[Ian Stuart Forsyth]

Ian Stuart Forsyth : This is quite deep I think I partially understand, there are lot of variable in digital always learning thanks for the info. !

Will also add that you have to be careful when trying to understand the histogram in a photo editor , what it means when the red is to the right and that many times it does not mean that your color red in no way means its clipping.
Here is a image

Now if we rely on the histogram of your image editing software it will tell you that you have a great deal of red clipping and that you should go about stopping this


all 3 of these colors could lead you to thinking that there is clipping in the red channel, my question to you are they really clipped. It just shows that red is at full saturation and or close to the limit of the color gamut
If you want to show these colors in your photograph than they are not and if I change them so that the histogram show no clipping then I will fundamentally change the those colors and if I am trying to accurately capture and show those colors in my image I may be decreasing the range of colors within my image if I use the histogram

Top right color #ff002f Color Hex
Top left color #ff00c2 Color Hex
Bottom color #ffc82f Color Hex

So then what the question again is what is it really mean with the red show a heavy count to the right in photo editing software ?

Here is a second image

Now if I look at the red in the histogram you will be given the idea that it all should be corrected because of how close the red is to the right

But if I want to show any range of red contained in this image I will need to have the red to the far right in the image. If not then I will lose that complete range of these reds shown in the above image

[GUB]

Will also add that you have to be careful when trying to understand the histogram in a photo editor , what it means when the red is to the right and that many times it does not mean that your color red in no way means its clipping.
Here is a image

Now if we rely on the histogram of your image editing software it will tell you that you have a great deal of red clipping and that you should go about stopping this


all 3 of these colors could lead you to thinking that there is clipping in the red channel, my question to you are they really clipped. It just shows that red is at full saturation and or close to the limit of the color gamut
If you want to show these colors in your photograph than they are not and if I change them so that the histogram show no clipping then I will fundamentally change the those colors and if I am trying to accurately capture and show those colors in my image I may be decreasing the range of colors within my image if I use the histogram

Top right color #ff002f Color Hex
Top left color #ff00c2 Color Hex
Bottom color #ffc82f Color Hex

So then what the question again is what is it really mean with the red show a heavy count to the right in photo editing software ?

Here is a second image

Now if I look at the red in the histogram you will be given the idea that it all should be corrected because of how close the red is to the right

But if I want to show any range of red contained in this image I will need to have the red to the far right in the image. If not then I will lose that complete range of these reds shown in the above image

Pretty well agree here Ian - your red here ( in 8 bit) is 255 with a bit of blue thrown in.
But the issue is the loss of texture / detail in a red flower and your image makes a good base to give an example of what happens with over exposure.
So in your image I have inserted a rectangle of red darkened 20 points (so 235). It is quite visible as you can see and this represents the idea of a tonal range in a red flower.
Now I brighten the image a bit with the colour curves tool ( think of this are increasing exposure in the camera) and the brightest red stays at 255 (which is as bright as it can get) and the 235 rectangle has also brightened to 255 and therefore is almost invisible -- hence loss of detail or that horrid flattening that you get from clipping.
And the thing is you cannot back away from this by darkening those reds again - now they are clipped together.
This is why these bright flower colour issues are simple overexposure in the camera as Clackers was saying on page one.
If your flower was pure white you would know immediately that you were overexposing.
These fully saturated colours have to be treated like they were white.
White of course is the three primarys mixed at full saturation so it makes sense. .

[timb64]

So many shades of “red”

[GUB]

Indeed there is so many shades of red.
Rhododendron arboreum, the National flower of Nepal, photographed in Nepal.
Heavy crop of a K-1 image , FA100mm 3.5 macro.
The natural colour of the flower is close to full saturation red.

The trick is to retain a tonal range so close to the full saturation.

Attached Images

View Picture EXIF

File Date

 

[Trickortreat]

Too much info here. Just shows how little i understand color for now

[GUB]

Still a heck of a lot to learn here too - but it is a fun journey!!.
Almost all colour clipping I see on PF is created from simple over-exposure in the camera with the occasional example being from someone over-enthusiastic on an editing slider without looking at the result.
Gaining a good understanding of histograms is a necessity and mastering the curves tool in conjunction with the histogram is the way to get the best out of a shot.
And also understanding your subject and the likely issues it will present is important too.
Like Ian said above sometimes the colours of the subject are right at the colour palette edges and you have to live with that.

[Ian Stuart Forsyth]

But the issue is the loss of texture / detail in a red flower and your image makes a good base to give an example of what happens with over exposure.

Sorry its not about over exposure its about displayed colors and that numeric values it occupies within a color space.

There is a difference between having a properly exposed image, the correct lightness in our final image and how we show those in a color space. Most of the time when we are photographing its the processing that has caused us the issues with reds this includes the camera profiles as I have shown when you look at the raw histograms in earlier pages.

When we are using the cameras Histogram during exposure in raw we are using it to measure saturation capacity so that we don't clip data. When using in camera histogram for in camera jpg we are trying to find the best exposure to fit the cameras color profile and color space you have selected for a given scene . With the histogram in the photo editor we are using it to placing our exposure within a color & tonal range of a color space.

There is a difference in how we use those histograms for in camera and photo editing software, If we do not use RGB values of 255 in our processed image we can be removing any details and tonal range that is found in our scene because to reproduce them those color values need us to use R 255 in our displayed image.

There is really no raw R255 value (what we can see is saturation levels of the RGB with special software) what we see in the raw converter in RGB values that is derived from using a color profile, the WB you select and how we place all that information using tonal, saturation,contrast sliders, along with the color space and bit depth will decides what data falls in at R 255.

We have to display images within a color space there can be many colors that fall into the range that R255 is required in a color space . Take this image below.
http://isfphotography.com/255.tif
everyone of those colors that we see here (much of the color ranges we have with red 255 is to large to show all of them in a image like this) we need to have red at 255 and several of those we also need either B or G to also have 255. If I where to use the histogram in my photo editorr (not cameras histogram used for judging exposure) and show R at 255 many would be telling you have clipped your reds.



In this image we need R 255 and also G255 B255, many of those variations shown above so we can see these different colors, to alter them so you wont show reds in the 255 histogram you will be clipping the size of color gamut for that color space to a smaller one, this would removing a great amount of the colors you see above , now some of them will not be altered like the ones at the very corners of the magenta, red, orange and yellows.

To put it simply if I take a photograph of the tiff file shown above which has get deal of color variations, I would use an exposure as not to clip the data we need to create the final image, Now if I want to process the file to accurately depicted how the original image looked we would need to have all of the red pixels at 255 or we will not correctly recreate the final image as it is seen in the tiff file. The reason why is that a great deal of the colors of that scene fall in the colors gamut's values that use R255 marker and a get deal of them needing ether B255 and or G255 also

Here I have deceased the red down to 254 to conform to the believe that there is clipping in the red
http://isfphotography.com/254.tif
In this smaller image the banding is less prominent because of the range of the limited colors because of the size of the image but here you can see banding in much of the colors and that is because you have clipping off the color gamut of those lost R255 values
Here I lowered it to 253
http://isfphotography.com/253.tif

All of that banding is because of the clipping of the color gamut. You will also see that in the reds, oranges and magentas that fall at the outermost regions of the image did not change in how we perceive those colors when lowered to 253. The major reason why this is, the color data that fails within those limits 255-253 values found at the outer boundary of the image are much more accurate at measuring variations than we can see.

There differences as to why we use the histogram (more importantly raw histogram) to evaluate the correct exposure for the camera and there are difference as to why we use a histogram in processing our raw data and finally the histogram in how we choose to display the final image.

Oddly enough the red colors that are the at most greatest risk of clipping when setting your cameras exposure are the ones found in the center 3rd of the first time tiff file and not the reds that everyone thinks that are clipping. Its the color gamut, your WB and other processing that does the clipping for you.

---

Last edited by Ian Stuart Forsyth; 05-16-2019 at 01:55 AM.

[GUB]

Sorry its not about over exposure its about displayed colors and that numeric values it occupies within a color space.

There is a difference between having a properly exposed image, the correct lightness in our final image and how we show those in a color space. Most of the time when we are photographing its the processing that has caused us the issues with reds this includes the camera profiles as I have shown when you look at the raw histograms in earlier pages.

When we are using the cameras Histogram during exposure in raw we are using it to measure saturation capacity so that we don't clip data. When using in camera histogram for in camera jpg we are trying to find the best exposure to fit the cameras color profile and color space you have selected for a given scene . With the histogram in the photo editor we are using it to placing our exposure within a color & tonal range of a color space.

There is a difference in how we use those histograms for in camera and photo editing software, If we do not use RGB values of 255 in our processed image we can be removing any details and tonal range that is found in our scene because to reproduce them those color values need us to use R 255 in our displayed image.

There is really no raw R255 value (what we can see is saturation levels of the RGB with special software) what we see in the raw converter in RGB values that is derived from using a color profile, the WB you select and how we place all that information using tonal, saturation,contrast sliders, along with the color space and bit depth will decides what data falls in at R 255.

We have to display images within a color space there can be many colors that fall into the range that R255 is required in a color space . Take this image below.
http://isfphotography.com/255.tif
everyone of those colors that we see here (much of the color ranges we have with red 255 is to large to show all of them in a image like this) we need to have red at 255 and several of those we also need either B or G to also have 255. If I where to use the histogram in my photo editorr (not cameras histogram used for judging exposure) and show R at 255 many would be telling you have clipped your reds.



In this image we need R 255 and also G255 B255, many of those variations shown above so we can see these different colors, to alter them so you wont show reds in the 255 histogram you will be clipping the size of color gamut for that color space to a smaller one, this would removing a great amount of the colors you see above , now some of them will not be altered like the ones at the very corners of the magenta, red, orange and yellows.

To put it simply if I take a photograph of the tiff file shown above which has get deal of color variations, I would use an exposure as not to clip the data we need to create the final image, Now if I want to process the file to accurately depicted how the original image looked we would need to have all of the red pixels at 255 or we will not correctly recreate the final image as it is seen in the tiff file. The reason why is that a great deal of the colors of that scene fall in the colors gamut's values that use R255 marker and a get deal of them needing ether B255 and or G255 also

Here I have deceased the red down to 254 to conform to the believe that there is clipping in the red
http://isfphotography.com/254.tif
In this smaller image the banding is less prominent because of the range of the limited colors because of the size of the image but here you can see banding in much of the colors and that is because you have clipping off the color gamut of those lost R255 values
Here I lowered it to 253
http://isfphotography.com/253.tif

All of that banding is because of the clipping of the color gamut. You will also see that in the reds, oranges and magentas that fall at the outermost regions of the image did not change in how we perceive those colors when lowered to 253. The major reason why this is, the color data that fails within those limits 255-253 values found at the outer boundary of the image are much more accurate at measuring variations than we can see.

There differences as to why we use the histogram (more importantly raw histogram) to evaluate the correct exposure for the camera and there are difference as to why we use a histogram in processing our raw data and finally the histogram in how we choose to display the final image.

Oddly enough the red colors that are the at most greatest risk of clipping when setting your cameras exposure are the ones found in the center 3rd of the first time tiff file and not the reds that everyone thinks that are clipping. Its the color gamut, your WB and other processing that does the clipping for you.

So my best practise for taking a shot like the Rhododendren above is to expose as far to the right on the Histogram as I dare with the red (or whatever colour) without risking hitting the right hand side.
This is done realising that in camera I am looking at a histo of the jpg extraction of the raw dictated by my jpg settings. So this is done with a margin of error built in.
For me best practise also includes being at base Iso (100) to maximise Dynamic Range. This includes times when base Iso is insufficient for correct exposure. (Utilising Iso invariance)
In PP I can drag the exposure more precisely to the optimum brightness but if I have already overexposed in the camera I can not bring the detail back - ie clipped.
If you put a colour dropper on the Rhodo you will see there is a certain number of pixels at R255 and in practise I have tolerated a small amount of clipping in recognition that in real life that is where the colour of the flower sits.
We seem to differ on the concept of over-exposure and I fail to see where you have offered any solutions to the over cooked reds we see around.
How about telling us your best practise approach to such a subject.

[stevebrot]

There is really no raw R255 value (what we can see is saturation levels of the RGB with special software) what we see in the raw converter in RGB values that is derived from using a color profile, the WB you select and how we place all that information using tonal, saturation,contrast sliders, along with the color space and bit depth will decides what data falls in at R 255.

I can confidently assert that the sensor capture data* is limited to 14 bits (all current model Pentax dSLRs) and that an exposure to any sensel resulting in a recorded value of 214 is clipped since it cannot, with confidence, be rendered in any color space without ambiguity. A color space may support that value, but only for captures at greater than 14-bit. White balance and gamut coercion are red herrings, despite anomalies introduced by both having similar appearance to true clipping.


Steve

* That would be the detector voltage after initial scaling as saved to file.

[stevebrot]

How about telling us your best practise approach to such a subject.

I know the question was not intended for me, but my practice has been to intentionally bracket to underexposure when saturated red or yellow are part of the scene and to always shoot RAW.* The rear LCD histogram is useful for detection as might also be "blinkies", but within the limits discussed on this thread. I no evidence, but capture (even RAW) to the wider gamut Adobe RGB might result in a more accurate histogram.


Steve

* No, I don't generally expose to the right and particularly not when I anticipate red channel clipping. I will, however, apply positive EC when the meter is obviously biased by large areas of sky or similar in the frame.

[TonyW]

Question to Lotech
Your last response 6 days ago. Are you any clearer on your red issues?

As you know I suspect that your issue is not clipped reds (likely I think processing) but if you think it is AND you are capturing raw then why not post a download link so we can take a look

[Ian Stuart Forsyth]

So my best practise for taking a shot like the Rhododendren above is to expose as far to the right on the Histogram as I dare with the red (or whatever colour) without risking hitting the right hand side.

If you are wanting to use your cameras histogram I would first look at understanding what the cameras WB is doing and what happens to the color channels
How red is your red ? - Page 3 - PentaxForums.com

If you are using photoshop I would find out how much of a BLE your camera is using this way you can override the highlight headroom that is built into and that PS is using to setup the raw processing covered in the link about. covered in the above link with a link showing how to setup PS
Also consider adjusting your WB in camera until you reach a WB setting of 1,1,1 in line c628 of your exif data of your photograph and store that in your camera so you can pull it up when you need it. Covered here
How red is your red ? - Page 3 - PentaxForums.com

This would give you a pretty good idea of what your raw data looks like, then you can go about setting up the camera and what to expect when looking at the cameras histogram, If you don't want to go thru this work you might look at buying fastraw viewer

Youtu.be

Youtu.be

This will give you the best way of knowing what your camera is doing and just as important you can find how the relationship with how the color channels clip and how they clip in different light sources, You will find in natural outdoor light the green channels is first to clip while red is less likely and the program is cheap I think for what it allows you to find out.

Once you know your raw files you can go about finding a way of setting up an exposure management system that will work for you. I use 2 basic methods one based on the histogram of my camera and the second on the cameras metering system. With the Histogram method I set it up as close as I can to what the raw data contains, again using a UniWB this way you will get a better representation how the 3 color channels fall that exposure ( use the WB for all your conditions of lighting as it does not matter). I would then go about setting up the cameras color profile to the most neutral setting. Then its just trial and error until you find out how your cameras histogram shows you raw data.

The one I mostly use is the cameras metering system, for this I find a white target while in Manual mode I meter on that white target and increase that exposure until it reads +3 in your OVF. Take a photo and look at it in your raw viewer and see how far that white target is from clipping and then keep increasing the cameras exposure until you see your first channel start to clip in the raw viewer. Now you know how many stop till clipping, for me I had to raise my camera 1 2/3 stops to clip and I like to have a safety net so I don't go the full 1 2/3 stops I use 1 to 1 1/3 and set my cameras EC to that. So now when I place my spot metering point on what I want to be white in my raw file it should read +2 2/3 to 3 with my cameras metering system.

[GUB]

If you are wanting to use your cameras histogram I would first look at understanding what the cameras WB is doing and what happens to the color channels
How red is your red ? - Page 3 - PentaxForums.com

If you are using photoshop I would find out how much of a BLE your camera is using this way you can override the highlight headroom that is built into and that PS is using to setup the raw processing covered in the link about. covered in the above link with a link showing how to setup PS
Also consider adjusting your WB in camera until you reach a WB setting of 1,1,1 in line c628 of your exif data of your photograph and store that in your camera so you can pull it up when you need it. Covered here
How red is your red ? - Page 3 - PentaxForums.com

This would give you a pretty good idea of what your raw data looks like, then you can go about setting up the camera and what to expect when looking at the cameras histogram, If you don't want to go thru this work you might look at buying fastraw viewer
Why You Need FastRawViewer If You Shoot in RAW - YouTube This will give you the best way of knowing what your camera is doing and just as important you can find how the relationship with how the color channels clip and how they clip in different light sources, You will find in natural outdoor light the green channels is first to clip while red is less likely and the program is cheap I think for what it allows you to find out.

Once you know your raw files you can go about finding a way of setting up an exposure management system that will work for you. I use 2 basic methods one based on the histogram of my camera and the second on the cameras metering system. With the Histogram method I set it up as close as I can to what the raw data contains, again using a UniWB this way you will get a better representation how the 3 color channels fall that exposure ( use the WB for all your conditions of lighting as it does not matter). I would then go about setting up the cameras color profile to the most neutral setting. Then its just trial and error until you find out how your cameras histogram shows you raw data.

The one I mostly use is the cameras metering system, for this I find a white target while in Manual mode I meter on that white target and increase that exposure until it reads +3 in your OVF. Take a photo and look at it in your raw viewer and see how far that white target is from clipping and then keep increasing the cameras exposure until you see your first channel start to clip in the raw viewer. Now you know how many stop till clipping, for me I had to raise my camera 1 2/3 stops to clip and I like to have a safety net so I don't go the full 1 2/3 stops I use 1 to 1 1/3 and set my cameras EC to that. So now when I place my spot metering point on what I want to be white in my raw file it should read +2 2/3 to 3 with my cameras metering system.

So in other words you are careful not to over-expose the shot !

[TonyW]

If your really want to override the highlight headroom and other raw editor additions BLE is only a starting point. Adobe process versions have changed over time to accomodate what Adobe feels represents an 'good/acceptable starting point'

A -0.5 BLE value to exposure (note this value may change with ISO) is not enough to give you a clear picture of what Adobe have changed in process versions. To look at the data closer to what was added you will need to make changes to the tonal curve and adjust Exposure to -1.0 + BLE, i.e. new Exposure slider value will be -1.50. WIth Contrast and Black sliders adjusted to -33, +25 respectively.

These adjustment figures will bring you closer to what would be seen in RawDigger and some other raw editors such as RawTherapee. The image below I think illustrates the 'problem' quite clearly. Initial capture relying on the cameras matrix metering, reveals an OK exposure in ACR with indications of more exposure headroom. Wait for the image change to reveal BLE compensation and exposure, contrast and black level nulling from previous versions and we have a totally different story a pretty much underexposed image IMO. Please forgive the banding that has been caused by dropping image straight to GIF

All editors must add their own interpretation of raw data as you really would not like what you see viewing actual raw data i.e. a flat and dark image prior to rendering to what someone deems as acceptable. An example of raw data dump up to demosaicing. Someone has to apply a rendering to this to get an acceptable looking starting point. Unless there a specific needsf or data to be shown in a certain way then rendering interpretations will look different between editors and other subjective views.

https://www.pentaxforums.com/forums/attachments/6-pentax-dslr-discussion/439...awdataview.jpg

[GUB]

Isn't it just simpler , once you start working in raw, to leave your camera jpg settings in a standardised state and gain a sense of what you can get away with with experience. One glance at the editor raw histo to see if the right hand end of the data is spiked and sawn off is all you need to check --- it doesn't need to get more complicated than that.