Histogram

Topic: Histograms

What is a histogram and how can it help? – this question was posted by a forum user.

The user noticed that in a large collection of photographs the histograms shown by PhotoPerfect varied a great deal. The ones with very distinct histograms could be optimized very successfully (Xe 847 etc.), but with the ones which have less distinct histogram profiles the optimization didn’t seem to work as well. He wondered what the reason may be for this and how such different histograms already get generated while the picture is being taken and wanted to know how to get the best optimization results with those images which have less distinct histograms. In fact, is there such a thing as a perfect histogram?


My answer:

Preliminary note:
The term ‘tone value’ is used in image processing contexts and simply refers to the degree of brightness of the pixels in a digital image.

There is no such thing as an ‘optimal’ histogram, but you can orient yourself by bearing in mind the following points:

1. With a 'normal' image (this is where the difficulties start – what do we mean by normal?) with an average brightness distribution the histogram should form a clearly defined bell-shaped (so-called ‘normal’) curve.

2. In processing, the histogram should not be compressed or stretched too much, otherwise tone value jumps (i.e. discontinuities) may result. It may help to apply a small degree of softening as a first step. De-noising can help to prevent overstretching the histogram.

3. Don’t use the contrast tools to push the tone values too far to the edges – there is a danger of too much evening out of the values with accompanying loss of information. But don’t push them together in the middle too much either otherwise a similar problem can arise.

4. The contrast, gamma and brightness settings should produce an image which appears as natural as possible – which means using your own sense of judgment and experimenting with the sliders! It is a good idea to store very promising intermediate results as TIFF images!

5. And in general you can assume that no photograph leaves the camera in an optimal state, i.e. every image can be improved in some way. Just think of how much time RAW data converters need even on a fast PC and then compare it with how quickly a camera stores a JPG image (don’t forget, that includes the basic storage process as well) – then it becomes clear that the camera’s ‘lean’ software can’t deliver everything that is theoretically possible, otherwise you would have to wait for ever.

6. Regarding the histograms: This has now been solved (2006); there is now a scaling slider next to the histogram.



The next user was looking for help in understanding how to use a histogram.

PhotoPerfect provides histograms which show how the colors are distributed – but how can one tell which slider one should use and which direction to slide it in so as to improve the picture?


My answer:

That’s a misunderstanding: A histogram only shows tone values (= brightness values), or to be exact the number of pixels which have the brightness of the given histogram step. Tone values range from 0 – left (= black) to 255 – right (= white). By observing the histogram one can see whether one is pushing tone values too far to the edges or bunching them together too much. When this happens it means that previously differing tone values are collected together, e.g. all the values from 240 to 255 are concentrated into 255. That means that there are no more grades of brightness between 240 and 255 any more, but only pure white.

Also worth noting:

Don’t forget to check our FAQs – i.e. Frequently Asked Questions in this forum where we are right now!
You can also find general topics there including information on how to use all the forum features.


General information on using a histogram:

Generally, the histogram shows the way the tone values are distributed (see above). That gives a general impression of the state of the image: Many normally exposed photos display the tone value distribution as a bell curve (‘normal curve’) with its maximum roughly in the middle. If most of the tone values are strongly biased to the left, that indicates underexposure, whilst a strong bias to the right is a sign of overexposure. Of course, that may simply be due to the subject of the picture, where large parts of it are either light or dark. So the histogram gives a first clue as to what may need to be corrected, and if so in which direction.

After this first step towards recognizing what needs to be done, the second step is to select the correct tools for the correction. This step is dealt with via the links posted above.



Note added by another user:
Most high end digital cameras now have a histogram display which can provide the experienced users with useful information about the exposure level.

Cheers

Ralf Eberle

Hi,

A histogram is a graphical representation showing a visual impression of the distribution of data. In other words, it is a summary graph showing distribution of data points measured that falls within various class-intervals. It provides important information about the shape of a distribution.

Thanks