Sony Histograms, Zebras and Highlight warnings Sony A1 experience

Interceptor121

Forum Pro
Messages
12,604
Solutions
8
Reaction score
9,603
Today I had time to do some tests in my light box. Apologies for the soft images the lens is super close and the edges blur a lot.

I wanted to stress the validity of the Zebra +109 highlight warning as well as the highlight warning in playback

WIth controlled light I took several shots as follows in multimeter with creative look Standard no picture profile

The results were as follows, as common factor ALL shots shows highlight warning on the camera playback.

I then imported the RAW and JPEG in lightroom and applied the color style ST to try and match the camera.

First observation adobe lightroom images, net of color differences are 0.25 ev darrker than camera jpeg. This is consistent with DxOmark observed ISO delta of 0.5 while adobe is applying a 0.25 correction.

First sample an image taken so that the zebra would not show at all at +109. Multi meter reading -0.3 ev



SOOC
SOOC





RAW
RAW

You will notice that distortion correction is not consistent either but thats another story

Here the same with adjustment of +0.25 ev which brings it closer to the SOOC



RAW +0.25
RAW +0.25

You can see there is no clipping in either despite the camera is already issuing highlight warning in playback

Second set is the image with meter at 0 Ev



SOOC 0 Ev
SOOC 0 Ev

You can see the small highlight warning which showed as very little zebra in the EVF



RAW 0 Ev +0.25 ev
RAW 0 Ev +0.25 ev

The RAW file however even with the additional 0.25 Ev is not clipping

I now increase exposure until I have a bit more zebra on the white bar



SOOC 0.3 Ev zebra on the white bar
SOOC 0.3 Ev zebra on the white bar

The raw file still not clipping

RAW filw 0.3 Ev +0.25
RAW filw 0.3 Ev +0.25

At +2/3 EV I get lots of zebras and clipping in the jpeg



SOOC 2/3 Ev lots of zebra
SOOC 2/3 Ev lots of zebra

There is now a little clipping on the RAW but only if I apply the 0.25 ev adjustment



RAW 2/3 Ev +0.25
RAW 2/3 Ev +0.25

Here a shot where the whole screen was full of zebra stripes



SOOC +1 Ev
SOOC +1 Ev

and corresponding RAW showing some more clipping



RAW +1Ev with 0.25 applied
RAW +1Ev with 0.25 applied



If you are familiar with baseline exposure correction you know that adobe increase the brightness of raw files of 1 stop and increase constrast lowering the black point it also applies a curve. If you reset those adjustments you see that even the +1 Ev file was not clipped



LIghtroom reset of presets and camera baseline
LIghtroom reset of presets and camera baseline

In summary if you are shooting a scene with black and white like the test chart here that has some dynamic range and you use zebra not to clip you may be over 1 stop below the clipping point.

This creates some challenges for those users that believe in ETTR as clearly the camera metering system is not well set and gives false positives.

In this instance a bracketing of 3 shots -1 0 +1 would have made the +1 shot the favourite and therefore lacking any better camera tools euristic approaches are preferrable

Other camera may exhibit different behaviours as manufaturers and software setting differs

--
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/
If you want to get in touch don't send me a PM rather contact me directly at my website/social media
 
what sensitivity did you set the zebras at ?
As per post 109+
thats not sensitivity.
Zebra has not sensitivity it has levels and so does peaking on my A1

The level for low is 109+ there is nothing higher

If you refer to exact zebra values there is a range but this is to nail a specific exposure i.e. skin at 65% +-5 not relevant for clipping monitoring where you want low limit that has no range
its the same as adjusting focus peaking for a more precise measurement.
it does not exist for highlight clipping on the A1 and I think on most models

Zebra is based on luminance there is not such a thing to sensitivity a pixel is or isn't a certain RGB value
i set custom to 109+ range +1 is the most accurate for raw files ,your a1 has the same control.
 
I love using zebra’s compared to using the histogram. For fast-paced shooting, it is far faster and easier… and you get to see exactly what in the image is getting clipped.

Since both zebra’s and histogram’s use JPG’s, I have learned, over the years, how much I can push each one, while maximizing dynamic range and avoiding clipping. I accept light zebra effects with no penalty. After awhile, you learn how much you can push the envelope.

--

Jeff
Florida, USA
http://www.gr8photography.com
 
Last edited:
what sensitivity did you set the zebras at ?
As per post 109+
thats not sensitivity.
Zebra has not sensitivity it has levels and so does peaking on my A1

The level for low is 109+ there is nothing higher

If you refer to exact zebra values there is a range but this is to nail a specific exposure i.e. skin at 65% +-5 not relevant for clipping monitoring where you want low limit that has no range
its the same as adjusting focus peaking for a more precise measurement.
it does not exist for highlight clipping on the A1 and I think on most models

Zebra is based on luminance there is not such a thing to sensitivity a pixel is or isn't a certain RGB value
i set custom to 109+ range +1 is the most accurate for raw files ,your a1 has the same control.
There is no range in the lower limit

If you use the medium and say +1 -1 it does not show 110 it only shows 108 and becomes the same as setting 108+ so you may as well just set 108+ lower limit
 
what sensitivity did you set the zebras at ?
As per post 109+
thats not sensitivity.
Zebra has not sensitivity it has levels and so does peaking on my A1

The level for low is 109+ there is nothing higher

If you refer to exact zebra values there is a range but this is to nail a specific exposure i.e. skin at 65% +-5 not relevant for clipping monitoring where you want low limit that has no range
its the same as adjusting focus peaking for a more precise measurement.
it does not exist for highlight clipping on the A1 and I think on most models

Zebra is based on luminance there is not such a thing to sensitivity a pixel is or isn't a certain RGB value
i set custom to 109+ range +1 is the most accurate for raw files ,your a1 has the same control.
There is no range in the lower limit

If you use the medium and say +1 -1 it does not show 110 it only shows 108 and becomes the same as setting 108+ so you may as well just set 108+ lower limit
Zebra only relies on green channel. It's based on raw, but only based on green primaries. Histogram on the other hand, is not your friend as well. It's based on JPEG.

If you have a bright red or bright blue it sometimes would not correctly show zebras on 109+. This might cause washed sky, and other unwanted over exposure artefacts.

So, there's really no way to do accurate ETTR on Sony cameras. Maybe some exposure bracketing can help.
 
what sensitivity did you set the zebras at ?
As per post 109+
thats not sensitivity.
Zebra has not sensitivity it has levels and so does peaking on my A1

The level for low is 109+ there is nothing higher

If you refer to exact zebra values there is a range but this is to nail a specific exposure i.e. skin at 65% +-5 not relevant for clipping monitoring where you want low limit that has no range
its the same as adjusting focus peaking for a more precise measurement.
it does not exist for highlight clipping on the A1 and I think on most models

Zebra is based on luminance there is not such a thing to sensitivity a pixel is or isn't a certain RGB value
i set custom to 109+ range +1 is the most accurate for raw files ,your a1 has the same control.
There is no range in the lower limit

If you use the medium and say +1 -1 it does not show 110 it only shows 108 and becomes the same as setting 108+ so you may as well just set 108+ lower limit
Zebra only relies on green channel. It's based on raw, but only based on green primaries. Histogram on the other hand, is not your friend as well. It's based on JPEG.

If you have a bright red or bright blue it sometimes would not correctly show zebras on 109+. This might cause washed sky, and other unwanted over exposure artefacts.

So, there's really no way to do accurate ETTR on Sony cameras. Maybe some exposure bracketing can help.
I don't think that is accurate because Zebra come out on screen and therefore will be based on luminance that with appropriate coefficients maps all RGB channels

The reason why bright red and blue do not saturate the zebra is because there are twice green than there are red and blu so even if you add a clipped red the zebra does not clip but this is an issue with any luminance based exposure aid, in essence they are only good for highlights not to monitor clipping of single channels which is related to saturation

if you look at the color checked you can see blue eventually clips and this matches raw digger

I am not supporting that there is ETTR tools on camera, in fact there is not on any camera I am just doing some tests to see what is actually going on the limitations you mention exist but they are common to any luminance based tool

--

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/
If you want to get in touch don't send me a PM rather contact me directly at my website/social media
 
what sensitivity did you set the zebras at ?
As per post 109+
thats not sensitivity.
Zebra has not sensitivity it has levels and so does peaking on my A1

The level for low is 109+ there is nothing higher

If you refer to exact zebra values there is a range but this is to nail a specific exposure i.e. skin at 65% +-5 not relevant for clipping monitoring where you want low limit that has no range
its the same as adjusting focus peaking for a more precise measurement.
it does not exist for highlight clipping on the A1 and I think on most models

Zebra is based on luminance there is not such a thing to sensitivity a pixel is or isn't a certain RGB value
i set custom to 109+ range +1 is the most accurate for raw files ,your a1 has the same control.
There is no range in the lower limit

If you use the medium and say +1 -1 it does not show 110 it only shows 108 and becomes the same as setting 108+ so you may as well just set 108+ lower limit
Zebra only relies on green channel. It's based on raw, but only based on green primaries. Histogram on the other hand, is not your friend as well. It's based on JPEG.

If you have a bright red or bright blue it sometimes would not correctly show zebras on 109+. This might cause washed sky, and other unwanted over exposure artefacts.

So, there's really no way to do accurate ETTR on Sony cameras. Maybe some exposure bracketing can help.
I don't think that is accurate because Zebra come out on screen and therefore will be based on luminance that with appropriate coefficients maps all RGB channels

The reason why bright red and blue do not saturate the zebra is because there are twice green than there are red and blu so even if you add a clipped red the zebra does not clip but this is an issue with any luminance based exposure aid, in essence they are only good for highlights not to monitor clipping of single channels which is related to saturation

if you look at the color checked you can see blue eventually clips and this matches raw digger

I am not supporting that there is ETTR tools on camera, in fact there is not on any camera I am just doing some tests to see what is actually going on the limitations you mention exist but they are common to any luminance based tool
TBF a easy fix would be just add some RGB waveform tools. Easy to see clips in these tools.
 
what sensitivity did you set the zebras at ?
As per post 109+
thats not sensitivity.
Zebra has not sensitivity it has levels and so does peaking on my A1

The level for low is 109+ there is nothing higher

If you refer to exact zebra values there is a range but this is to nail a specific exposure i.e. skin at 65% +-5 not relevant for clipping monitoring where you want low limit that has no range
its the same as adjusting focus peaking for a more precise measurement.
it does not exist for highlight clipping on the A1 and I think on most models

Zebra is based on luminance there is not such a thing to sensitivity a pixel is or isn't a certain RGB value
i set custom to 109+ range +1 is the most accurate for raw files ,your a1 has the same control.
There is no range in the lower limit

If you use the medium and say +1 -1 it does not show 110 it only shows 108 and becomes the same as setting 108+ so you may as well just set 108+ lower limit
Zebra only relies on green channel. It's based on raw, but only based on green primaries. Histogram on the other hand, is not your friend as well. It's based on JPEG.

If you have a bright red or bright blue it sometimes would not correctly show zebras on 109+. This might cause washed sky, and other unwanted over exposure artefacts.

So, there's really no way to do accurate ETTR on Sony cameras. Maybe some exposure bracketing can help.
I don't think that is accurate because Zebra come out on screen and therefore will be based on luminance that with appropriate coefficients maps all RGB channels

The reason why bright red and blue do not saturate the zebra is because there are twice green than there are red and blu so even if you add a clipped red the zebra does not clip but this is an issue with any luminance based exposure aid, in essence they are only good for highlights not to monitor clipping of single channels which is related to saturation

if you look at the color checked you can see blue eventually clips and this matches raw digger

I am not supporting that there is ETTR tools on camera, in fact there is not on any camera I am just doing some tests to see what is actually going on the limitations you mention exist but they are common to any luminance based tool
TBF a easy fix would be just add some RGB waveform tools. Easy to see clips in these tools.
You need RGB real time histograms and those ar not provided in camera

The other way is to use UNIWB and then luminance is correct however pictures look horrid
 
+1
 
As I understand it, and I might be wrong, the histogram readings (and therefore the zebras and clipping) are based on the jpegs. Therefore, if you are shooting RAW, you have some headroom left.

As I understand it, and again, I might be wrong and I'd welcome a Sony guru contributing to this, you can adjust the Sony inbuilt histogram to more accurately reflect RAW by using the DRO settings. Adding DRO changes the dynamic range of the jpegs so that the reading can be set so they are closer to what you will get from RAW. DRO settings only affect jpegs so your RAW files wont be altered. I don't know what the appropriate DRO settings should be to do this.
DRO affects raw as it changes exposure
But the answer is complicated. Sony has always been coy in explaining just what DRO does. I'm going back many years here, but I seem to remember that Sony first licenced and eventually bought the rights to the tech from British firm.

Here's more discussion on it.

https://www.dpreview.com/forums/thread/3860468

In my experience, and from what I have read over the years, DRO makes no difference at all to RAW images unless you use processing that takes it into account. The question here is whether or not clipping warnings are based on RAW or jpeg. I'm suggesting that they are based on the jpeg files, which is what you see in the evf even if you are shooting only RAW. I'm also suggesting that DRO settings change the clipping indicators because they are reading from the adjusted jpegs. ie., the dro settings can be used to give more accurate clipping indication, especially of highlights, without changing the rAW file.

But I'd welcome more input on this from others. If I remember correctly, and I'm going back to Sony A700 era, this question was asked specifically of Sony and no definitive answer was forthcoming.
I revisited this topic as there was discussion on zebra and ettr which in my view is not a good combination
I agree. I have always wanted a definitive answer on the use of DRO, RAW and clipping readings.
Hi Mike

I am very familiar with the inner workings of the DRO technology as the same was available for Panasonic that has now moved away from it.

DRO is mostly for jpegs or video it does two things assuming it is the same as panasonic

1. Lifts shadows locally in the image not overall so fixes dark areas only

2. Underexposes so that you do not clip from my tests up to 2/3 Ev

I read all those threads that it does not affect raw files but it does you just have to find the right scene like a sunset and you will see the camera drop the exposure as you increase DRO strength. The algorithm makes decisions for you however this is optimised for Jpegs in camera and it does not affect the histogram directly unless it decides to change exposure

Shadow boost is performed applying gamma curve correction to parts of the image afterwards so it is similar to using masks

I believe this technology is not very dated considering that masking is very strong even in lightroom so I would not loose my sleep on it
 
I wanted to stress the validity of the Zebra +109 highlight warning as well as the highlight warning in playback

WIth controlled light I took several shots as follows in multimeter with creative look Standard no picture profile
What is the quality of the light source you use? Probably it would be better to use full spectrum daylight outdoor to find how zebra works for everyday exposure.
This creates some challenges for those users that believe in ETTR as clearly the camera metering system is not well set and gives false positives.
If you use in-camera light meter, you have to use spot metering for the highlight areas, and then calculate the number of stops of exposure that should be added to get as close as possible to highlight clipping. This is cumbersome compared to using zebra pattern, so I use zebra all the time for setting the best exposure.

The problem with in-camera metering is that you measure reflected light from the scene, so the color and brightness of the subject/scene is crucial. Multi pattern metering helps, but this is also a kind of guesswork. Also, you could use a separate incident light meter.

You could use a grey card, but then what about highlights like white, backlit clouds for landscape scenes?

From my experience with stills photography, using zebra +109 is within 0.5 stop precision for highlight clipping when using RAW/Lightroom, and way more consistent than any auto exposure. Also, about 0.5 stop "safety zone" for highligh clipping when using raw is good practise, no matter metering method.

For video, zebra pattern is set according to the Picture Profile that is used, and rarely above a value of 100. You need to know the base ISO for the actual Picture Profile to set correct value for the zebra pattern.

For video, I use zebra for highlight warning, but you could also set the zebra value much lower and use this as a warning for skin tones. The problem is that skin tones differ a lot, also for the same face when lit directly or backlit.

As Sony says: Creative Look for stills, Picture profiles for video.

From my experience, professional work, zebra works fine for both stills and video. You have to set the zebra value according to what you want the zebra pattern to tell you, though, and also from how the camera is set up.
 
I wanted to stress the validity of the Zebra +109 highlight warning as well as the highlight warning in playback

WIth controlled light I took several shots as follows in multimeter with creative look Standard no picture profile
What is the quality of the light source you use? Probably it would be better to use full spectrum daylight outdoor to find how zebra works for everyday exposure.
I use a light box with daylight illumination as the only source of light
This creates some challenges for those users that believe in ETTR as clearly the camera metering system is not well set and gives false positives.
If you use in-camera light meter, you have to use spot metering for the highlight areas, and then calculate the number of stops of exposure that should be added to get as close as possible to highlight clipping. This is cumbersome compared to using zebra pattern, so I use zebra all the time for setting the best exposure.
No that is not how it is done. I have an exposure checker I meter on the mid grey first and then see where the white is.
The problem with in-camera metering is that you measure reflected light from the scene, so the color and brightness of the subject/scene is crucial. Multi pattern metering helps, but this is also a kind of guesswork. Also, you could use a separate incident light meter.

You could use a grey card, but then what about highlights like white, backlit clouds for landscape scenes?
Real life scenes are what they are there is no right way to expose it, if the scene exceeds the camera dynamic range you need to make sure what matters is exposed well. The camera meter is a suggestion hence the need to understand what else can be done other than bracketing
From my experience with stills photography, using zebra +109 is within 0.5 stop precision for highlight clipping when using RAW/Lightroom, and way more consistent than any auto exposure. Also, about 0.5 stop "safety zone" for highligh clipping when using raw is good practise, no matter metering method.

For video, zebra pattern is set according to the Picture Profile that is used, and rarely above a value of 100. You need to know the base ISO for the actual Picture Profile to set correct value for the zebra pattern.

For video, I use zebra for highlight warning, but you could also set the zebra value much lower and use this as a warning for skin tones. The problem is that skin tones differ a lot, also for the same face when lit directly or backlit.
I know exactly how video works I shoot in slog3.cine no issues with video as what you see is what you get
As Sony says: Creative Look for stills, Picture profiles for video.

From my experience, professional work, zebra works fine for both stills and video. You have to set the zebra value according to what you want the zebra pattern to tell you, though, and also from how the camera is set up.
From what I can see zebra can be a support tool to make sure important details are not blown but it is not the way you determine the exposure overall
 
From what I can see zebra can be a support tool to make sure important details are not blown but it is not the way you determine the exposure overall
When you set exposure for where you want to keep highlight detail, shadow detail will follow as deep into the shadows as the sensor is good for.

Where to keep (and blow) highlights is a choice, just like you have to make a choice when using a light meter. I find using zebra for optimum RAW exposure much faster and just as precise as calculating from light meter readings.

For a wider dynamic range than what the sensor is capable of, techniques like HDR exposure is available.

So yes, using zebra pattern is excellent to ensure the best RAW exposure,
 
From what I can see zebra can be a support tool to make sure important details are not blown but it is not the way you determine the exposure overall
When you set exposure for where you want to keep highlight detail, shadow detail will follow as deep into the shadows as the sensor is good for.

Where to keep (and blow) highlights is a choice, just like you have to make a choice when using a light meter. I find using zebra for optimum RAW exposure much faster and just as precise as calculating from light meter readings.

For a wider dynamic range than what the sensor is capable of, techniques like HDR exposure is available.

So yes, using zebra pattern is excellent to ensure the best RAW exposure,
agree 100%
 
From what I can see zebra can be a support tool to make sure important details are not blown but it is not the way you determine the exposure overall
When you set exposure for where you want to keep highlight detail, shadow detail will follow as deep into the shadows as the sensor is good for.

Where to keep (and blow) highlights is a choice, just like you have to make a choice when using a light meter. I find using zebra for optimum RAW exposure much faster and just as precise as calculating from light meter readings.

For a wider dynamic range than what the sensor is capable of, techniques like HDR exposure is available.

So yes, using zebra pattern is excellent to ensure the best RAW exposure,
It is not as straightforward like that because zebra being monochrome will not show clipping if one channels clips and the others don't which is one of the points I am making here. It can be demonstrated shooting a red or blue patterned object but this is not in my scope

Anyway light meters are not required as those only apply in studio scenes and in those scenes you never have a problem
 
As I understand it, and I might be wrong, the histogram readings (and therefore the zebras and clipping) are based on the jpegs. Therefore, if you are shooting RAW, you have some headroom left.

As I understand it, and again, I might be wrong and I'd welcome a Sony guru contributing to this, you can adjust the Sony inbuilt histogram to more accurately reflect RAW by using the DRO settings. Adding DRO changes the dynamic range of the jpegs so that the reading can be set so they are closer to what you will get from RAW. DRO settings only affect jpegs so your RAW files wont be altered. I don't know what the appropriate DRO settings should be to do this.
DRO affects raw as it changes exposure
But the answer is complicated. Sony has always been coy in explaining just what DRO does. I'm going back many years here, but I seem to remember that Sony first licenced and eventually bought the rights to the tech from British firm.

Here's more discussion on it.

https://www.dpreview.com/forums/thread/3860468

In my experience, and from what I have read over the years, DRO makes no difference at all to RAW images unless you use processing that takes it into account. The question here is whether or not clipping warnings are based on RAW or jpeg. I'm suggesting that they are based on the jpeg files, which is what you see in the evf even if you are shooting only RAW. I'm also suggesting that DRO settings change the clipping indicators because they are reading from the adjusted jpegs. ie., the dro settings can be used to give more accurate clipping indication, especially of highlights, without changing the rAW file.

But I'd welcome more input on this from others. If I remember correctly, and I'm going back to Sony A700 era, this question was asked specifically of Sony and no definitive answer was forthcoming.
I revisited this topic as there was discussion on zebra and ettr which in my view is not a good combination
I agree. I have always wanted a definitive answer on the use of DRO, RAW and clipping readings.
Hi Mike

I am very familiar with the inner workings of the DRO technology as the same was available for Panasonic that has now moved away from it.

DRO is mostly for jpegs or video it does two things assuming it is the same as panasonic

1. Lifts shadows locally in the image not overall so fixes dark areas only

2. Underexposes so that you do not clip from my tests up to 2/3 Ev

I read all those threads that it does not affect raw files but it does you just have to find the right scene like a sunset and you will see the camera drop the exposure as you increase DRO strength. The algorithm makes decisions for you however this is optimised for Jpegs in camera and it does not affect the histogram directly unless it decides to change exposure

Shadow boost is performed applying gamma curve correction to parts of the image afterwards so it is similar to using masks

I believe this technology is not very dated considering that masking is very strong even in lightroom so I would not loose my sleep on it
I have been traveling and missed your reply. Thanks for that.

I agree that DRO lifts the shadow areas in jpeg.

My question then remains, if Dro is activated, as it only works on jpeg and jpeg is what we are seeing through the evf, will the histogram readings more accurately reflect the heard reality of the RAW image, when we are shooting RAW?
 
Somebody posted a reply that the zebras were based on raw, but I had read before that (like the histogram), it is based on the jpg being emulated.

On that note, some folks way back were advising to setup the color space as AdobeRGB and not sRGB. Also, it would make sense to setup the camera neutral profile (and not camera standard). Could you check if the zebra warnings change if you do this?

ps: in the Fuji system this makes a difference. But then they have 3 different histograms for RGB which is much nicer.

Thanks
 
The color space has no impact on zebra because the simulation works on srgb anyway

i have checked the difference between natural and standard

mostly the level of saturation is less while the difference in the clipping is fairly small around 1-2%
i can lost some samples no problem
 
I have created a post in the the forum with the creative looks there are differences but not as much as you would think I guess
 

Keyboard shortcuts

Back
Top