The following is at best half-baked. I need to experiment with it to see how good it is. I thought I’d share it with you anyway, since I often get good advice from you all.
When figuring out an exposure strategy based on ETTR and the effect of in-camera ISO settings on signal-to-noise ratio (SNR), one of the first things to do is to identify two points on the curves that I derived in the preceding posts. I don’t have a catchy name for these curves – let’s call them “ISO Improvement” curves for now — but what they’re showing is the relative improvement in SNR that’s available for the specified tonality by changing the camera’s ISO setting vs making an equivalent change with the Exposure slider in Lightroom or Adobe Camera Raw.Here’s an example:
In my tests, I picked a light and a dark gray similar to Zone VI and Zone III of a normally-exposed scene using the Zone System.
DxO posts curves showing SNR vs ISO setting for Zone V (which they call 18%), but the resolution of their curves is too low to figure out the relative benefits of increasing the in-camera ISO setting vs boosting Exposure in LR of ACR. DxO does not subtract out the Poisson Noise component of the SNR, as I do in my ISO Improvement curves, so, even with increased resolution, you’d be reduced to eyeballing slopes in the DxO curves. As far as I know, the only way to get the information that the ISO Improvement curves provide is to make them yourself. I will be posting the Excel spreadsheet I use to generate these curves together with instructions on how to make the exposures.
I’ve had a reader point out that the fact that the five cameras that I’ve tested offer little benefit from boosting in-camera ISO doesn’t mean that that situation obtains for all cameras. The reader says that Canon cameras usually can benefit from in-camera ISO settings well above the base ISO. I have no way of verifying that since I don’t have access to any Canon cameras, but also no reason to think that it’s untrue.
The way that I suggest using the ISO Improvement curves is as follows. Look at both the Zone VI and the Zone III curves, and for each find the point nearest the base ISO where the curve gets close to its maximum. Call that point the Peak ISO. Then find the point after which there is a substantial drop in the curve. Call that point the Max ISO. Then take the smaller of the two numbers representing the performance in each tonality (this is a judgement call on my part, and I could easily be convinced that the average (or the geometric mean, or something else) is the right number; the point is to reduce the number of ISO values that you need to think about when making an exposure). Here’s the way I see the curves I’ve created for the five cameras I’ve tested:
So now, for each camera, you only have two ISO numbers to think about, the Peak ISO and the Max ISO. Here they are for the five cameras I tested:
With that in mind, here’s an exposure algorithm calculated to give you near-optimum SNRs:
- Use ETTR at base ISO if you can manage it. Everything that follows is about how to best manage the loss in image quality you’ll get if you can’t do this.
- When circumstances force you to make exposures under the ETTR/base ISO exposure, start turning up the ISO. When you get to the Peak ISO, stop. If the Peak ISO is the same as the base ISO, skip this step.
- When you’ve reached the peak ISO, stop changing it. Let the histogram slide to the left until it starts to be hard to see the image in the electronic finder or the chimping screen, then start increasing the ISO. You have a lot of latitude here; there are only small penalties in increasing ISO early (loss of headroom) or late (harder to judge the shadows in the histogram, or see the undeveloped images in Lightroom).
- When you get to the Max ISO, stop increasing the ISO and let the histogram continue to slide to the left.
John Scherer says
One are of photography I would like to experiment with is the milky way night scape shots you see all over now. How do these methods work in these VERY low light situations. Is your exposure algorithm still the way to go?
A friend did a series of deep sky tests with the Nikon D800E, and found that there were very slight improvements from increasing the ISO setting up to 1250. After that, you were better off pushing in post. That’s about two stops over the unity gain ISO. So, the answer to your question is, “Pretty much, but stop a little sooner.”
I have difficulty understanding the last two steps, I’d appreciate it if you could shed some light on it. Let’s say I have a D800E paired with the Zeiss 135/2 and I am 2 stops short of achieving ETTR at IS0=100, f2 &1/160. I think you suggest not to increase ISO to 400 to ETTR but lower the shutter speed first and increase the ISO afterwards. Can you expatiate on the method a little more.
In that situation, I’d just take the picture, let it be two stops under, and fix it in post. If it were more than two stops under, I’d consider increasing the ISO setting.