Expose to the right really important?

[sieracki]

I think it might be important in certain situations. I am thinking specifically of long SBR (subject brightness ratio) subjects we often get in the landscape. My Kodak DCS SLR/c has about 6 stops of useful latitude at ISO 200. If I encounter a scene with more than 6 stops I have a decision to make. I think it's best to allow considerable lifting of shadow areas in post processing, so I don't want my shadows to be exposed much below zone III. What I have done is the following:

I shot a sequence of frames of gradually changing exposure in RAW mode. I started with a neutral evenly lighted subject such as a grey card or the blue sky at zenith. If I expose down in 1/3 stops from that point, down as many stops as possible for my shutter speed range, I'm getting frames that have a known exposure relative to 'middle grey' exposure. For instance, I can choose the frame exposed at two stops below the middle toned exposure, this is zone III by definition. I open that frame in Photoshop CS RAW editor. I can set the proper white balance since I roughly know the color temp of the light I shot in (e.g. daylight is 5500K). Now I increase exposure with the exposure slider and view the frame at say 50% zoom. For me, I see that I can move up the slider about 1 1/3 stops before I see significant posterization artifacts. So I conclude that for me, I can place my shadows in Zone III then I can at least move them up by one stop. If I look at the Zone II shot, then I can see that I can move that up by no more than 2/3 of a stop.

I also make use of the frames shot above Zone V. I can take a frame that is three stops above middle tone and move it down 3 stops and it looks natural. If I take anything that is strongly color saturated such as the blue sky, I can see that my movement is less than if I chose something neutral such as a grey card as my subject. I can place a blue subject in Zone VII 2/3 and still recover the frame back to a lower exposure with the slider. So I know I can go out and shoot blue sky as highlight and place on Zone VII 2/3 and I'll get a useable frame.

So if I encounter a 8 stop SBR scene, I need to do something fairly drastic. I can risk noise in my shadow areas if they are small enough or I can blow the highlights if they are small enough. Otherwise I'm going for a grad filter, or thinking of digital blending. I'd rather not blend as I haven't had that much experience or success. If I am at a lower SBR then I have a good deal of flexiblility. I think using the first stop of brightness is a good idea if you expect quite a bit of post processing. Keep in mind also that future RAW converters might do more with these frames that are blown and seen too far gone for us now, but that does nothing for useable results today. What I do is bracket around the value for the highlights I have established. I can place my highlights in Zone VIII (+3.0 exposure steps), and I bracket around this value. I may go from +2.0 to +4.0 for important shots. I am shooting digital, so the cost of this is minimal.

So in summary ETTR is a correct strategy but the method requires a certain flexibility. I bracket judiciously around a fairly high value. With my camera I can recover highlights way into the +3.7 level, but this may not be desirable in all circumstances.

[didger]

The issue of shooting for the right end of the histogram has arisen here countless times and I've come to understand the theoretical significance of it; namely that you're using more of the bits that you've paid for.  It's exactly the same thing as recording digital audio as close to max levels as possible to avoid a grainy sound due to effectively only using part of your potential bit width.  It's also the exact same risk of clipping.

HOWEVER, for the case of digital photography in real life, I've often wondered just how important this issue actually is.  The concurrent thread about ETTR shows that there's a very substantial down side and a big risk with ETTR, so it's important to know just how much real life gain there ever is shooting to the right.  I inquired on the forum and got strong opinions backed by no actual facts or test photos.  The top image here is one that I had to shoot very dark to avoid blowing out the sky on the left side.  I could have bracketed and blended, but the histogram showed that I was not off the left side and I should have no loss of shadow detail.  The white border shows approximately the crop on the bottom.  I did a huge levels adjustment on the full image to get the brightness and contrast about where I'll ultimately want it and then did the crop.  There's been no sharpening or any other tweaking at all at any stage.  Compared to what I would expect from a normally exposed foreground, or even a shot way to the right of the histogram foreground, I'm not seeing a problem with this crop and it's nearly black in the original raw conversion.  Obviously, if you're shooting raw with a 1ds at 100 ISO you have enough bit width even if you don't make a point of using it all, just as in audio you can get away with 50% recording levels with 24 bit recording, but you have to go for close as possible to 100% levels with 16 bit recording.

I'm not planning to do any more real gambling or shooting things over and over to crowd the histogram right to the right edge.  In the kind of shooting I mostly do, that will clip your blue channel almost every time and though this is generally just sky and is easy to deal with, I'm still not seeing any real (non theoretical) reason for pushing the histogram as far right as possible, at least with a 1ds.  A 1dsMKII is supposed to be even better at handling deep shadow areas without noise or loss of detail, though now that I've done this test, I can't see that I care much about this 1ds and 1ds2 difference.  The picture here is really extreme.  I  usually bracket in a situation like this, but the light was changing too fast and I wanted to get some more shots.

[61Dynamic]

At the time I'm writting this, your images are not showing...

I could have bracketed and blended, but the histogram showed that I was not off the left side and I should have no loss of shadow detail.

Ah, but you do have a loss in shadow detail. Assuming proper focus if you were to compare that frame with another bracketed with the sky blown out and thusly that shadow detail brighter, you most certainly will have more noticeable detail in the shadows when you correct in post.

As someone who shoots candid photography as part of my living I often am shooting in varying low-light situations. I often have to shoot in AV with exposure corrections in post and sometimes push exposures. I try to use ETTR as much as possable and when I do, I always get better shadow detail and also better mid-tones and highlights. It's less noisy, sharper and more detailed.

The question is, how important that shadow detail is and what your situation will allow. In your case, you had to under-expose to capture the image. A loss in shadow detail is acceptable in this case since one picture is better than no picture. Furthermore, you are pleased with the results. However, if the detail in the shadows is important for a particular image then you'd certainly benefit from ETTR.