Z7ii vs Z8/9 Dynamic Range - A bit deeper

[JoelKlein]

Good morning.
I would like to understand from a technical point, a more deeper understanding about the loss of 1/3th of a stop in DR up to ISO 500.

Let’s say I’m trying to capture a sunset.
I’m at ISO 64 to get the best DR.
Where would the Z8/9 DR play a difference vs the Z7ii? The highlights on the Z8/9 would clip 1/3th before the Z7ii? Which means I would have to stop down the exposure on the Z8/9 with 1/3th? Or is it in the shadows zone?

Thank you.

P.s.
I already ordered the Z8 and be selling the Z7ii to a friend. I want to understand the difference.

 

[John Navitsky]

i don’t mean this as an answer, but i think until we see revolutionary changes in sensor design, differences between dynamic ranges between sensors is mostly something we can see/measure in “the lab” or on paper.

i think if someone showed 99.9% of people real images from sensors being compared, they wouldn’t be able to tell which is which

 

[Nimi]

Good morning.
I would like to understand from a technical point, a more deeper understanding about the loss of 1/3th of a stop in DR up to ISO 500.

Let’s say I’m trying to capture a sunset.
I’m at ISO 64 to get the best DR.
Where would the Z8/9 DR play a difference vs the Z7ii? The highlights on the Z8/9 would clip 1/3th before the Z7ii? Which means I would have to stop down the exposure on the Z8/9 with 1/3th? Or is it in the shadows zone?

Thank you.

P.s.
I already ordered the Z8 and be selling the Z7ii to a friend. I want to understand the difference.

1/3 of a stop, even if real, is meaningless. As long as you're above 10 or so for stills and have the color bit depth, you're in golden territory almost regardless of how the image is viewed.

As mentioned above, with dual gain sensors and other techniques it will be continuously improved. For reference, the highest-end cine camera (inarguably the Arri Alexa) is probably around 16 stops.

 

[DRwyoming]

Or is it in the shadows zone?

Dynamic range differences show up in the shadows and specifically show up as higher noise in the shadows. IOW,. dynamic range quantifies the range between the brightest brights and some defined signal to noise ratio, the brights don't get any brighter and camera settings for proper exposure don't change for different dynamic range cameras but how much noise is visible in the deeper shadows does change.

So if you look at a dynamic range chart for the Z9 vs the Z7 II you'll see that yes there's roughly a 1/3 stop difference between them at say ISO 100 and virtually no difference between them at high ISO settings. But that difference at ISO 100 is the difference between roughly 11 stops of DR on the Z7 II to roughly 10.6 stops on the Z9. In either case that's a massive dynamic range and you'd be hard pressed to pick up that difference in shadow noise even though it's measurable in the lab.

Another way to look at this is most wildlife photographers wouldn't hesitate to shoot either a Z9 or Z7 II at ISO 1000 or even quite a bit higher. At ISO 1000, both cameras have a DR of 8.49 stops and those images aren't captured with excessive shadow noise. At ISO 100 either camera has at least an additional stop and a half or more DR which is quite a bit.

 

[JoelKlein]

Thank you @DRwyoming
Let me repeat to make sure I understood it.

I’m photographing a landscape, ISO64 f/8-/16 in post I’m lifting the shadows.
The Z8/9 would produce a tiny bit more noise at the darkest area lifted?
How much to I have to open the shadows before I see the noise?

 

[DRwyoming]

Thank you @DRwyoming
Let me repeat to make sure I understood it.

I’m photographing a landscape, ISO64 f/8-/16 in post I’m lifting the shadows.
The Z8/9 would produce a tiny bit more noise at the darkest area lifted?
How much to I have to open the shadows before I see the noise?

It's not really about lifting the shadows to see noise though that could happen.

The DR curve basically says the noise in shadows in some number of stops below bright white (255, 255, 255 in 8 bit RGB representation). So you have some defined signal to noise ratio (SNR = 20 represents the noise threshold that sets the lower end of the DR curve for the photonstophotos curves).

In practical terms if you have a DR of 10 stops on these curves there would be 10 stops of captured detail between the brightest brights and the shadows where noise starts to become noticeable. That's a massive range of tonal values that are holding detail before you get so deep into the darkest area of your images where noise starts to become noticeable.

But sure if you start pulling up the darker shadows you'll cut into that range further as you selectively raise the brightness of the darker shadow areas. So maybe if you start with a 10 stop dynamic range and do two stops of shadow pulling you'd have an effective DR of roughly 8 stops between the brightest whites to where shadow noise becomes noticeable. But that's still a huge range as most photographic print processes can't capture more than about 5 to 6 stops of dynamic range.

How much DR you 'need' also depends a lot on the tones in your scene and the lighting. On a soft light day (e.g. high overcast or maybe shooting in the shadows) the scene itself might only have 3 to 5 stops of total dynamic range depending on the various colors and tones in the scene. But on a bright sunny day shooting one of the worst case scenarios, a bride in a bright white silk wedding dress standing next to a groom in a black tuxedo the inherent dynamic range of the scene can be a lot bigger and if you want detail in both the dress and the tux without excessive noise you'll want more dynamic range in the camera. But even then 8 plus stops of dynamic range can typically capture a scene like that if the image was well exposed at the time of capture. If you miss the exposure in the field and have to start raising the exposure or dramatically pulling the shadows then yeah, a bit more dynamic range wouldn't hurt.

 

[Cameron T]

It's not really about lifting the shadows to see noise though that could happen.

The DR curve basically says the noise in shadows in some number of stops below bright white (255, 255, 255 in 8 bit RGB representation). So you have some defined signal to noise ratio (SNR = 20 represents the noise threshold that sets the lower end of the DR curve for the photonstophotos curves).

In practical terms if you have a DR of 10 stops on these curves there would be 10 stops of captured detail between the brightest brights and the shadows where noise starts to become noticeable. That's a massive range of tonal values that are holding detail before you get so deep into the darkest area of your images where noise starts to become noticeable.

But sure if you start pulling up the darker shadows you'll cut into that range further as you selectively raise the brightness of the darker shadow areas. So maybe if you start with a 10 stop dynamic range and do two stops of shadow pulling you'd have an effective DR of roughly 8 stops between the brightest whites to where shadow noise becomes noticeable. But that's still a huge range as most photographic print processes can't capture more than about 5 to 6 stops of dynamic range.

How much DR you 'need' also depends a lot on the tones in your scene and the lighting. On a soft light day (e.g. high overcast or maybe shooting in the shadows) the scene itself might only have 3 to 5 stops of total dynamic range depending on the various colors and tones in the scene. But on a bright sunny day shooting one of the worst case scenarios, a bride in a bright white silk wedding dress standing next to a groom in a black tuxedo the inherent dynamic range of the scene can be a lot bigger and if you want detail in both the dress and the tux without excessive noise you'll want more dynamic range in the camera. But even then 8 plus stops of dynamic range can typically capture a scene like that if the image was well exposed at the time of capture. If you miss the exposure in the field and have to start raising the exposure or dramatically pulling the shadows then yeah, a bit more dynamic range wouldn't hurt.

But if one is serious enough to actually need every bit of dr, it's better to just get a medium format camera with even more dr. A third of a stop isn't noticeable, especially with multiple exposures and blending them.

 

[JoPoV]

on a bright sunny day shooting one of the worst case scenarios, a bride in a bright white silk wedding dress standing next to a groom in a black tuxedo the inherent dynamic range of the scene can be a lot bigger and if you want detail in both the dress and the tux without excessive noise you'll want more dynamic range in the camera. But even then 8 plus stops of dynamic range can typically capture a scene like that if the image was well exposed at the time of capture

So, if I understand well you say a 10 dynamic range sensor is able to capture any scene with only reflective lighting without any noise at base iso if well exposed ?

And what about landscapes with direct lighting inframe (means sunset for example) - as Joel maybe asked ?

I’m photographing a landscape, ISO64 f/8-/16 in post I’m lifting the shadows.

 

[Cameron T]

So, if I understand well you say a 10 dynamic range sensor is able to capture any scene with only reflective lighting without any noise at base iso if well exposed ?

And what about landscapes with direct lighting inframe (means sunset for example) - as Joel maybe asked ?

It depends on the exact scene, but still possible to do. Your other options are to bracket, or go to medium/large format.

The difference between ~11.3 and ~11.6 stops at ISO 64 is not noticeable by anyone under normal conditions, and this falls under spec sheet anxiety.

 

[JoPoV]

0.3 IL is not that much, of course.
1 is twice the amount of light.

 

[JoelKlein]

It's not really about lifting the shadows to see noise though that could happen.

The DR curve basically says the noise in shadows in some number of stops below bright white (255, 255, 255 in 8 bit RGB representation). So you have some defined signal to noise ratio (SNR = 20 represents the noise threshold that sets the lower end of the DR curve for the photonstophotos curves).

In practical terms if you have a DR of 10 stops on these curves there would be 10 stops of captured detail between the brightest brights and the shadows where noise starts to become noticeable. That's a massive range of tonal values that are holding detail before you get so deep into the darkest area of your images where noise starts to become noticeable.

But sure if you start pulling up the darker shadows you'll cut into that range further as you selectively raise the brightness of the darker shadow areas. So maybe if you start with a 10 stop dynamic range and do two stops of shadow pulling you'd have an effective DR of roughly 8 stops between the brightest whites to where shadow noise becomes noticeable. But that's still a huge range as most photographic print processes can't capture more than about 5 to 6 stops of dynamic range.

How much DR you 'need' also depends a lot on the tones in your scene and the lighting. On a soft light day (e.g. high overcast or maybe shooting in the shadows) the scene itself might only have 3 to 5 stops of total dynamic range depending on the various colors and tones in the scene. But on a bright sunny day shooting one of the worst case scenarios, a bride in a bright white silk wedding dress standing next to a groom in a black tuxedo the inherent dynamic range of the scene can be a lot bigger and if you want detail in both the dress and the tux without excessive noise you'll want more dynamic range in the camera. But even then 8 plus stops of dynamic range can typically capture a scene like that if the image was well exposed at the time of capture. If you miss the exposure in the field and have to start raising the exposure or dramatically pulling the shadows then yeah, a bit more dynamic range wouldn't hurt.

Thank you for your detailed explanation.
I love sunset (who doesn’t?!) and exposing for the yellow/orange sun, resulting in the grass, trees in front pretty dark. So in ACR I mask the bottom and move the shadows slider almost all the way to the right.

 

[DRwyoming]

Thank you for your detailed explanation.
I love sunset (who doesn’t?!) and exposing for the yellow/orange sun, resulting in the grass, trees in front pretty dark. So in ACR I mask the bottom and move the shadows slider almost all the way to the right.

Yup, that approach to processing benefits from a lot of dynamic range but still you should be fine shooting the Z9/Z8 or Z7II in those situations as long as you keep the ISO below 800 or so.

If you use a graduated ND filter or shot an exposure bracketed pair of images, one exposed for the sky and one exposed for the foreground and combine them in post you could get by with even less DR.

Lots of ways to approach these kinds of HDR scenes but even going with a single exposure and hard shadow pulls, 8 to 10 stops of dynamic range is typically more than enough.

 

[FB101]

Thank you for your detailed explanation.
I love sunset (who doesn’t?!) and exposing for the yellow/orange sun, resulting in the grass, trees in front pretty dark. So in ACR I mask the bottom and move the shadows slider almost all the way to the right.

If you are shooting on a tripod, bracket 2 or 3 shots and blend in post. If you shoot handheld, a graduated neutral filter is exactly what you need.
If you shoot 1 exposure only, without filter, this is exactly the case where you shoot exposure to the right (as high as you can before clipping anything) and then develop the file 2 ways to blend in post - that’s not as good a solution because you are limited by the max DR of the sensor, but that’s how you’ll squeeze most DR out of it.
Remember when exposing to the right that the histogram you’ll see after exposing uses the jpg profile, so use the flattest looking jpg mode you can find (whatever that is in Nikon) to get a better approximation of the “raw” histogram and check if you properly exposed to the right.

 

[JoelKlein]

Most of the time when I see that sunset it’s changing and gone in minutes. I may be in different places with different lenses available. I have a 10 stop ND, but not a graduated. I would need to invest in a slide in filter system to accommodate most of my lenses. Snapping another exposure for the bottom half then HDR it, should be ideal, but it hasn’t given me the results I’m after, it looks fake. But that might be me missing the technique. Pushing the slider all the way is easier… and it looks like one natural exposure