Nature Photographers Network
This is the preserved archive of the original Nature Photographers Network. The community continues at naturephotographers.network.

can you rely on a Histogram?

Discussion

4 replies
Posted by Prashant Khapane on April 24, 2009 at 7:30 AM
If I understand correctly RGB histograms are gamma encoded based on the camera-generated JPEG settings. Can you then rely on them for critical exposures? What is your ETTR technique? I do not rely on them particularly while using GNDs. What is your take?
Posted by Matt Anderson on April 24, 2009 at 8:47 AM
Camera histograms are only as accurate as your White Balance. Change the WB, you will see the histograms change. Thats why it is imperative to get your WB set as accurately as possible, that is, as close to your final intention in post processing. Otherwise, you may experience channel clipping. (Red flowers are a great example of needing to accurately nail down your WB prior to setting your exposure)
Posted by Jerry Greer on April 24, 2009 at 10:32 AM
Prashan,

Make sure that the histogram is set to RGB and not luminance. This way you can see and control all 3 channels to insure that none are blown. Also, my camera never leaves the custom WB setting of 5250K. I only change the balance/colors in Photoshop. There are many reasons that I do this. one is that I hate overly warm images!
Posted by Prashant Khapane on April 24, 2009 at 12:16 PM
Thanks Matt and Jerry,
I do carry a white balance card (WhiBal. smallest) however it is not always feasible to use it.

Hey jerry, I also set it to 5250, guess you mentioned it somewhere before.
Posted by Chris Rice on May 8, 2009 at 1:56 PM
Jerry's right about the RGB so you can see all three bands overlaid. The histogram is a visual chart of the distribution of your pixel values. In statistics, a normal distribution is when all values tend to converge on the mean, known as a Gaussian distribution. If your histogram looks like this distribution (bell curve), then your exposure should be spot on, except this won't ever be the case as we tend to shoot things with millions of colors and brightness values, so the histogram will have outliers based on these extreme values (not a 'smooth' curve, but one with lots of peaks and valleys). This si fine as long as the bulk of the bars are near the center. But any histogram that is skewed right will be overexposed and skewed left, underexposed.

Hope this helps. Sorry for the geek talk, I've been working with statistics lately so everything comes back to it....