Unless you use a specialized camera (e.g. the Betterlight scanning back), the histograms that you’re seeing on the back of your camera are not presented in terms of the primaries of the camera’s native color space. The way that most digital photography experts put it is that the histogram you see on the back of the camera is computed from a JPEG image.
If you’re using a camera that computes the histogram before the exposure is made, like a mirrorless interchangeable lens camera, I don’t believe that the above explanation is strictly accurate, although it’s on the right track. What I think is happening – and the camera manufacturers are not particularly forthcoming, so I don’t know for sure – is that the histogram that you see is derived from part of the image processing chain that produces the JPEG file. I think that the image processing ahead of the histogram computation shares many operations — demosaicing, conversion to a standard color space (in most high-end cameras, the choices are sRGB or Adobe 1998 RGB), white balance, contrast settings, and special effects — with the JPEG processing. To produce the histogram, the camera subsamples (to make the computation easier), and assigns pixels to buckets. To produce the JPEG file, the camera performs the 8×8 discrete cosine transform JPEG compression and formats the data in the standard JPEG file format.
If you’re using a DSLR and using live view, the above applies. If you aren’t using live view, you make the exposure and then you look at the histogram. In that case, even if you have your camera configured to save the image in raw format, it’s likely that you are looking at a histogram created from the JPEG preview image in the raw file.
If you’re trying to use the histogram in the back of the camera to get the correct exposure using the expose-to-the-right (ETTR) method, having the camera present the histogram of the image after extensive processing is not at all what you want. Almost all the time, the camera’s histogram will show clipping at the right side of the image well before the actual raw data is clipped. You can make your histogram less inaccurate by choosing the widest color space available (usually Adobe 1998 RGB), the lowest contrast setting available, and getting the right white balance. It may seem silly to set white balance that will have no effect on the data in your raw file, but you will see a much closer approximation to the true raw histogram, which is what you want to see if you are using the histogram to aid in ETTR exposures.
Next: Image processing for in-camera histograms
Sergey says
Such great insights in every and each one of your articles. I will try setting up my Sony A7 to shoot with Adobe RGB and minimum contrast settings. Will also need to make import presets in Lightroom to compensate for such changes.
Jim says
“Will also need to make import presets in Lightroom to compensate for such changes.”
Maybe not. Those changes won’t affect the raw file, just the preview and the histogram. Don’t forget, ETTR only makes sense at base ISO.
Jim
Brian Patterson says
I shot PP Off video in the Adobe RGB space on my a7iv well under the ETTR limit and it still overexposed highlighted. Any ideas?
JimK says
Sorry, I know very little about video.
Xuan Zheng says
someone on capatureOne website vouched for sRGB colorspace in-camera provides better accuracy, mind elaborate why choose AdobeRGB? currently my red balance is 1.02 and blue balance is 1.
contrast =low saturation =low and sharpness = normal
JimK says
There is no way to get close to a raw histogram on your EVF without a calibration procedure that leaves your EVF white balance looking green in D50ish light.
Xuan Zheng says
Hi thanks for the reply, so simply answer, once I had archieve UniWB(red=1.02, blue=1.0), should I go for AdobeRGB or sRGB for the best blowout(ETTR) accuracy?
do I have the correct contrast, saturation, and sharpness settings? SONY A7R IV. thanks a lot
JimK says
The main thing is to get the WB right. After that, I don’t think it matters much, but low contrast, saturation, and sharpness can’t hurt, and might help.
Xuan Zheng says
I had zebra meters set to flickering at 100+ in my sony.
JimK says
That’s a good move.
Xuan Zheng says
ok, maybe I didn’t express my concern right, once had zebra=100+, contrast =lowest, saturation=lowest, sharpness normal, and in UniWB, the closest bet would be AdobeRGB right? since it somewhat closest to the RAW RGB clipping point?
JimK says
I use Adobe RGB. But I don’t think there would be much difference with sRGB. The primaries aren’t that far apart.
Xuan Zheng says
Thanks a great deal. will sharpness also affect the clipping point(this is the only thing I worry about), I understand that lowering the contrast, saturation will help minimize the JPEG processing over/under messing with the RAW channels clipping point, but for contrast, I’d have a hard time to set an optimal value, some people who uses UniWB vouched for the sharpest setting, while others like you prefer the lowest value?!
Again, your UniWB really helped my ETTR approach.
JimK says
I would suggest avoiding unnecessary processing of the JPEG, which can affect the histogram in ways unrelated to the raw file. That includes sharpening.