How to Read and Use Image Histograms

Published on May 25, 2011 by Michael Zhang

PetaPixel

How to Read and Use Image Histograms

May 25, 2011 · Michael Zhang

Here’s a great video lesson by Greg Cazillo on understanding and using image histograms to improve your photography.

(via Fstoppers)

http://twitter.com/lifebypixels Mark Heaps

Ha ha, love his commentary between the educational parts…”…if anyone tells you that, they’re full of crap”. Nice.
http://blog.stopped-motion.com Amber

One of the things he says is “Shoot to right” and he says that “if you’re under exposed you’re losing data.” This doesn’t sound right to me. I always heard the opposite was true when shooting a digital because the image isn’t inverted.

Any thoughts?
http://www.facebook.com/Xvant Pedro Rojas Jorquera

that’s because in the instruct video also explains that most of the programs to develop RAW files are programed to recover as much information from the light área of the histogram, and that the black levels are much more easy to compensate in this programs, like lightroom, and the black slider, contrast, etc.
Anonymous

That’s true. And because of the way digital sensors work they contain exponentially more information in the bright areas than in the shadows. So if you expose to the right side of the histogram and then use software to recover that information your image will have less noise in it.
It is completely different to the film days.
Nicola

I echo what Amber said … what I understood.
Nicola

I echo what Amber said … what I understood.
http://twitter.com/ericsknape ericsknape

Nothing like watching a video of someone talking in front of a dry-erase board … and using Wikipedia definitions. Great information, for sure, but really could be disseminated in a more interesting an memorable way.
ShamB

Shooting to the right does nothing (except ruining color accuracy). The way data is arranged in a RAW image matches the way we see… we dont see the noise in dark areas because our eyes cant see into the dark (which is why RAW uses less data there), and we are very sensitive to bright noise (which is why RAW uses lots of data there).

By shooting the the right all you do is move non-visible ‘dark noise’ areas up into bright areas where the noise now becomes more apparent, and you NEED the addition data contained in the bright areas just to make the noise look the same.

The only advantage is shooting to the right is if you intend to digitally manipulate the image later (especially dodging).

See http://howgreenisyourgarden.wordpress.com/2011/01/10/shooting-to-the-right-and-why-it-can-be-a-bad-idea/
http://www.imrezbalint.com/ Imre Z. Balint

Not that long ago I made a video like this: http://www.youtube.com/watch?v=0U-xwC4AEcQ and a blog post to go with it: http://binarygraphite.blogspot.com/2011/03/part-2-histogram-photography-with-imre.html
http://twitter.com/cazillo Gregory Cazillo

Thats incorrect, its a well known fact that 50% of the image data is in the brightest stop. Google it ;)
Futuremedia

Hi Gregory.

Sigh.
Yes it is and yes I have previously. I understand the point enough to know its not actually relevant: data s/n in an image is not proportional to how much noise we perceptually see because our eye response is *not* linear! That is the exact reason why the camera response is also not linear!

1. If a camera took an image with no anthomorphic bias (totally linear response) with just as much noise in the shadows as the highlights, we would percieve more noise in the highlights because our eyes are more sensitive at the brighter end.
2. The camera has to cater for (1) by using more data when capturing highlights.
3. By shooting to the right, you reduce the signal to noise considerably (which is the point you make, and correct *but only in terms of amount of data*).
4. Because our eyes are more sensitive to bright images, shooting to the right actually makes us *more* sensitive to the noise at about the same rate that the s/n gets better – so there is no point in shooting to the right, as we see the lower s/n better, and therefore we actually see the same amount of noise (my point).
5. To which you will say ‘ah, but after shooting to the right, I underexpose back to actual in post, so it doesnt matter: there is now more data in the image AND it is the same brightness!’
6. And I will say ‘yes, but in doing so, you forget that the exposure response for every camera is actually different for each channel (R, G, B), so reducing exposure *also compromises color accuracy. You get more data yes, but it is now wrong (try it with a color card to see for yourself)!

Upshot: there is no free lunch except in very particular post processessing operations: shooting to the right has very little advantage.

Far better to just shoot for the key, and forget shooting to the right.