Nightski

Maybe I had just arrived in this spot from shooting eagles and hadn't changed my settings yet. I usually do try to get my shutter speed up once I'm in the small bird spot. I often shoot at F6.3 and use lower ISO's when possible. However, this image proves that if you ETTR properly, you can avoid noise at higher ISO's. I have one that's at ISO 6400 in this spot and it's incredibly unaffected by noise. Call it a little field experimentation if you like, Winter.

winterrose Loc: Kyneton, Victoria, Australia

Sorry, but I think you are missing the point that I am trying to make. Given that the f stop is set, one would select a high ISO in order to provide a fast shutter speed but then in an effort to reduce the noise resultant from the high ISO one chooses to deliberately overexpose the shot, (which is what ETTR is after all), and so accept a slower shutter speed. So why not just use a lower ISO, which would reduce the visibility of noise and maintain the high shutter speed?

selmslie Loc: Fernandina Beach, FL, USA

I think you missed Rob's point.

This image was captured in broad daylight (note distinct shadows and reflection of the sun in the bird's eye). The Sunny 16 Rule suggests an exposure of 1/ISO @ f/16. In this case, ISO 250 would work out to 1/250 @ f/16 which is the same as 1/500 @ f/11.

Changing the ISO without changing the shutter speed will not change the S/N and it will not change the level of visible noise. It will only increase the risk of blowing the highlights.

Neither will changing only the ISO affect the ratio of light areas to dark shadows. That comes with the scene. You just have to use your raw editor to recover whatever shadows you want.

Nightski

So ... Selmslie and Winter ... What are the technical issues that you detect in this photograph? Isn't it true that there are many combinations for a correct exposure? Selmsie... while I agree with you that there was a more optimal exposure, I disagree with you about recovering shadows. Yes you can recover them, but if you used a high ISO it will look like crap. The fact is that noise resides in shadow. I do use at least 800-1600 when shooting birds so that I can get my shutter speed up in case they fly.

selmslie Loc: Fernandina Beach, FL, USA

There are no glaring technical issues. The focus and DOF are sufficient. The reflection of the sun in the bird's eye is at 255,255,255 (as is should be) and brightest feathers are close to being blown. The only indication that they might actually have been blown is that there is little detail or texture in some of the white areas but not enough to detract from the image. The shadow under the bird is lit with only skylight so it received only two stops less illumination. This is not a problem since it has a mostly medium reflectance. Note that the underside of the branch is actually brighter than the shadow under the bird so there was probably something (snow?) reflecting light to fill in the shadows.

The only problem with the composition is that the birds is dead-center. There is nothing distracting about the branch behind the bird or the tree trunk on the right. They are so out of focus that they don't matter. Cropping just a little off the right (but not all the way to where the top of the branch emerges from the bottom) will move the bird off center.

R.G. Loc: Scotland

Some interesting points on using high ISOs here.

http://petapixel.com/2017/03/22/find-best-iso-astrophotography-dynamic-range-noise/

In particular:-

"So a reduction of light by the shutter or the aperture is the reason that the image appears noisier. It’s not noisier because of the higher ISO. This reduction of light is a reduction of signal and a reduction of signal yields an overall lower signal-to-noise ratio and thus a noisier photo".

If ETTR is achieved by increasing the amount of captured light (by using a slower shutter speed and/or a wider aperture), it increases the signal-to-noise ratio. Increasing the ISO helps to counteract the effect of downstream electronic noise (noise contributed by the camera's electronics).

However, increasing ISO also reduces the camera's dynamic range capabilities, so if the scene has extremes of brightness it would be advisable to avoid very high ISO values.

selmslie Loc: Fernandina Beach, FL, USA

Exactly correct. ETTR does not reduce noise. It's the additional exposure that increases the S/N ratio thereby reducing the noise.

ETTR merely moves the histogram to the right. That can be risky since you can't always be sure if the histogram is correct until you copy the raw file to your computer. Even then, if you have a lot of specular highlights, the histogram is not a good indication.

Here are some images from a recent outdoor car show. All of the color images were taken at ISO 400 with a Sony A7 II and a 35 mm Zeiss f/2.8 Biogon set to f/11. I used aperture priority and "Multi" metering mode (similar to Nikon's Matrix metering). I expected all of the images to be exposed around 1/800 second (because I used f/11 in daylight) so I did not need ICIS (in camera image stabilization). I got no noise, no blown highlights other than specular reflections and the shadow information was easily recovered when processing the raw file.

The B&W images were taken with the same lens on a Leica M6 using TMax 400 developed in Xtol. The Leica uses some form of weighted average metering off the front of the shutter. When in doubt I just set the shutter speed to 1/1000.

I hand-held both cameras.

R.G. Loc: Scotland

Strictly speaking, the randomness, interference and inconsistency that produces noise is there all the time. What changes is how visible that noise is.

Brighter = stonger signal and darker = weaker signal, which is why in the case of a bright exposure (or more specifically the bright parts of an exposure) the signal predominates and the noise becomes unnoticeable, whereas with a dark exposure (or the dark parts of an exposure) the noise becomes more noticeable to the point where it can predominate (depending on the level of darkness).

The sensor is the source of the signal, so what we're looking for is as strong a sensor signal as we can get (to give the best SNR), with the ever-present requirement that we don't blow any important, detail-carrying highlights (specular highlights are often regarded as expendable). The way we increase the sensor signal is to increase the amount of light being captured during the exposure, and the two factors that control the amount of captured light are the shutter speed and the aperture. If exposing to the right (ETTR) is achieved by using the shutter speed and aperture, the sensor signal will increase which in turn increases the signal-to-noise ratio (the SNR). However, ISO amplification is another matter...

ISO amplification takes place one step away from the signal source and it doesn't add significantly to the noise. What it does do is boost the signal before the remaining electronics adds its own noise (the downstream noise), so as far as the downstream noise goes, increasing ISO increases the SNR. However, the ISO amplification amplifies the sensor noise along with the sensor signal, so the ISO amplification does nothing to improve the SNR of the sensor signal.

The main implication of the above facts is that there is an incentive to maximise the exposure using shutter speed and aperture. ETTR (and exposing beyond the right - EBTR) are both ways to do that, but the possibility of blown highlights means that ETTR and EBTR have to be used with caution. If manufacturers can come up with a safe way to ETTR or EBTR they will be providing us with the most effective way to minimise noise.

Another implication of these facts is that they provide an incentive to know your camera's Extra Raw-Available Dynamic Range (ERADR). It would be nice if cameras came with that information, but it's possible to evaluate it yourself. ERADR is a very inconsistent parameter, even between different examples of the same model, so cameras would have to be assessed individually.

selmslie Loc: Fernandina Beach, FL, USA

The bottom line is that if you can't see it, it does not exist. If you can ignore it, it does not matter. And if you can remove it without damaging the rest of the image it's not a problem.

With a modern camera in daylight with a low ISO, noise will not be visible. In fact an incident reading or matrix metering may be all you need.

If this image had been captured with the same exposure (1/500@f/11) at ISO 250 the final image would have looked almost the same. The only difference would be that the brightest feathers (like the back of the head, the middle of the back) would have retained their detail and texture. Because of the high ISO they are actually blown out, but you have to look closely to see this. It does not really detract from the overall image unless you obsess over it.

winterrose Loc: Kyneton, Victoria, Australia

I have no specific criticism of your pic nightski, it is quite a well executed study.

I was compelled to bring this up because I sensed that there was a degree of misunderstanding regarding the relationship between ISO and noise. The information and explanations that selmslie and RG provided were excellent and there is no doubt that for many their work would benefit from a better understanding of the subject.

selmslie Loc: Fernandina Beach, FL, USA

It is probably more useful to know what your camera considers to be middle gray. You can test this easily with the help of RawDigger. You can download a free trial version but the price is so low you might just want to purchase it.

The procedure is simple:

1. Take an image of a uniformly gray or white surface. You can use a long lens and fill the frame - might as well turn off autofocus and focus at infinity. Use auto-exposure with the exposure compensation (EC) to set to 0.
2. Take a second exposure with the EC set to +3
3. Open the raw file with RawDigger and display the histograms.

You will see a narrow histogram or spike centered in the display. For my two cameras the first image's histograms are half way between a 14-bit raw value of 1000 and 2000. In the second image the spike is centered between 8000 and 16000.

You can learn a lot more about you camera by spending some time with RawDigger than you will ever learn by trying to nail down your ERADR.

R.G. Loc: Scotland

I agree with most of what you're saying. However, I don't agree with the above comment because it's not true of the noise that appears in shadows when you attempt to lift them. Shadows that appear to be noise-free SOOC can prove to be difficult when you try to recover them.

Just because an exposure appears to be noise-free SOOC, that doesn't mean that there's nothing to worry about when it comes to editing. As a very general rule I would say that the more pushing and pulling an image can take in PP the better. That's one of the reasons why I go to the bother of using exposure bracketing (when possible), merging the resulting files and then pushing and pulling them into shape in PP. Coming back to the original topic, the bottom line is that a single exposure got from ETTR will be more noise-free than one got from a neutral exposure.

Your comment above is true of SOOC images, and it's worth reminding ourselves of the relevance aspect of performance figures, but PP is a very real aspect of most people's workflow, and that's what I would base my relevance evaluations on. Because of that I keep coming back to the fact that ETTR and EBTR, when done properly using aperture and shutter speed, result in noise being far less of a problem.

selmslie Loc: Fernandina Beach, FL, USA

You might be less skeptical about my comments after you look at this thread, Don't Give Up on Dark Images.

Those two images were taken in open shade at 1/2000 @ f/11 ISO 200. A quick calculation of the exposure would make that about 5 stops underexposed. I was able to pull the shadows up in Capture One without any noise becoming visible. The credit probably goes to the camera, D610.

R.G. Loc: Scotland

That's very impressive. Hopefully if things carry on improving we'll get to the point where noise and dynamic range aren't a problem any more.

Having said that, the images do look a bit soft. I wouldn't concede to being completely impressed unless I could compare normal exposures with your recovered images. And for those of us using less impressive hardware, ETTR and EBTR will still be the low noise options.

selmslie Loc: Fernandina Beach, FL, USA

Noise and DR are not a problem for me with either of my digital cameras, D610 and A7 II. I keep my ISO low and I often use a tripod. My usual subject matter is landscape and other relatively stationary subjects.

The softness is from the Nikon 75-300 f/4.5-5.6 at 300 mm. It's my only long lens for this format and I hardly ever use it. For $133 it's just OK.