5D and the histogram

OlgaJ

I'm having a heck of a time with the 5D and the histogram it shows you after you take a shot. I am getting a lot of images with blown highlights although I get no such indication from the camera's histogram.

So, first, let me ask this: Why would the histogram from raw be any different than the histogram from jpg in DPP? The in-camera histogram, after I take the photo, does not show me the raw (per DPP) histogram, but the jpg (per DPP) one.



When I look at the JPG file with Photoshop or Breezebrowser, the histogram looks like DPP's Raw file's histogram, not like DPP's JPG file's histogram. I end up with blown highlights if I rely on the camera's histogram.

Here's another example.

Two lower left histograms are from camera. PS and BB histograms of the same jpg are there too.



Anyone else with the 5D having the same trouble as I am? It seems that with a good histogram, I have to dial in -1/3 or -1/2 of EC if I want to avoid blown highlights.

Thanks!

erich6

It's possible you are seeing the difference in how various software scripts bin the histograms. Think of it as how finely sampled the histograms are. Depending on the amount of binning you will get slight variations in the histogram shape.

The camera should be giving you blinking dots in parts of the image where you are saturating. Otherwise, if you take a picture and see the histogram pushing to the right it's generally a good idea to EC 1/3 to a full stop lower than the camera meter. This gives you plenty of headroom

I seem to remember reading that RAW shots typically give you about -1 stop and +2 stops in post processing so you can always "push" the exposure in software to get what you want (or apply local contrast/brightness enhancement).

Erich

colourbox

It seems to depend on how a specific camera model is programmed to calculate the histogram. I've read that many camera histograms are calibrated to represent the way the in-camera circuitry produces JPG, though I certainly don't know what the 5D is doing (I have an XT). Another variable is whether the histogram you're looking at is representing sRGB, Adobe RGB, or something else entirely. That's important because if a color clips in one color space but not another, and the camera histogram displays using the color space that doesn't clip that color, but you convert the Raw to a color space that does clip that color, the camera histogram wouldn't give you any warning (though the raw converter histogram should indicate the clipping; Adobe Camera Raw's histogram and clipping displays do change if you change the output color space).

I read this long and complicated but interesting thread, which includes some discussion of histograms. The sense I got from that thread is that you might want to run some old-school tests to determine what your camera's histogram really means when it indicates clipping, similar to the way people used to take a test roll of film to figure out what certain exposure settings actually produced on a specific type of film.

Red Bull

Is there a site with a very in-depth information about histograms and how to read them. I've read the page on Luminous Landscape, but I still don't get it. What should the histogram of a well exposed image look like?

erich6

Red Bull wrote:

Is there a site with a very in-depth information about histograms and how to read them. I've read the page on Luminous Landscape, but I still don't get it. What should the histogram of a well exposed image look like?

Here's one I think is pretty good....

http://www.cambridgeincolour.com/tutorials/histograms1.htm

Red Bull

That was a very good explination on the histogram. Just what I was looking for. Thank you!

OlgaJ

erich6 wrote:

Depending on the amount of binning you will get slight variations in the histogram shape.

But I've never seen this type of variation before, and believe me, I've had more digital cameras and dSLRs than Andy.:D I've never had an LCD histogram fool me as much as the 5D's does.

OlgaJ

Red Bull wrote:

Is there a site with a very in-depth information about histograms and how to read them. I've read the page on Luminous Landscape, but I still don't get it. What should the histogram of a well exposed image look like?

Here are the ones that I've bookmarked over the years:

http://www.quiknet.com/~frcn/Histograms.html
http://shortcourses.com/how/histograms/histograms.htm
http://www.luminous-landscape.com/tutorials/understanding-series/understanding-histograms.shtml
http://www.earthboundlight.com/phototips/histogram.html

Andy

OK I made a test and find no inconsitency

The background of this grab is ACR, the inset left is DPP histo, and the inset upper middle right is the camera's histo.

OlgaJ

Thanks Andy!

I should add that I have several examples that don't show that inconsistency. It doesn't happen all the time. I'm still trying to figure out what triggers it, when it does happen.

QuinnPorter

I don't have any of my equipment with me to test my idea, but could it be your picture style setting or your contrast setting. I'm guessing your in-camera JPEGs are being processed with the lowest contrast setting.

OlgaJ

QuinnPorter wrote:

I don't have any of my equipment with me to test my idea, but could it be your picture style setting or your contrast setting. I'm guessing your in-camera JPEGs are being processed with the lowest contrast setting.

While that could explain the different Raw v. JPG histogram in DPP, it doesn't answer the question of why that same jpg (with low contrast and any other parameters I diverted from) produces a different histogram in PS and BB (and other programs) than in DPP and in-camera.