What does dpreview latitude test really measure?

[Interceptor121]

The recent post from a forum member about the GH6 dynamic range reopens the point on the validity of the dpreview studio scene but more than anything also shows that there is no direct relationship between such test and dynamic range.

As I do not trust GH6 studio scene as this is linked to a beat of ACR I looked at four cameras that are already supported today. The G9 the GH5M2 the OM-1 and the EM1MKIII

In terms of DR all cameras come close but if we look at ranking they are ranked like this

1. G9
2. GH5M2
3. EM1MKIII
4. OM-1

The gap between top and bottom at ISO 200 is 1/3 Ev so nothing major.

Those are the cameras at 0 EV. The first thing you notice is that the OM-1 does not produce the same deep blacks of the other 3 cameras indicating a possible higher level of base gain.

When you look at the 3 cameras at +3 you are presented with the following situation and things are very much in check with the same trend continuing

At 6 Ev the situation takes a turn with the G9 a=and the EM1MKIII turning green. The OM-1 and GH5M2 seem to hold better

This is more apparent if you look at another patch

Results are more apparent at 6EV

So why is that the OM-1 and GH5M2 held better than others models that should have equal or better DR?

That is because this test does not show DR but only how a camera manages variations in black levels

black level range

It is no coincidence that the OM-1 and GH5M2 have a lower variations of blacks and therefore behave better. However this does not mean that in normal conditions those cameras will have better dynamic range.

Cameras with good behaviours with blacks are good for night photography assuming they do not have other issues due to dark current.

So for example I take my GH5M2 at night not my G9

Dpreview already drew wrong conclusions in their review of the S5

https://www.dpreview.com/reviews/panasonic-dc-s5-review/4

Our exposure latitude testshows no surprises; similar to the Panasonic S1, the S5 turns in a slightly better performance than the Sony a7 III, and easily surpasses the Canon EOS R6. In essence, this is an extremely capable sensor that will stand up to extreme shadow pushes in its Raw files.

But that is actually not true the S5 has only a better managements of black level range over the already outstanding A73 and is in fact a very good camera for night photography

So what is dpreview latitude test of the GH6 really telling us is that this camera at base ISO does not have a strong management of black levels as its predecessor from the GH range which I am sure most people fill find disappointing yet we are not going to use ISO 100 in low light so what is interesting is to see what does the camera do at higher ISO. But it is also true that conventional sensor do not get better with blacks as gain goes up so I am not that convinced there will be an improvement on this front.

--
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
Deer Photography workshops https://interceptor121.com/2021/09/26/2021-22-deer-photography-workshops-in-woburn/

 

[Cafe Racer]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

 

[Interceptor121]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows



This is the GH5M2 -3 EV corrected

This is the same file with shadows at 100

This is easier to see with the +1 EV

and 100 slides on the same without changing ev

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall. The higher you go more there is a possibility that issues with black level create false colours in the shadows



--
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
Deer Photography workshops https://interceptor121.com/2021/09/26/2021-22-deer-photography-workshops-in-woburn/

 

[Cafe Racer]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows

Yes it does.

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

Yeah, that's the whole point of what dpreview is demonstrating.

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

I don't know about you but I'm not expecting any magic, I expect a visual representation of the level of noise introduced.

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall.

Yes it's useful. When I'm street shooting a fleeting moment can mean a quick grab shot, which gives no time to consider exposure. In this case highlights and shadows may require adjustments in post.

 

[Interceptor121]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows

Yes it does.

Where exactly you see that?

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

Yeah, that's the whole point of what dpreview is demonstrating.

It is not demonstrating anything as I just proved to you with examples how the two things have vastly different impacts on the image

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

I don't know about you but I'm not expecting any magic, I expect a visual representation of the level of noise introduced.

You need to look elsewhere for example downloading a file at 0 or +1 and changing the slider and you will not see much difference in any of those cameras

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall.

Yes it's useful. When I'm street shooting a fleeting moment can mean a quick grab shot, which gives no time to consider exposure. In this case highlights and shadows may require adjustments in post.

It is totally pointless to do those tests if you want to manipulate a tone curve. For that you need large dynamic range to start with and increased tonal depth that a 12 bit sensor does not have

This is why MFT files 'break' sooner the variation between all those cameras in the example is minimal when you have exposed your image correctly

If you have not exposed the image correctly latitude tells you the amount of correction you do to the overall image NOT a shadow adjustment which is an alteration of the curve itself that alters how the sensor captured the image in the first place and hence is more problematic

 

[Cafe Racer]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows

Yes it does.

Where exactly you see that?

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

Yeah, that's the whole point of what dpreview is demonstrating.

It is not demonstrating anything as I just proved to you with examples how the two things have vastly different impacts on the image

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

I don't know about you but I'm not expecting any magic, I expect a visual representation of the level of noise introduced.

You need to look elsewhere for example downloading a file at 0 or +1 and changing the slider and you will not see much difference in any of those cameras

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall.

Yes it's useful. When I'm street shooting a fleeting moment can mean a quick grab shot, which gives no time to consider exposure. In this case highlights and shadows may require adjustments in post.

It is totally pointless to do those tests if you want to manipulate a tone curve. For that you need large dynamic range to start with and increased tonal depth that a 12 bit sensor does not have

This is why MFT files 'break' sooner the variation between all those cameras in the example is minimal when you have exposed your image correctly

If you have not exposed the image correctly latitude tells you the amount of correction you do to the overall image NOT a shadow adjustment which is an alteration of the curve itself that alters how the sensor captured the image in the first place and hence is more problematic

Go back and read my original post as you seem to be completely ignoring it.

What does dpreview latitude test really measure?

The recent post from a forum member about the GH6 dynamic range reopens the point on the validity of the dpreview studio scene but more than anything also shows that there is no direct relationship between such test and dynamic range. As I do not trust GH6 studio scene as this is linked to a...

www.dpreview.com

 

[Interceptor121]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows

Yes it does.

Where exactly you see that?

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

Yeah, that's the whole point of what dpreview is demonstrating.

It is not demonstrating anything as I just proved to you with examples how the two things have vastly different impacts on the image

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

I don't know about you but I'm not expecting any magic, I expect a visual representation of the level of noise introduced.

You need to look elsewhere for example downloading a file at 0 or +1 and changing the slider and you will not see much difference in any of those cameras

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall.

Yes it's useful. When I'm street shooting a fleeting moment can mean a quick grab shot, which gives no time to consider exposure. In this case highlights and shadows may require adjustments in post.

It is totally pointless to do those tests if you want to manipulate a tone curve. For that you need large dynamic range to start with and increased tonal depth that a 12 bit sensor does not have

This is why MFT files 'break' sooner the variation between all those cameras in the example is minimal when you have exposed your image correctly

If you have not exposed the image correctly latitude tells you the amount of correction you do to the overall image NOT a shadow adjustment which is an alteration of the curve itself that alters how the sensor captured the image in the first place and hence is more problematic

Go back and read my original post as you seem to be completely ignoring it.

https://www.dpreview.com/forums/post/66060033

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

I have read it and posted a long demonstration with images and screenshots of why what you say is not true. Which part are you missing or not understanding?

What does dpreview latitude test really measure?

The recent post from a forum member about the GH6 dynamic range reopens the point on the validity of the dpreview studio scene but more than anything also shows that there is no direct relationship between such test and dynamic range. As I do not trust GH6 studio scene as this is linked to a...

www.dpreview.com

is it clear the two things do not relate to each other as one is an overall bit multiplication and the other is an alteration of a tone curve so two operations totally different that will NOT produce similar results nor one can use to forecast the other

 

[Cafe Racer]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows

Yes it does.

Where exactly you see that?

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

Yeah, that's the whole point of what dpreview is demonstrating.

It is not demonstrating anything as I just proved to you with examples how the two things have vastly different impacts on the image

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

I don't know about you but I'm not expecting any magic, I expect a visual representation of the level of noise introduced.

You need to look elsewhere for example downloading a file at 0 or +1 and changing the slider and you will not see much difference in any of those cameras

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall.

Yes it's useful. When I'm street shooting a fleeting moment can mean a quick grab shot, which gives no time to consider exposure. In this case highlights and shadows may require adjustments in post.

It is totally pointless to do those tests if you want to manipulate a tone curve. For that you need large dynamic range to start with and increased tonal depth that a 12 bit sensor does not have

This is why MFT files 'break' sooner the variation between all those cameras in the example is minimal when you have exposed your image correctly

If you have not exposed the image correctly latitude tells you the amount of correction you do to the overall image NOT a shadow adjustment which is an alteration of the curve itself that alters how the sensor captured the image in the first place and hence is more problematic

Go back and read my original post as you seem to be completely ignoring it.

https://www.dpreview.com/forums/post/66060033

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

I have read it

And yet you've managed to ignore half of it and not understand the context!

and posted a long demonstration with images and screenshots of why what you say is not true.

Which part are you missing or not understanding?

https://www.dpreview.com/forums/post/66060075

is it clear the two things do not relate to each other as one is an overall bit multiplication and the other is an alteration of a tone curve so two operations totally different that will NOT produce similar results nor one can use to forecast the other

--
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
Deer Photography workshops https://interceptor121.com/2021/09/26/2021-22-deer-photography-workshops-in-woburn/

 

[Interceptor121]

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

it does not give you any representation of the level of noise when lifting shadows

Yes it does.

Where exactly you see that?

Generally artefacts that are generated by increasing shadows are higher than raising exposure because you are changing the overall curve in a different way than captured.

Yeah, that's the whole point of what dpreview is demonstrating.

It is not demonstrating anything as I just proved to you with examples how the two things have vastly different impacts on the image

So when you see an image with say +3 EV and expect the camera to do magic when you raise shadows it won't

I don't know about you but I'm not expecting any magic, I expect a visual representation of the level of noise introduced.

You need to look elsewhere for example downloading a file at 0 or +1 and changing the slider and you will not see much difference in any of those cameras

You capture a correctly exposed image with dark shadows as you did not want to clip and then increase shadows in post this may break anyway due to 'holes' in the tone curve.

Latitude is still a good thing if for some reason your exposure was off overall.

Yes it's useful. When I'm street shooting a fleeting moment can mean a quick grab shot, which gives no time to consider exposure. In this case highlights and shadows may require adjustments in post.

It is totally pointless to do those tests if you want to manipulate a tone curve. For that you need large dynamic range to start with and increased tonal depth that a 12 bit sensor does not have

This is why MFT files 'break' sooner the variation between all those cameras in the example is minimal when you have exposed your image correctly

If you have not exposed the image correctly latitude tells you the amount of correction you do to the overall image NOT a shadow adjustment which is an alteration of the curve itself that alters how the sensor captured the image in the first place and hence is more problematic

Go back and read my original post as you seem to be completely ignoring it.

https://www.dpreview.com/forums/post/66060033

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

I have read it

And yet you've managed to ignore half of it and not understand the context!

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software

It does not do it

What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness.

A latitude test on the other side i.e. overexposed cannot give that either for the same reasons

In essence a latitude test only tells you what happens when you move the exposure slider nothing more nothing less. Doing adjustments of the tone curve breaks raw files because the adjustment is not linear. For it to work better you need higher bit depth and ability of the camera to produce a wider tonal depth none of those are tested in this dpreview studio scene

In essence the studio scene is only good to have a visual representation of the image that is produced with camera defaults it is not a proxy for DR nor for noise or anything else. In order to do that you need to perform different measurements and use different tools for example use the kodak step chart connected to measuring equipment for DR and SNR. This is why their conclusions are off what dxomark of photonstophotos produce and should not be trusted

and posted a long demonstration with images and screenshots of why what you say is not true.

Which part are you missing or not understanding?

https://www.dpreview.com/forums/post/66060075

is it clear the two things do not relate to each other as one is an overall bit multiplication and the other is an alteration of a tone curve so two operations totally different that will NOT produce similar results nor one can use to forecast the other

--
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
Deer Photography workshops https://interceptor121.com/2021/09/26/2021-22-deer-photography-workshops-in-woburn/

 

[WileEC_ID]

I think the usefulness of these shots is limited AND only for comparison - as they are, for one system to another. It is a very rudimentary comparison - solid entrance in the better than nothing category.

Drawing conclusions about how far they can be pushed, How much dynamic range, etc. has too many variables for one to make any of those kinds of suppositions from a single studio setup. And unless you shoot in these exact conditions - it is practically useless for how you shoot.

Personally, I find it best to take my own shots - and do - in real light, not neutral studio light. And, if I want to explore the times when I take a bad exposure, which I put on me - regardless of the situation*, then I take some shots slanted that direction, then evaluate them with my software and skill. Trying to use someone's studio, relatively flat (dimensionally) images and draw conclusions is, in my opinion and experience not very usable.

*I read so often of people excusing their bad exposures, so the need to push files later. That is just a load of dingo kidneys, 99% of the time. We are in a situation and know the kind of shot we want, or are looking for - if we are not ready for it, that is on us. Certainly, after the fact we can see where a third or half an EV tweak or white balance tweak is prefered - that is different - that is about what we want verses what the camera sees.

 

[Cafe Racer]

I have read it

And yet you've managed to ignore half of it and not understand the context!

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software

It does not do it

Are you aware that a visual representation is not the same as a measurement!

What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness.

A latitude test on the other side i.e. overexposed cannot give that either for the same reasons

It can be used to give a visual representation when comparing the ability of raw converter's to recover details in bright parts of an image where detail is not visible prior to adjustment. I.e. Which raw converter performs the best on the same image.

In essence a latitude test only tells you what happens when you move the exposure slider nothing more nothing less.

This comment clearly demonstrates that you don't understand my original post as that's what I'm suggesting, which is why I said "using Adobe software." I'm clearly talking about software adjustments, which involves moving sliders!

Many people process images using software and they get a visual representation of what happens when they move sliders! Have you ever tried it?!

Doing adjustments of the tone curve breaks raw files because the adjustment is not linear. For it to work better you need higher bit depth and ability of the camera to produce a wider tonal depth none of those are tested in this dpreview studio scene

In essence the studio scene is only good to have a visual representation of the image that is produced with camera defaults it is not a proxy for DR nor for noise or anything else.

VISUAL REPRESENTATION - that's what I said in case you missed it!

In order to do that you need to perform different measurements

I'll say it again, a visual representation is not the same as a measurement.

and use different tools for example use the kodak step chart connected to measuring equipment for DR and SNR. This is why their conclusions are off what dxomark of photonstophotos produce and should not be trusted

 

[Interceptor121]

I have read it

And yet you've managed to ignore half of it and not understand the context!

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software

It does not do it

Are you aware that a visual representation is not the same as a measurement!

What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness.

A latitude test on the other side i.e. overexposed cannot give that either for the same reasons

It can be used to give a visual representation when comparing the ability of raw converter's to recover details in bright parts of an image where detail is not visible prior to adjustment. I.e. Which raw converter performs the best on the same image.

In essence a latitude test only tells you what happens when you move the exposure slider nothing more nothing less.

This comment clearly demonstrates that you don't understand my original post as that's what I'm suggesting, which is why I said "using Adobe software." I'm clearly talking about software adjustments, which involves moving sliders!

Many people process images using software and they get a visual representation of what happens when they move sliders! Have you ever tried it?!

Doing adjustments of the tone curve breaks raw files because the adjustment is not linear. For it to work better you need higher bit depth and ability of the camera to produce a wider tonal depth none of those are tested in this dpreview studio scene

In essence the studio scene is only good to have a visual representation of the image that is produced with camera defaults it is not a proxy for DR nor for noise or anything else.

VISUAL REPRESENTATION - that's what I said in case you missed it!

In order to do that you need to perform different measurements

I'll say it again, a visual representation is not the same as a measurement.

and use different tools for example use the kodak step chart connected to measuring equipment for DR and SNR. This is why their conclusions are off what dxomark of photonstophotos produce and should not be trusted

It is not a visual nor factual representation of you moving the slider because shifting exposure and moving a slider for shadows or highlights are two different things

There is essentially no direct correlation between that test and noise measurements in any part of the image. This is why when something comes up it generates just endless subjective discussions. They could at least do a full latitude test which would be more useful and leave observation on noise and dynamic range alone as they are not equipped to measure them or to give visual indications

 

[Interceptor121]

Here are 6 images with the GH5M2 the camera that has the most latitude according to this dpreview test

This is the -5 under corrected in post. it looks pretty green

This is the 4ev under

This is the 3ev under

the 2ev under

The 1ev under

and finally the 1ev under with 100% shadows slider

A quick 100% look at the bottom down shows that the adjustment of the 100% shadows slider does not look similar to any of the 5/4/3/2/1 and actually looks worse so none of those can be taken as proxy

Why is that? Because while exposure adjustment is a linear operator and although there is loss that affects mostly the dark tones this is then propagated everywhere the shadow adjustment only works on one part of the tones so darks get shifted into midtones area creating crowding and generating a result that overall looks worse even of more severe under-exposures

So the argument I frequently use shadows at 50 so a camera with 4 stops latitude will be better is not a valid one. You can easily destroy tones in a correctly exposed image if the camera does not produce a sufficient tonal range by moving a slider

I found that a good proxy of the impact of specific slider on an image is DxOMark Tonal Range when combined with dynamic range. If your image is clipped on darks or lights or both recovery of tones will generate issues no matter what you do

--

instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
Deer Photography workshops https://interceptor121.com/2021/09/26/2021-22-deer-photography-workshops-in-woburn/

 

[Cafe Racer]

I have read it

And yet you've managed to ignore half of it and not understand the context!

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software

It does not do it

Are you aware that a visual representation is not the same as a measurement!

What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness.

A latitude test on the other side i.e. overexposed cannot give that either for the same reasons

It can be used to give a visual representation when comparing the ability of raw converter's to recover details in bright parts of an image where detail is not visible prior to adjustment. I.e. Which raw converter performs the best on the same image.

In essence a latitude test only tells you what happens when you move the exposure slider nothing more nothing less.

This comment clearly demonstrates that you don't understand my original post as that's what I'm suggesting, which is why I said "using Adobe software." I'm clearly talking about software adjustments, which involves moving sliders!

Many people process images using software and they get a visual representation of what happens when they move sliders! Have you ever tried it?!

Doing adjustments of the tone curve breaks raw files because the adjustment is not linear. For it to work better you need higher bit depth and ability of the camera to produce a wider tonal depth none of those are tested in this dpreview studio scene

In essence the studio scene is only good to have a visual representation of the image that is produced with camera defaults it is not a proxy for DR nor for noise or anything else.

VISUAL REPRESENTATION - that's what I said in case you missed it!

In order to do that you need to perform different measurements

I'll say it again, a visual representation is not the same as a measurement.

and use different tools for example use the kodak step chart connected to measuring equipment for DR and SNR. This is why their conclusions are off what dxomark of photonstophotos produce and should not be trusted

It is not a visual nor factual representation of you moving the slider because shifting exposure and moving a slider for shadows or highlights are two different things

Now you're setting up a straw man argument, trying to saddle me with something I didn't say.

I said nothing of 'shifting exposure' because I'm fully aware that exposure is at the image capture stage and not the raw development stage.

If you don't like dpreview calling their test 'exposure latitude', then take it up with them.

There is essentially no direct correlation between that test and noise measurements in any part of the image.

Another straw man! For the umpteenth time, a visual representation is NOT the same as a measurement.

This is why when something comes up it generates just endless subjective discussions. They could at least do a full latitude test which would be more useful and leave observation on noise and dynamic range alone as they are not equipped to measure them or to give visual indications

 

[Interceptor121]

I have read it

And yet you've managed to ignore half of it and not understand the context!

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software

It does not do it

Are you aware that a visual representation is not the same as a measurement!

What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness.

A latitude test on the other side i.e. overexposed cannot give that either for the same reasons

It can be used to give a visual representation when comparing the ability of raw converter's to recover details in bright parts of an image where detail is not visible prior to adjustment. I.e. Which raw converter performs the best on the same image.

In essence a latitude test only tells you what happens when you move the exposure slider nothing more nothing less.

This comment clearly demonstrates that you don't understand my original post as that's what I'm suggesting, which is why I said "using Adobe software." I'm clearly talking about software adjustments, which involves moving sliders!

Many people process images using software and they get a visual representation of what happens when they move sliders! Have you ever tried it?!

Doing adjustments of the tone curve breaks raw files because the adjustment is not linear. For it to work better you need higher bit depth and ability of the camera to produce a wider tonal depth none of those are tested in this dpreview studio scene

In essence the studio scene is only good to have a visual representation of the image that is produced with camera defaults it is not a proxy for DR nor for noise or anything else.

VISUAL REPRESENTATION - that's what I said in case you missed it!

In order to do that you need to perform different measurements

I'll say it again, a visual representation is not the same as a measurement.

and use different tools for example use the kodak step chart connected to measuring equipment for DR and SNR. This is why their conclusions are off what dxomark of photonstophotos produce and should not be trusted

It is not a visual nor factual representation of you moving the slider because shifting exposure and moving a slider for shadows or highlights are two different things

Now you're setting up a straw man argument, trying to saddle me with something I didn't say.

I said nothing of 'shifting exposure' because I'm fully aware that exposure is at the image capture stage and not the raw development stage.

If you don't like dpreview calling their test 'exposure latitude', then take it up with them.

it is half c*** exposure latitude as it is not complete and as such not really worthy of consideration like their ISO invariance that is also flawed

There is essentially no direct correlation between that test and noise measurements in any part of the image.

Another straw man! For the umpteenth time, a visual representation is NOT the same as a measurement.

This is why when something comes up it generates just endless subjective discussions. They could at least do a full latitude test which would be more useful and leave observation on noise and dynamic range alone as they are not equipped to measure them or to give visual indications

it visually looks different see images in the main thread

 

[ChrisFB]

https://www.dpreview.com/forums/post/66060033

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

I have read it and posted a long demonstration with images and screenshots of why what you say is not true. Which part are you missing or not understanding?

https://www.dpreview.com/forums/post/66060075

is it clear the two things do not relate to each other as one is an overall bit multiplication and the other is an alteration of a tone curve so two operations totally different that will NOT produce similar results nor one can use to forecast the other

I agree with most of what you said Interceptor - they are not the same. But of course, pushing the global exposure and seeing how the sensor behaves gives SOME prediction of how well the sensor/raw file handles pushing the shadows. Even if not a linear prediction. Besides that, the latitude test isn't only about pushing the shadows. Understanding how well you can compensate the exposure in post is a useful indication too (although more for shots that were accidentally badly exposed).

If the GH6 handles global exposure pushes badly, you can proxy/predict that pushing the shadows will not suddenly go much better than in other sensors. Probably worse too. So while no linear prediction, of course it says something. Even if I agree they are very different things.

I predict that sensors that handle the exposure push (the dpreview test) very well (/better than other cameras), also handle a shadow push well (/better than other cameras).

 

[Interceptor121]

https://www.dpreview.com/forums/post/66060033

Dpreview's exposure latitude test gives me a visual representation of the level of noise I can expect when lifting shadows at a camera's given 'base ISO' using Adobe software: assuming dpreview haven't made the mistake they made with the G90. What it doesn't give is a visual representation of highlight recovery, which limits it's usefulness. I find the ability to discern detail in highlights is largely dependant on the raw converter used, but shadow noise can be represented differently too. That being said I do like visual representations in the form of photographs, even if it's limited in its scope. I just take the limitations into consideration.

I have read it and posted a long demonstration with images and screenshots of why what you say is not true. Which part are you missing or not understanding?

https://www.dpreview.com/forums/post/66060075

is it clear the two things do not relate to each other as one is an overall bit multiplication and the other is an alteration of a tone curve so two operations totally different that will NOT produce similar results nor one can use to forecast the other

I agree with most of what you said Interceptor - they are not the same. But of course, pushing the global exposure and seeing how the sensor behaves gives SOME prediction of how well the sensor/raw file handles pushing the shadows. Even if not a linear prediction. Besides that, the latitude test isn't only about pushing the shadows. Understanding how well you can compensate the exposure in post is a useful indication too (although more for shots that were accidentally badly exposed).

If the GH6 handles global exposure pushes badly, you can proxy/predict that pushing the shadows will not suddenly go much better than in other sensors. Probably worse too. So while no linear prediction, of course it says something. Even if I agree they are very different things.
I predict that sensors that handle the exposure push (the dpreview test) very well (/better than other cameras), also handle a shadow push well (/better than other cameras).

Not really as you can see from the examples in the main thread changing the shadow to 100 is much kore invasive than changing exposure which is what latitude shows

The GH6 may as well have issues of all sorts but the whole train of thought +6 stretch means less noise and more dynamic range is flawed

generally cameras with good blacks perform well in low light however a deep shadow in an overall bright scene is not the same as a real low light situation

what latitude shows is what it says on the tin how the whole image holds an error in exposure it doesn’t forecast what the shadow slider does but does tell you how the camera will do if you shoot at night