Histogram question

[Abitconfused]

A histogram that shows BOTH blacks and whites extending (respectively) beyond black point and white point is indicative of what?

[lukaswerth]

This shows both black shadows (without detail) and clipped highlights. It means the dynamic range of the scene extended that of your sensor at that particular ISO. It is left to your interpretation how adverse this is: are the clipped highlights just due to the sun or a light source in the picture which should anyway be white, do you want detail in that shadow where there are none? Are the clipped highlights only of one channel or of all three (a picture of a bunch of red roses easily produced some clipped reds)?

Just to mention it: If you shoot raw, it pays to see whether you can retrieve some highlights of shadows not only by moving the shadow and highlight sliders, but also the exposure slider.

Lukas

[Abitconfused]

Now I can't recall where the photo is (I have eight terabytes here) but it looks quite good despite the histogram. So, the dynamic range of the scene was a little too much for the sensor. I get that. Perhaps three photos -1/2 EV, EV, +1/2 EV and made into a HDR photo would reveal more tones. As to the esthetics, that's another point.

[Abitconfused]

OK, here we go. I imagine the individual color channels could be adjusted here.

[Colin Southern]

Perhaps three photos -1/2 EV, EV, +1/2 EV and made into a HDR photo would reveal more tones.

No. .5EV will do approx nothing. If you're going to construct an HDR photo, shoot at a minimum of 2EV steps.

Keep in mind that the histogram you have is the result of adjustments that you've already made and keep in mind too that there's no ideal histogram; in my opinion, you probably want less detail in this shot, not more (the midtones are looking quite washed out in places.

[Colin Southern]

It means the dynamic range of the scene extended that of your sensor at that particular ISO. I

Not necessarily; it just means that that's the current state of the image -- which isn't necessarily how it was when it was captured.

[chauncey]

Grab a cuppa and watch this...http://www.youtube.com/watch?v=EjPsP4HhHhE

[IzzieK]

Grab a cuppa and watch this...http://www.youtube.com/watch?v=EjPsP4HhHhE

Thanks for this William, I will learn from watching this one too...I'll find a time for me.

[ied]

Thanks for the link. I'll make it a point to watch this as well.

Irene

[dje]

Ed sometimes when you have black or white clipping it can be useful to turn on the exposure "blinkies" in the editing software you are using. This shows the areas of the image where clipping is occurring with a colour overlay or perhaps flashing light.

In your image, the black clipping is probably occuring in that dark area under the bridge and the white clipping may be some highlights on the bridge itself or the bridge in the background. And as Colin says, they may have been caused or exaggerated by the editing adjustments.

Dave

[DanK]

Not necessarily; it just means that that's the current state of the image -- which isn't necessarily how it was when it was captured.

To add to this: that's why it is helpful to check the histogram on the camera if you think your scene may have more dynamic range than the sensor can handle. Since the histogram reflects a jpeg thumbnail, you can make it more accurate by picking a very neutral picture style.

[Colin Southern]

To add to this: that's why it is helpful to check the histogram on the camera if you think your scene may have more dynamic range than the sensor can handle. Since the histogram reflects a jpeg thumbnail, you can make it more accurate by picking a very neutral picture style.

+1

I might add too through that although the in-camera histogram is based on a JPEG - and is thus conservative - that "conservativeness" (is that a real word?) equates to "safety margin". Most composite histograms won't indicate a single channel over-exposure - and in certain types of shots (eg very warm sunsets) you're likely to blow one channel long before others.

At the shadow end it's even more conservative - usually with a RAW capture you can dig quite a bit more out of the shot, although it may get noisy if pushed too far.

[fenix]

Just to mention it: If you shoot raw, it pays to see whether you can retrieve some highlights of shadows not only by moving the shadow and highlight sliders, but also the exposure slider.

Ed,

If you're serious about knowing the true exposure of the sensor in your shot, I would heartily recommend a utility called RawDigger which can show you a histogram of all four raw channels R,G,B,G1. Since the histogram is of the actual raw data you can easily see if the sensor itself was saturated by the exposure and therefore has unrecoverable highlights. I find this a better starting point rather than trying to guess what the raw converter did to my image.

A case in point is IR imaging (OT for most folks) with my Sigma SD10 DSLR. Usually, RawDigger shows no blown highlights in my raw data (I tend to under-expose for IR). When I look at the Review image in SPP's histogram, the green is totally killed *** making the image a horrible purplish-red, as one would expect, with a lot of blown reds. Point here is that the review image is completely unrepresentative of the raw data - also as one would expect; after all, it has just been converted!

*** my Foveon sensor does respond to near IR in all three layers (channels), approximately in the RGB ratio 6:2:1.

RawDigger is also good for sorting through bracketed shots, to know for sure which ones to reject.

Eric

[fenix]

To add to this: that's why it is helpful to check the histogram on the camera if you think your scene may have more dynamic range than the sensor can handle. Since the histogram reflects a jpeg thumbnail, you can make it more accurate by picking a very neutral picture style.

To get even closer to raw on the in-camera histogram, one can set a custom in-camera WB to UNIWB. Not a procedure for the faint-hearted and one could be forgiven for being a bit confused ;-)

But the prize is being able to almost believe the in-camera histogram and thereby approach the grail of ETTR.

UNIWB . . . don't leave home without it.

Eric

[Glenn NK]

Mostly I only pay attention to the individual RGB histograms. And as Colin has aptly pointed out, the Histogram is based on a JPEG so if one is shooting RAW, there is a bit more leeway (more conservative) than with JPEG shooting.