What Is Happening Under The Histogram Hood

Ok, I'll try to break the ice on this forum and start a thread.

This thread is not really a question but a description of my understanding of what histograms are and what they represent.

Ok, here goes:

It should be helpful to most people if they understand what the histogram actually is and what it is telling you.

The two main histograms I use are:

1. RGB Histogram - which is nothing more than a frequency count of the 0 - 255 values in all the pixels in the image. Visually on its own, it tells you nothing at all regarding exactly where in the image any data point on the histogram came from.

2. Luminosity Histogram - is a frequency count of the brightness value of each pixel calculated from the rgb values for a given pixel.

Personally I prefer to use the luminosity histogram to set white and black points for an image.

Anyway, to demonstrate what the far more commonly used RGB histogram actually is, consider this simple and basic image with pure red (255,0,0), green (0,255,0) and blue (0,0,255) pixels on a pure white (255,255,255) background.

Obviously the rgb data in the image will contain only the values 0 or 255.

Here is the RGB histogram for the above image.

You can see that the only data in the histogram are the two vertical bars at 0 and 255. The height of the bars representing the total number of 0 values in the rgb data and the number of 255 values in the rgb data.

The histogram on its own cannot tell you what the colors are in the actual image or where in the image (white, red, green, blue elements) any particular histogram data point came from.

If someone was given just the above histogram without the image it came from it would be totally wrong to say the image contained only the colors black and white.

From an RGB histogram you can interpret whether the image is overall lightish, darkish or somewhere in between and whether it possibly lacks contrast.

duplicate post deleted

No illustrations, Danno. The code looks wrong.

[![rgbDemo.jpg](/a/DWMP9n4gJIbKodFy0PtLyCmwC07u0znWE17vOyIjoK8IMFmQc5b5n10Rm7NibvKJ/413/?shva=1)]
(/a/DWMP9n4gJIbKodFy0PtLyCmwC07u0znWE17vOyIjoK8IMFmQc5b5n10Rm7NibvKJ/413/?shva=1)

I just logged in again now and the images appear in my post.

But if I logout and browse this and other threads as a "guest" then the images do not appear in posts.

Are you logged in when you view this thread?

I don't know why there is a line of code below the first image.
You should see only 2 images on my post. The red green blue patches on a white background and the histogram image.

I'm logged in and I see neither image. Just some image placeholders.

Unless special provisions are taken to change the finder white balance, JPEG-derived in camera histograms, whether one or three channel, are unreliable for predicting raw file clipping.

blog.kasson.com/using-in-caera-histograms-for-ettr/

Ok, that's interesting.
The only thing I can of as to the cause is that after uploading the images from my pc and then selecting to insert the image in the post by clicking the attached images' icons, I deleted the attached images. But the images then still appear in the post when I view the post.
I will be returning home soon and will try to repost the images and this time not delete the attached images.

I think "unreliable" is a bit harsh, but certainly not totally accurate for raw clipping.
But I know from experience that after the camera histogram shows data is about to be clipped, based on the camera generated jpeg, I can get about 1/2 to 2/3 of extra exposure* before the raw data will be clipped.

But I also use the luminosity histogram in post processing with Photoshop Elements.

• exposure - amount of light striking the sensor per unit area while the shutter is open.

How much extra headroom you have is dependent on the spectrum of the illuminant and the reflectance spectrum of the subject.

This is getting beyond what a beginner wants to worry about, but the luminance histogram in Ps Elements (or Lr or Ps) doesn't tell you if the raw file is clipped or not. That histogram is after raw development.

It would be nice to know the camera being used, when viewing raw data in PE there are many under the hood processing being done to the image.
One is the baseline exposure, with many cameras this places the recorded raw values into the correct tonal values in the final processed image. For many cameras this can be anywhere from 0.5-2 stops

BLE

Long way to find uniWB yet? 😀👍

Yes, that is true but that is not what this thread is about. It's about what a histogram is and the data it represents.
When taking a shot (I shoot raw) I aim to get as much light onto the sensor as I can within my DOF and motion blur requirements without clipping important highlights. The camera's histogram helps me do that nothwithstanding its limitation.

My main camera is the Canon 90D and I use the Adobe Camera Raw (ACR) that comes with Photoshop Elements to process the raw data.

For the 90 D at iso 100 the BLE is I think around 0.16
Where are my midtones

If you follow the above link you can workout how much headroom you have -1-0.16(BLE)=-1.16 to your exposure slider

The histogram in the raw converter really does not tell you anything as to how much or how close you are to clipping unless you set it up to show you one of them being uniwb and the other being what is happing behind the scene in the converter

here is an image taken using ETTR and ARC is telling me that there is clipping all over

but the reality is that less than 1/2 percent of the green is clipping and that the red is a stop from clipping

Next what color space are you using


both are the same image but using 2 different color spaces, so what is the histogram really telling you

If I understand you correctly we are saying pretty much the same thing but in different ways.

The RGB histogram is just a display of the frequency count of the individual red, green, blues values for each pixel. Nothing more. The RGB histogram doesn't tell you from which colour or where in the image the data point came from. For example, a data point at level 255 in the histogram image could have come from any of the red, green, blue or white coloured elements in the first image.

IMHO adding "... and how to use it", or even to start with "how to use histograms" (while shooting, culling, editing) might be useful.