If you’re a RAW shooter you should already know about ETTR, Expose to the Right. The theory behind ETTR is that the best image capture for the most possible information is when the histogram is pushed to the right, to the “bright” side. Just keep adding exposure until that histogram is over to the right side of the graph.
OK, but how far to the right? You definitely don’t want to clip the highlights. DSLR cameras have a clipping warning display, the “blinking highlights,” or “blinkies” as they are commonly called, which shows up on the camera’s LCD. Many cameras will even display the blinkies for each individual color channel, besides the composite luminosity. But remember that the image displayed on the LCD is not the actual RAW file; it’s a jpeg thumbnail created on the fly by the camera. Camera manufacturers have coded in some headroom with the blinkies, as they don’t want customers to be angered at blown out highlights.
Fine. But I would suggest running a test to determine exactly what the correlation is between when the blinkies start, and the actual clipped highlights in the RAW file. You can easily run a test to determine this. Set your camera to Aperture Priority, lock it firmly on a tripod, and aim at any scene. Increase exposure until the first blinkies appear. Note this frame (probably the easiest solution would be to delete any previous frame you shot to get to this point). Now shoot several more frames, using Auto Compensation to add 1/3 stop to each successive frame. Open this series in your RAW file software, such as Lightroom or Adobe Camera Raw, turn on the clipping warning in the software, and check each frame. I’ll bet that the starting frame, the one the camera blinkies said was clipped, is actually not clipped at all. In fact, you might be surprised at just how much headroom you have. With my particular Nikon D800E and a medium toned test subject, I have to add 1.3 stops beyond the blinkies before the RAW file has clipped highlights.
So what’s the point of doing this? Why worry? Well, with digital capture, noise lives in the dark exposures. If you want the best possible data, start with the best possible exposure. With my particular camera I’ll add some extra exposure whenever the subject is such that I can, especially when working at higher ISO values where noice is always a problem. Recently I shot a landscape at ISO 1600. I shot at both the metered ETTR exposure, and at my “extra 1.3 stop” ETTR settings. The difference was remarkable. The first image needed noise reduction. When I looked at the second shot, the one when I had added 1.3 stops, the image on the LCD appeared almost washed out. But when I reduced the exposure in Lightroom (my standard RAW file software), all the noise was gone. In fact, my tests suggest there is even a slight difference when the camera is set at base ISO 100, where I use the camera the most.
Does this really matter? The answer depends on how compulsive you are about quality, the realities faced in the field, and on how the photograph is to be used. Just remember, ETTR is for RAW capture only. And you don’t want to lose an image by blowing out the highlights. There are indeed some scenes with the highlights already maxed out.
Once you know how the blinkies in your camera correlate to the actual RAW histogram, a simple and safe solution — particularly for static subjects such as landscapes — is to set your camera to bracket another frame that is 2/3 stop more exposure than the ETTR histogram on the LCD. When I shoot with my D800E, I know that I’ll be adding that extra bit of light whenever I can.
Extraordinary tip from the best cutting edge thinker we have in photography. Thank you, John. I owe all my photography technical knowledge to you. You continue to inspire me. I appreciate you setting the pace.
I appreciate the extremely nice comments. Thank you.
Thanks for the tip. I’ve known about ETTR and use it regularly but was not aware of the manufacturers head room.
Thanks John. In some pictures (e.g. a field of poppies), even if I set the exposure so that the COMPOSITE luminosity histogram is close to the right end (without blinking), the individual RED channel histogram has blown out highlights. I need your help to understand if such exposure is correct or not. When we expose to the right, do we have to consider the composite luminosity histogram only? or also each individual color channel histogram? Thanks for your fantastic books.
Take a test photo so that one of the channel histograms (such as the red channel you mention) shows the blinkies. Open the file in ACR or Lightroom’s Develop module, and see if the channel is actually clipped, or if it can be recovered.
That is an insightful tip John. Thank you!
I’ve always wondered why I am able to recover blown out highlight in the sky/clouds using ND filters in LR. Your write-up gives me a simple explanation.
I will certainly try your simple exercise to figure out exactly how much headroom I have.
Great thinking, John!
My D4 showed one full stop of headroom relative to reference ACR and CS6 histograms.
I have yet to actually study some highlight detail to ensure that those reference histograms are accurate. Have you any thoughts on that?
The histograms in ACR and Lightroom’s Develop module are indeed accurate, since these are the histograms of the actual file. If there is highlight detail, it is available.
Hi John, very helpful info as always. Would be very interested to hear your thoughts on Uni White balance and ETTR? BTW, have you made any decisions about your new book yet? Also, a very happy, healthy and creative new year to you and Pam. Best regards, Ash & Ger
As a novice I must be missing an important point contained within your post that you could help me with.
If you are shooting at high ISO and have the ability to over-expose then why not simply drop ISO?
If the scene and subject give you the opportunity to over-expose in the first place then by definition you have some latitude with the exposure which implies that you would be able to drop ISO so the device itself delivers improved IQ.
Is there some other benefit of pushing the camera to higher ISO that you’re trying to obtain?
The point you’re missing it that ETTR does not mean you are “over-exposing.” ETTR is a way to record the most data possible, and consequently the best exposure possible. If you’re shooting RAW, at any ISO you ideally want an exposure so that the histogram is pushed to the right, since that is where the most information lies. ETTR does not mean going to a higher ISO. Shoot at ISO 100, I want the histogram to the right; shoot at ISO 1000, I still want it to the right.
John, it is possible to calibrate your camera such that the camera’s histogram is derived from the RAW image and not the jpg image. Do a google search for “uniWB” if you care. Sounds like you already have workable solution though.
Even with uniWB the histograms are still derived from the jpeg image, but they are closer to what the actual RAW file would be. However, all the images now have a strong greenish color cast which drives me crazy. Since my white balance default import setting in Lightroom is “as shot,” I either have to evaluate images which are off-color, or reset WB for each individual shot.
I didn’t even realize I was doing ETTR, but on my cheap Nikon D5100 I started to overexpose +1\3 to +2\3 almost all the time with matrix metering and even at ISO’s that would make most people squeamish, that noise disappears when I pull he exposure down and add some black in Lightroom. So much data on the right side of the scale!
Recent ETTR posts regarding HDR subjects suggest that multiple exposures – starting with only the right 20% (approx.) of the histogram showing data, and then backing off one stop at a time until a “normal” histogram appears, then blending the images into a single 32 bit image in Photoshop will produce a superior image file. Any thoughts? Thanks John – Your “Landscape Photography” and “Focus on Nature” rarely stay on my bookshelf for long!
I’m not sure what comprises a “superior” file. I rarely do HDR, mainly because my newer cameras can capture such a wide dynamic range. If I can record all the data in one shot, I see no reason to take more.
I’ve done some experimenting with a thought similar to this. I take the same shot with two different sets of exposures to see how well the HDR software handles it. The first set I do one correct exposure, one 2 stops over, and one 2 stops under for a three shot HDR. The second set I do the same thing and throw in two additional exposures for a five shot HDR. The two extra exposures are one 3 stops over and one 4 stops over. The HDR software produces much less grain and much more detail with the five shot HDR that has the two extra overexposed frames (I think because of all the extra data sensors pack onto the right of the histogram). I can always add more black and shadows for contrast, but if you have to add whites and highlights you start introducing more and more noise. I’ve done the opposite and ran the five shot HDR with two extra underexposed frames and they had almost no effect compared to the standard three shot of the same scene.