Histowhatagrams?

[Brownbear]

Hi Victor,

I guess we have a lot to thank Bill and Manfred for their very informative posts. What makes them easy to grasp is because of the sample photos showing what they mean.

First time I realized histograms can be indicative of the "mood" conveyed by the images. Like to the left , equates to moody; while, to the right, to happiness , upbeat mood.

Like in music. The minor chords, the "sadness" side, and the major chords, the "happy, joyous" side.

Yes, indeed. I've been following this thread with great interest. I believe I'm a right side person, allowing for the occasional day on the left when my bird photos turn out to be duds.

Thank you Bill and Manfred.

[drjuice]

Hi, DDK -

Below, I've bolded the two sentences I just added.

In what I've read through I found that a number of replies assumed you know what a histogram is about and that may not be the case. A histogram is a statistical visual representation of the frequency of values in a large data set. Some shade of gray will show up where e Sun moving across the sky causes mountain ridges to cause the eastern side of the mountain to be cast in a shadow and the western side to be in full sun. Even if everything on both sides of the mountain is literally the same color, when the eastern side is in shade, it will be a different representation of the same color, usually darker. So, your histogram would show two spikes, one for the brighter color and one for the darker color. Of course, that's a kind of extreme situation, but it should get you started thinking about what your histograms look like.

The total number of different colors that conceptually can be represented in most camera systems is 256*256*256 or approximately 16 million. The color(s) that appear more frequently will cause the very narrow column associated with value for that color to stick up higher than the numbers for other colored picture elements (pixels) having a smaller number of pixels of that color.

However, none of us (unless somebody's working on that big new sensor that was announced a couple of weeks ago and hasn't told the rest of us) can take a picture that actually contains 16 million different colors, simply because of the size of our camera's sensor. If you had a sensor measuring 1024x1024 and didn't do any post-processing with a camera, you could represent 1 million colors out of that 16 million total. But, the other colors don't get represented in the image or in your histograms.

In histogram pictures, the y axis represents the number of pixels of the given color at each of the 256 positions on the x axis. 00 is the least value of any color for any pixel, and FF (the base 16 equivalent of 255) is the very brightest of that color's pixels.

So, in the histograms that William W uploaded at #3 in this thread, the left to right distance on all of the histograms will be 256 units (00 to FF) while the lower three have the same bottom to top scale along their y axis and the overall histogram at the top has a very much greater range on its y axis

As an experiment, take the following pictures without a flash and without any automatic or manual adjustment of the white balance for any of the environments and be absolutely certain to write down the order in which you took each one, for example, a-c, a-d, a-e, a-f, a-g.

a. Any person (head shot works best)
b. Any object (a single flower in a vase works best)

in the following environments:

c. Incandescent light (most people have at least one somewhere in their house or apartment)

d. CFL (compact fluorescent light; again most people have at least one)

e. Out-of-doors, not in direct sunlight

f. If you have a flash unit or built-in flash, put your diffuser on it (or, if you don't have a diffuser, wrap a clean tissue around it so that there's only a one thickness of the tissue over the flash part of the unit, and attach to your flash unit using blue painter's tape so you won't have sticky goo left on the camera flash housing or the flash unit housing).

g. Mercury vapor lights (in the big city) ONLY for the person picture, take one under this type of light (it should make the person look almost baby-**** green).

I'd make a protected copy of all my images before starting. Customarily, I put those images in a folder called "holding" inside the folder where the images from my camera are resting.

Load all the images into a directory name LearningHistograms so you can toss everything when you're finished. Make screenprints and then crop them so each one of them doesn't need a whole page and you can print and label the histograms.

You should have ten images and their corresponding histograms before you start looking at them. DO NOT look at them with your collection of reading materials just yet!.

Start by selecting one of the images with a high of one color. Then, look at the other images in that set and see what's different.

If you look at the histograms for William's first image you'll see that the blue histogram has a very small pixel count on the right side of his histogram while both the green and red histograms have a maximum number more towards the middle of their histograms. Why the red and green have quite similar relative distributions is that a lot of the pixels in the image are yellowish. That may be because of the light source. But you can see the resulting effect on the histogram as you change the histogram for the green colors.

Choose one of your two image sets and look at their histograms. For the incandescently lit picture, you should either see more green or more red and green depending on the skin tones of the person. Pick a particular value and slide the corresponding histogram bar left and right up and down while watching what happens to the corresponding image. Do that for each of the three image component colors.

Next, do the same for your other set.

You should have a pretty reasonable idea now what you can do to cause one effect or the other for the light sources you're working with. When you've figured that out, I'd suggest going out and taking a bunch of pictures of the type you normally take and doing a similar exercise with those pictures.

Hope this helps.

virginia

[Glenn NK]

If anyone doesn't know what a histogram is and why it's useful, this may help:

http://schewephoto.com/ETTR/index.html

Also:

http://www.luminous-landscape.com/tu...se-right.shtml

Glenn

[William W]

If one shoots RAW, the camera doesn't put out a JPEG image on the CF/SD card, but it does create a JPEG to display on the LCD (RAW not being an image that will display).

The JPEG image used for the LCD has been processed . . . etc

Glenn,

Thank you for following up and explaining why you use "Contrast" set to minus three for the JPEG file, even though you use the raw file for you final image.

In this previous thread I made similar points – including mentioning the Blinkies.

What I wrote there is applicable here, too.

REF Post #13

WW

[Glenn NK]

Glenn,

Thank you for following up and explaining why you use "Contrast" set to minus three for the JPEG file, even though you use the raw file for you final image.

In this previous thread I made similar points – including mentioning the Blinkies.

What I wrote there is applicable here, too.

REF Post #13

WW

Hi Bill:

Actually I use the minus contrast BECAUSE I shoot RAW. If I shot only JPEG, It would be a zero. And the same things would (likely) apply to the blinkies.

I was thinking of Manfred when posted that because I believe he shoots both RAW and JPEG at the same time. I wonder what he does when he shoots both. It seems that one can't use ETTR (and get the full benefit) when the camera is writing both types of file.

It never concerns me at all because I shoot only RAW.

I'm sorry, but I didn't read the linked post until just now - and that's right as far as I know.

Glenn

PS - we're getting quite a few queries about exposure, and I'm wondering if it's time to have an exposure sub-heading on the forum.

[William W]

Double post, sorry.

[William W]

Actually I use the minus contrast BECAUSE I shoot RAW.

That is what I thought my words meant.

I meant: you set out to use ONLY the raw file for your final image - but you purposely set the contrast to minus 3 for the JPEG file - and the reason for that is - because the JPEG LCD Image and Histogram and Blinkies will be more accurate for you when you use those features on the camera.

As I understand - EVEN THOUGH you choose to "select" only raw in the camera's selection - the camera will still make a JPEG file - otherwise you could not view the preview on the LCD or the Histogram or the Blinkies.

It is worth noting that HTP (High Tone Priority) features also in this scenario. i.e. to say setting HTP affects the JPEG file which in turn affects the Histogram and Blinkies.

***

I was thinking of Manfred when posted that because I believe he shoots both RAW and JPEG at the same time. I wonder what he does when he shoots both. It seems that one can't use ETTR (and get the full benefit) when the camera is writing both types of file.

I can’t answer for Manfred but I do shoot raw + JPEG(L).

My answer is in that other post to which linked and I have repeated it here:

“I use Canon DSLRs and capture ‘raw + JPEG(L)’. Because sometimes I use the JPEG image with minimal Post Production, I set the Picture Style to suit the JPEG output that I require. I have made several (arduous) tests and I now have a set of general lighting scenarios which will then determine what Picture Style I will choose to use.”

In other words, I do not use the LCD Image or the Histogram all that much, as reckoning tools.

WW

[Glenn NK]

Bill:

Read your post - we're on the same track.

Except that I use the RGB histo quite a bit, but I'm shooting only RAW files.

If I shot JPEG, I would not change the contrast setting, and I would not use the histo.

Glenn