Raw Histogram??

Again my understanding of this whole idea of buckets is that they relate to photon collectors on the sensor.. These collectors can count so high before they overflow (so to speak.) That is the height as I am using it .... You seem to want to teach a graduate class in this elementary school!! ..

WhyNot

I really don't care what the software developer is thinking or intent is as I look through the viewfinder at that bird on a branch ... So you're saying that the Raw Histogram is of no interest as I start to press that shutter button ....

WhyNot

Every pixel on the sensor is used to calculate the number of pixels with a particular coded value.

As to the coded values themselves, for raw files, they can range from 0 to the maximum value that the ADCs can encode. I wouldn't call that the height of the bucket. The x axis of a raw file with a linear x-axis presentation (not often a very useful way to look at things) contains all the possible values that the ADC can put put divided by the bin width.

It's not what the software developer is thinking, but what the person who is going to be controlling the raw developer -- presumably, the photographer -- is thinking. You want to capture an image that you'll be able to process into what you want.

That means avoiding most, but probably not all, clipping of raw values. Understanding what raw values can be clipped is an art that photographers have to learn through knowledge, vision, and experimentation.

I withdraw the comment - assuming that the elephant in the room actually does put 00 to FF on the displayed histogram. But I may have misunderstood "x-axis" or the derivation thereof.

I, and possibly many others, am more used to histogram graphics that show no units for the x-axis or decimal numbers and (shame) I did not read the links, tsk.

Being an octogenarian scientist, I would have said "x-coordinate data values" to the great irritation of some members here, LOL.

I thought that would be confusing, so I converted to decimal values and put those on the displayed histogram.

Let me say that I have been a student from kindergarten through graduate school. I have spent decades in technical fields and many of those concerned with software and systems but I gave that up many years ago and am now technologically incompetent – a beginner – a student again and this discussion has not enlightened me much … I'm still not sure what I get as a photographer from the RAW histogram that I would not have gotten from the one I see in my viewfinder … Generally, I am thinking about the subject and the exposure I need to set to get that vision that I have right.. Like Adam's advised I will include in that thinking how I will crop and process the file I get.. I only need a file that I can process to what I think the final picture should be … and I have finite time in which to make that decision, usually …

This is a beginners forum as it is titled … The teacher needs to make sure the student understands the lesson … but then I need to go out and find a bird on a limb I've spent to much time here already ….

I thank you for your patience and the information shared .. I will continue to read responses to this post later .. Thank you all for participating .. have fun …

WhyNot

You want to get the most light on your sensor that won't cause clipping in significant highlights in the raw file. Unless you take special pains, you can't reliably tell by looking at the live view histogram on your camera whether the raw file is close to clipped, or is indeed clipped.

If you are an AA fan, you probably have used the Zone System. In that system, the optimum exposure depends on the intended development.

Yeah, reading back through the thread, I get it. I think you'd like to know the difference between a raw histogram and what's usually depicted, no? And then, of what use is it?

Just to anchor the explanation, I'm talking about a raw image as an two-dimensional array of values that represent the light from a scene. Typically, the image you look at has been transformed from the scene measurements by a sequence of tools, e.g., demosaic, white balance, tone curve, etc. So, the values in the displayed image are not the same as the corresponding values measured at the scene by the camera. contained in the raw file.

A histogram is a visual representation of the distribution of a data set. With an image as the data set, we're most interested in the distribution of tone values, and that's what an image histogram tells us: distribution of tones. Thing is, the distribution of the final rendition does not tell us the distribution of the light in the scene, due to all those tools used to make the image. The tone curve, in particular is usually a "lift" of the mid and high tones to higher than their original scene values. So, the histogram calculated from the raw data, which represents the light at the scene, is different than that of the final image, which has been produced to look right on a monitor or paper.

Soooo.... looking a a histogram of the display image doesn't tell you what you need to know about the light at the scene. Particularly, the display histogram may show more pixels saturated due to the lift of the tone curve, leading one to believe they overexposed the image when the original raw data still had some room before reaching saturation. And, you can't just dial in a particular exposure compensation and forget about it; the situation is different from scene-to-scene.

But... looking at a raw histogram with software after-the-fact just tells you about the exposure decision you made, too late to influence it. That's why there's desire to have an in-camera raw histogram, to provide that information before the shot is captured. Like I said in a previous post somewhere, I think I'd rather have raw blinkies than the raw histogram; a bit more actionable.

Anyway, an attempt to answer your original question.

ok, let's start at the beginning.

Each pixel on a sensor is covered by either a red, green or blue filter

source: www.researchgate.net/figure/Color-online-Bayer-filter-array_fig1_311861609 (19 May 2023)

While the shutter is open each pixel on the sensor collects light photons from the scene. Each pixel will have a maximum number of photons it can collect before becoming full and "overflowing". "Overflowed" photons are not recorded by the sensor pixel, so ideally you don't want to fill up the sensor pixels to full capacity while the shutter is open.

After the shutter closes the voltage or electrical charge (I always forget which it is) that the photons for each pixel generate is converted to a digital number via whatever internal processing the camera does to output to a raw file for that pixel. For 14 bit raw data that number will be between 0 - 16384 in theory. It will actually be ~16100 for reasons that people like JimKasson, Illiah Borg, Bobn2 will be able to explain.

Now, let's assume for simplicity that a sensor's pixel with a value of 16100 is deemed to be full.

A raw histogram will look similar to this:

source: @JimKasson's post dprevived.com/t/how-to-get-the-best-raw-exposure-part-2/3224/3/#post-34285 (11 May 2023)

This is a histogram of the actual unprocessed raw data. An in camera raw histogram will enable you to better maximise the exposure* before clipping the raw data than using an RGB histogram, which is of processed raw data into an actual image, because more often than not an RGB histogram will indicate clipping of highlights when the raw data is not actually clipped.

So with an in-camera raw histogram you will be able to push the raw data as close as possible to 16100 while maintaining your DOF and blur requirements. Pushing the histogram as close as possible to 16100 and within your artistic intents results in you maximising the Signal to Noise Ratio for that particular shot thus minimising visible noise in the final image.

In summary, from my experience I know I can add up to 1/2 a stop of extra exposure* past where the in-camera histogram indicates clipping before the actual raw data is clipped.

* exposure - amount of light that struck the sensor per unit area while the shutter was open

explain please.

in the photocell, a photon may or may not cause an electron to be created. That electron will go into that photocell's capacitance C and thereby increase it's charge Q. This causes the capacitance's voltage V to increase because, as is well-known in Electronics 101, Q=CV .. ergo V=Q/C.

A bit vague, Danno, not sure that Beginners will get it, sorry.

So it's the voltage. Thank you 🙂

I think most beginners will understand what I said. Basically, the voltage from each sensor pixel is eventually converted to a digital number that is allocated to that sensor pixel. It is the collective set of the number allocated to each sensor pixel that is referred to as the raw data.

For the purposes of the op's query, "I need a definition for Raw Histogram", exactly how the voltage is converted to a digital number is not relevant imo. I am trying to KISS 🙂 since it appears the op is hoping for an easy to understand explanation without too many technicalities.

Not quite.

This forum's name/title is "Beginners' Question Discussion" and its description is:

"This forum is for followup explanations from Beginners Questions (BQ). Posts may be moved from BQ to here once the discussion evolves beyond a few posts of answers.
[It used to be called Tutorials]."

You really should have posted this thread in "Beginners' Questions" where its description is:

"the place for beginners to ask questions and seek advice. Questions and their answers may be moved to the Beginners Question Discussion forum (below this one) if the answers and subsequent discussion evolves beyond a few posts."

In the "Beginners' Questions" forum then ideally yes. Here in "Beginners' Question Discussion", not necessarily.

Hi danno. just shot about 50 doll shots in the studio with white and black backdrops. To be honest sony has the best implemented histogram and zebras
of any camera brand with near perfect measurmensts. but lets face it ,in the real world you just take the good enough image into your raw convertor
and adjust the image how you want with practically no image degrading even if you not spot on with exposure. but i must say the zebras set to 70 are absolutly
spot on for skin tones, the histtogram tells you nothing inbetween highlights and shadows.

What is the difference between a histogram and zebra? I thought they were the same thing.

San x