Histogram - Long, flat and wide

[Brownbear]

What does it mean when your histogram stretches the full length of the graph (wide tonal range?) BUT lies very low at the bottom of the graph? ie; no peaks of light or darkness anywhere... almost of uniform height (of very short stature)

I have read the tutorials on histograms, but I can't figure this one out. Uniform lighting? No contrast?

Thank you in advance.

[davidedric]

Hi Christina,

Could you post the image, with EXIF if you can? So we can have a look.

Dave

[DanK]

The horizontal axis is luminance, running from black at the left and white at the right. The vertical axis is a count of pixels. So, a flat histogram means that the count of pixels is quite uniform across the range of brightness. There are two tutorials on this under the tutorials tab on this site. From the Tutorials menu, pick editing and postprocessing.

In general, the tutorials on this site are superb. Some of us found the site because we were pointed to them.

[Inkanyezi]

Actually, that histogram doesn't say a lot. Most histograms don't. The caveat with the histogram is that it doesn't say a thing about where those pixels appear. However, if you have it spread all over, it means your image does have decent contrast, that there are pixels of all or almost all kinds somewhere in the image.

If I could generalise a bit about histograms, a good contrast one often is rather even, with no large peaks, but most of it more or less equal height. A very high contrast histogram can have two peaks, one far to the right, the other far to the left. A low contrast image has one, only one peak, and its position is at the density level of the low contrast image, while the rest of the histogram might be flat at the bottom. A high peak to the left will tell you that there are rather many dark pixels.

In most of our images, we can disregard anything in the middle and at the left side of the histogram, but the very rightmost part of the histogram may have a history to tell. Generally, we shall not let it climb the right wall - unless we accept that parts of the image are burned out. We have that creative liberty. But in most cases, even though me may accept a few pixels hitting the right wall, the histogram should be decreasing, sloping downward toward the lower right corner. There shouldn't be a large empty flat space to the right, unless we deliberately make the image low key. But in all those cases, the most information we get out of the histogram is that lower right corner.

But apart from that, the histogram does not supply much valuable information. Evaluation of the image on a calibrated screen tells a lot more. The histogram, just as highlight blinkies, is only a tool that may give some advice on whether exposure was within the ballpark or if we blew it.

[Brownbear]

Thank you everyone.. Yes, the tutorials are truly excellent (thank you Cambridge) but I am trying to learn how I could have done this shot (and others) better, and for some reason I can't make the connection between the histogram and this image

Here is the original image in a screen shot (A 5.6, iso 800 exp. +.7... aperture priority)

[Mike Buckley]

I agree with what others have said and perhaps it might help to say the same thing in different words. The histogram that you described indicates that all luminosities are represented somewhere in the image and that all of them are displayed in relatively equal amounts.

However, I don't understand why the histogram displayed in your screenshot appears as it does. I saved the screenshot to my computer and cropped the image to include only your photo. Notice the histogram that is displayed in my primary editing software. A very similar histogram was also displayed in my cataloging software.

[Brownbear]

Yes, putting it in different words is very helpful. Thank you Mike.

Sorry, I don't know.

[Steve S]

Hi christina. If you bring up the curves tool, an eyedropper will appear, when you move the curser over the photo. If you click and hold, you will see where that point is, on the histogram in the curves box. (this is helpful when making adjustments to a specific point in an image) Seeing where the points fall on the histogram, may help you, in understanding it.

[DanK]

Christina,

but I am trying to learn how I could have done this shot (and others) better

You didn't do anything wrong. The flatness of a histogram reflects the flatness (evenness) of the lighting in the scene. Changing exposure would shift the histogram right or left but would not bunch it up. When you have a really flat histogram, you often have to do work in postprocessing to compensate. For example, you would want to increase contrast.

Re the shape of the histogram that Mike got--I don't think it is really different. I got one almost exactly like his in Photoshop. However, if you look closely at his and yours, you will see that the shapes of the histograms are similar. Both show more density in the midrange and less in the extremes (left and right). What differs is the vertical scale. Since there are no numerical values on the Y-axis--there is no set number of pixels represented by the top of the axis--the histogram is really only showing relative density across the luminance scale on the X-axis. Given that, it would make sense to scale the histogram to fill the vertical space, in order to best show the variation as you go from left to right. I am guessing that Photoshop does this scaling and that Picassa, for some reason, doesn't.

Dan

[Brownbear]

Thank you... First time I've done this, great to know. very helpful. thank you.

Hi christina. If you bring up the curves tool, an eyedropper will appear, when you move the curser over the photo. If you click and hold, you will see where that point is, on the histogram in the curves box. (this is helpful when making adjustments to a specific point in an image) Seeing where the points fall on the histogram, may help you, in understanding it.

[Brownbear]

wonderful explanation. thank you Dan

[praphul]

Christina,



You didn't do anything wrong. The flatness of a histogram reflects the flatness (evenness) of the lighting in the scene. Changing exposure would shift the histogram right or left but would not bunch it up. When you have a really flat histogram, you often have to do work in postprocessing to compensate. For example, you would want to increase contrast.

Re the shape of the histogram that Mike got--I don't think it is really different. I got one almost exactly like his in Photoshop. However, if you look closely at his and yours, you will see that the shapes of the histograms are similar. Both show more density in the midrange and less in the extremes (left and right). What differs is the vertical scale. Since there are no numerical values on the Y-axis--there is no set number of pixels represented by the top of the axis--the histogram is really only showing relative density across the luminance scale on the X-axis. Given that, it would make sense to scale the histogram to fill the vertical space, in order to best show the variation as you go from left to right. I am guessing that Photoshop does this scaling and that Picassa, for some reason, doesn't.

Dan

Dan , i would like to know how to boost/amplify those pixels(appearing in the middle predominantly shown by yellow and red) along the y axis. is there any method to do so?

[Inkanyezi]

You might find some of the advice and information a bit contradictory, and to illuminate what I said about a "flat" histogram, with no peaks indicating reasonable contrast, you might take a look at this image: http://uploads.ifokus.se/uploads/659...b/img-3964.jpg

The image is not mine, copyright notice is in its EXIF, and it was originally posted in a discussion thread in foto.ifokus.se. If you look at the image, you'll see that its highlights reach saturation, and the shadows are deep and black. Its histogram is low, low, almost flat.

But as I said, histograms say nigh to nothing about how an image looks, you'll have to evaluate it with your eyes from the screen, and the histogram is only a helper to show statistics of how tones are distributed in the image. You'll have to look at the image to see where the tones are placed.

And Praphul, yes, you can boost those colour channels in that specific place, although I have no idea why you would want to. Contrary to what you might believe, "boosting" them by placing more pixels at that level, creating a peak, will flatten contrast in that particular region, and it would seldom lead to a desired result regarding the image. You might mean something else, as perhaps you might want to emphasize some part of your image? That would mostly be done in an adjustment layer, eyeballing the image while working, to get the effect you want. However, it might look very different from what you think in the histogram that would result. Emphasizing certain tones or colours in a particular region would stretch them out along the histogram x-axis, flattening that particular part of the histogram.

Sometimes, you can use the histogram to see where an alteration of the curves might have the desired effect. Then you should realise that the action you take on the curve will be opposite to the one that will result in the histogram. You can raise contrast by lifting the curve where there is a peak, thus lowering the histogram peak.

And Christina, your image has a high bouldering histogram which indicates that there is a rather decent spread of tonality, although it is a bit "flat" - one high peak is a flat image and one large boulder is a somewhat less flat one. You may raise contrast with the "Curves" tool, by making the curve steeper in the middle, effectively giving it more S-shape, i.e. lifting the right side of the curve and lowering the left, keeping lower left and upper right corners anchor points as they are.

[Mike Buckley]

it would make sense to scale the histogram to fill the vertical space, in order to best show the variation as you go from left to right. I am guessing that Photoshop does this scaling and that Picassa, for some reason, doesn't.

Ahhhh. That would explain the different shapes. I'll have to remember in the future that some programs apparently do that and some don't. Now that I think of it, I believe this issue came up in the past. Very helpful, Dan!

[Dave Humphries]

Hi Christina, Praphul,

I think the histogram should be treated as a tool to answer specific questions only, I do not find it at all helpful to try to make a better image. Also be wary of the luminance/monochrome versions over individual, or overlaid colour versions.

Questions it can answer (particularly if separate RGB);

a) where is the black point?
b) are any/all colours blowing?
c) have I got room (a bit of flat space either side) to apply Local Contrast Enhancement (LCE) without losing image detail in shadows (left) or highlights (right)

... and that's about it.

I view the entire image full screen, or at 100%, to answer other specific questions that arise in PP and I formulate these questions in my mind by looking at the entire image and see what strikes me about it;

i) good composition, does it need a crop?
ii) any distractions need removing?
iii) does it lack "pop"?
iv) what pixel size is it after the crop? this determines what size I should downsize to and post it at and it still look good (after sharpening)
v) is it sharp enough, or over sharpened?

My belief is that if you can't 'formulate the questions' and feel the need to resort to 'measurements' (even graphical ones) then your seeing too much wood and not enough trees/forest and you should stop editing (save it as psd or tif, not jpg) and return to it later. It happens to me too.

Practising critiquing other's pictures (as well as your own) and saying why you think something works, or doesn't, can be helpful, even if you don't actually post what you're thinking here. If you do, make it clear your purpose, we wouldn't want to offend anyone. By all means use a histogram (available in FireFox as an Add-on), but only to answer a), b) or c), not to try to emulate, for the reasons Urban, Dan, Steve and Mike have demonstrated and/or explained above.

Cheers,

[Mike Buckley]

Excellent post, Dave. I was unsuccessfully trying to wrap my arms around what Christina might be trying to accomplish by viewing the histogram, so I intentionally avoided that part of her initial post. You summed up everything very nicely.

[praphul]

tanku Domeiji and Dave

[Brownbear]

Thank you Dave, (excellent advice, although I don't feel I'm ready to critique others photos, yet.. I will try if only in my thoughts)

If I may ask for one last clarification?

For this particular image, I would've liked the ducks to stand out more from the water (pop)... If I used a faster shutter speed the image would have been too dark. (iso 800 is high for my camera). If I upped the exposure comp the water would've been overexposed, and I shot as wide open as I could.

If I had changed my camera settings for the tone compensation to a lower contrast setting, could I have accomplished this? I'm assuming that a lower contrast photo as depicted by the lower portion of the boat photo example (and histogram shape) in the histogram tutorial https://www.cambridgeincolour.com/tu...istograms1.htm may have helped by boosting the highlights?


The upper region contains the most contrast of all three because the image is created from light which does not first reflect off the surface of water. This produces deeper shadows underneath the boat and its ledges, and stronger highlights in the upward-facing and directly exposed areas. The middle and bottom regions are produced entirely from diffuse, reflected light and thus have lower contrast; similar to if one were taking photographs in the fog. The bottom region has more contrast than the middle—despite the smooth and monotonic blue sky—because it contains a combination of shade and more intense sunlight. Conditions in the bottom region create more pronounced highlights, but it still lacks the deep shadows of the top region. - See more at: https://www.cambridgeincolour.com/tu....pLBlZfPH.dpuf


Also when photographing in high contrast situations, ie; dark birds in flight against a bright blue sky.. If I noted that the histogram was of high contrast, could I not change my tone curve to a low contrast curve thus eliminating problems of chromatic aberration?

And in the case of photographing in low contrast situations, ie; foggy days, could I not change the tone curve in my camera to a high contrast curve to fix it right then and there?

I was under the impression, that the understanding the histogram, beyond its use for exposure compensation could be useful for other photographic parameters/variables but perhaps not?

[Inkanyezi]

I was under the impression, that the understanding the histogram, beyond its use for exposure compensation could be useful for other photographic parameters/variables but perhaps not?

I think you got the core of it there. The histogram is just a statistic image, a diagram of how different values are represented in the image. It does not give a lot of information, and above all, it doesn't say a thing about the image; other than whether it is well within bounds of the sensor dynamic range, or if it is heavily over- or underexposed. Of course you will see that just by looking at the image as well, but the more precise levels of over-exposure are where the histogram may tell you whether you could expect more structure and detail when processing or not.

Those particular properties you mention of your image cannot be reached by help of the histogram.

[Brownbear]

Urban, thank you for confirming... I'm not sure why I thought this... And its time for me to move on to learning something else