.
More generally, whenever an image can benefit from multi-image stacking, a sensor with greater dynamic range is also beneficial because it may eliminate the need for stacking altogether.
In situations where the image can benefit from multi-image stacking, but stacking may not be feasible due to subject movement in the foreground, higher sensor dynamic range becomes even more valuable.
Personally, I would like to see higher dynamic range at high ISOs.
At low ISOs, subjects are often amenable to multi-image stacking. At high ISOs, however, subjects are frequently moving, making these techniques impractical. Since dynamic range is already reduced at high ISOs, this only exacerbates the problem.
Jun 21, 2026 at 11:46 PM
p.10 #17Utility of dynamic range beyond a certain point
You lose three and a half stops of PDR going from base to ISO 1000 in the new a7r6. The 2013 a7r is down a half stop of PDR to the 2026 a7r6 at ISO 6400 - to 5.91 from 6.42.
Jun 22, 2026 at 12:04 AM
p.10 #18Utility of dynamic range beyond a certain point
philip_pj wrote:
You lose three and a half stops of PDR going from base to ISO 1000 in the new a7r6. The 2013 a7r is down a half stop of PDR to the 2026 a7r6 at ISO 6400 - to 5.91 from 6.42.
...and the 2015 A7RII has the same PDR as the A7RVI at high ISO. This suggests that the read noise has remained practically unchanged in the last five generations of the cameras.
Regarding "You lose three and a half stops of PDR going from base to ISO 1000" - this is perfectly natural, as increasing ISO from 100 to 1000 means a loss of 3 and 1/3 stops of light. The PDR is calculated from the signal-to-noise ratio, as has been noted already in this thread. The PDR changes linearly with the signal, and hence with the exposure and the exposure index (ISO).
Jun 22, 2026 at 03:02 AM
p.10 #19Utility of dynamic range beyond a certain point
the pixel surface area of the a7iv sensor is 10% more than the a7rv . measuring DR from noise is not how i would measure it. and quite frankly sensor size is not a measure to actuall pixel surface area / light collecting area. no one even mentions this why !
Jun 22, 2026 at 04:11 AM
p.10 #20Utility of dynamic range beyond a certain point
A74me wrote:
the pixel surface area of the a7iv sensor is 10% more than the a7rv . measuring DR from noise is not how i would measure it. and quite frankly sensor size is not a measure to actuall pixel surface area / light collecting area. no one even mentions this why !
Please don't mix pixel pitch i.e. distance between sensels centers, microlens area and actual sensel area ('cause CMOS always have some borders in between and it strongly depends on manufacturing process as well as cirtuits schematics etc.)
the pixel surface area of the a7iv sensor is 10% more than the a7rv . measuring DR from noise is not how i would measure it. and quite frankly sensor size is not a measure to actuall pixel surface area / light collecting area. no one even mentions this why !
Please don't mix pixel pitch i.e. distance between sensels centers, microlens area and actual sensel area ('cause CMOS always have some borders in between and it strongly depends on manufacturing process as well as cirtuits schematics etc.)
Jun 23, 2026 at 11:46 AM