Digital Light & Color

Curious about the Histogram display after cropping an image.

I have an image that has 2:3 proportions. The Histogram Tool shows the histogram predominantly on the right side, and actually bumping up to the right edge. Along this right edge, the histogram does not go all the way to the top, as it is "stepped down" lower than the pixels immediately to the left. This "stepped down" portion of the histogram is the topic of this post.

I cropped this image to a square proportion and, as I was placing the crop into a 9 x 9 image panel, I was comparing the histogram of this new crop to other images to see how compatible they were.

(a small tip for those who do not know - if you have multiple images open and you wish to compare their histograms, you can click on one image, then click on the Histogram Tool to display its histogram - however, if you then click on the Tool again, the button will "de-activate" but the histogram will remain - you can then click on a second image and again click on the Histogram Tool and generate a second histogram for the second image. This can be done for multiple images as long as you remember to click on the button to de-activate the histogram between each image - if you do not, each time you click on an image, the histogram itself changes to reflect the active image).

Anyway, in comparing the 2:3 image to the new cropped image, I noticed that the right edge of the histogram (where the "stepped down" portion was) was now higher than in the original image.

I understand that a histogram will change and show a different distribution of tones for a cropped segment of an image, but I don't understand how the cropped version would show more height on the right edge. It suggests, to me at least, that the cropped image, although it actually lost pixels, added some brightness. Other than the cropping, I made no changes to the second image.

Is there an explanation for this curiosity?

Marv

Histograms are scaled automatically so the largest value goes all the way to the top. Thus, you cannot look at two histograms side-by-side and see which ones have higher or lower values. For example, if you cropped out an area that had a lot of one value which was causing a big spike in the histogram, you could remove the spike that was forcing the histogram to a smaller scale and all the other values would get larger. Also, make sure you were using the same level of histogram expansion in each of the histograms you want to compare.

Aah, that explains things. I did not realize that each Histogram is scaled so the largest value reaches the top. Makes sense though now that I think about it.

Thanks, Jonathan