I would have used the live histogram to push the data in the highlights very close to the right, if not actually actually touching the right and dial it back a little in post if necessary.
The histogram of a jpeg will normally show clipping before the raw data actually clips so as long as the camera histogram does not show clipping then highlights will not be clipped and you can set them to whatever you like in post.
The main principle to maximising the quality of the raw data, whether for a set of HDR shots to be blended or for a single shot of a scene, is to get as much light onto the sensor within dof and blur constraints without clipping important highlights.
As I posted earlier. If ETTR'ing the shadows in a HDR scene it doesn't matter if the highlights in the sky are completely blown because the sky in the final image will come from a different shot that is best exposed for the sky.
I'm no longer a subscriber to this unwritten rule that every raw capture should be maximising the capability of the kit, or this assumption that getting the maximum quality data as defined by the top-trump data that we use to define the "best camera" actually translates into a better photo. I have not found that to be the case. Quite the opposite really, I find that exposing to retain the highlights and shadows at the extremes of the image very quickly becomes the mantra for processing for the simple reason that we fail to question it. And so that highlight and shadow detail we strove for also must be revealed as it was a major reason of the effort in our exposure.
The image above that I posted in the C&C thread is a good visual to what I'm trying to say. When I was composing the image the sun was not out, that only happened briefly a little later, I was looking at making sure that I had the complete branches on the left in the photo together with the foreground leaves on the right. But the trouble with that is you're looking at the edges of the frame to define the composition rather than the centre and it ended as it often does with an unbalanced photo and the "damn I wish I'd framed a little further to the right," thump of fist on the editing desk. Reader I've learnt, and so also look at the centre for the composition ignoring the edges, they fall where they fall. (That LCD screen on digital is a major step forward in the learning process, along with the live histogram on M/L). Point is that if you define your composition by the edges it impacts the centre, and that is place your viewer is really looking at.
Also worth noting that it's mostly slightly out of focus, a little soft, f7.1 and focussed on the foreground leaves. I could've got out my tables and picked the ideal focal point for maximum DOF, maximised the sharpness of the lens, and again extracting maximum performance from the kit is dictating the look of the final picture. I like the softness of the distance, it works really well as a contrast to the detail on the foreground leaves, and in softening the acutance the distance is less busy/cluttered and I think the soft colours really benefit.
It's the same with the highlights and shadow, we look to the extreme edges to maximise the performance of the camera. That has an impact on the real subject of the photo, you shift everything towards the middle for the sake of the extremes that people aren't really looking at. You end up squeezing it all in and making sure that it shows in the end result simply because you don't question the effort that went into obtaining it, it must be worthwhile or you wouldn't strive to do it. (Then you can spend days on photo forums defending your actions as valid without really looking past the histogram 😀 ).
I tend to work the other way, I use the E/C button with the live histogram and chimp. I wanted the effect of light through the leaves, highlights are blown in the RAW but so what? There's enough blue to carry it. I also managed to keep the impression of light without much PP, and PP always subtracts. I also expanded the subject to fill the DR in post rather than compressing it to accommodate inconsequential detail. I simply don't think about it, close is good enough, but ISO800 and lots of shadows, hey I see no noise...
Take another step back and see, what's the worst that can happen? The danger of the technical approach is that it confines you to a box defined more by camera capability than visual understanding that often you fail to notice. You can end up chasing a small SNR that's invisible in rthe fional image and yet the processing you've used to obtain it has a large impact nort only on the final image but also in part defines your whole approach.
P.s. I know about maximizing the SNR for DOF and blur requiorements while maintaining important highlight details, I really don't want it to become an earworm!! 😫
No-one is saying that it is a rule or something that must be achieved on every shot.
More often than not, near enough is good enough (especially when pointing and shooting) but my general goal for every shot is to aim to get as much light onto the sensor within my dof and blur constraints without clipping important highlights.
This is not rocket science or time consuming. I find it quite simple with the following general setup.
When pointing and shooting I normally use aperture priority, set the widest aperture that gives my dof, set a minimum shutter speed to be what I can safely hand hold at to eliminate motion blur and/or camera shake and auto ISO.
This way I don't have to think about aperture, shutter speed or ISO until something changes that requires me to adjust aperture or shutter speed.
When I press the shutter button after first composing the frame, the camera first meters the scene with base ISO. If the shutter speed the meter wants to center/zero the meter needle is slower than the minimum shutter speed I set then the camera sets the minimum shutter speed and raises ISO to center the meter needle.
If the shutter speed the meter wants to center/zero the meter needle is faster than the minimum shutter speed then base ISO is maintained and the meter's chosen shutter speed is set.
This setup also allows the maximum amount of light onto the sensor within my dof and blur constraints without clipping highlights without me having to think about aperture, shutter speed or ISO.
It doesn't sound like you understand at all because you don't maximise the SNR for DOF or blur requirements.
The larger the exposure*, the higher will be the SNR. The higher the SNR the less visible will be the noise.
Because it is a low SNR that makes noise visible, you maximise the SNR within dof and blur requirements without clipping important highlights in order to minimise visible noise.
exposure* - amount of light striking the sensor per unit area during a shutter actuation.
Actually Dan I think it's you who doesn't understand here. I still think in film exposure, I tried the digital mantra and it didn't work for me.
I don't worry too much about DOF and because I don't think that maximum sharpness results in a good photo, things fall where they fall and are often improved by it. Also my exposure is not constrained by it.
I don't worry about the highlights, they fall where the fall and nearly always result in increasing the impression of light and reducing PP, which also kills it. And so my exposure is not constrained by it.
So, not being constrained by exposing for the extremes that nobody really notices my exposures are very probably 1 or 2 stops higher than your's already.
I don't need to raise shadows to the level of daylight.
In fact I don't find any problem with noise in digital photography unless shooting in ridiculously low light or trying to retain rediculous highlight detail, paradoxically chasing maximum IQ is creating more noise...
I am sure you are right, but my process of planning and taking a photo is entirely non-intellectual.Yes, I try to get things in the right part of the frame, and in focus; but SNR is the last thing I consider.
Recently, I have found that I can use silly high ISO and hand hold for architectural detail shots of frescoes and such. Up to about 3200, noise is not a problem.
Aftrr setting the maximum exposure as described earlier, the iso setting the auto iso sets is totally irrelevant as long as important highlights are not clipped.
SNR is probably the last thing people consider, at least directly, when setting up a shot but while choosing my aperture and shutter speed I do think about getting as much light onto the sensor as possible within my dof and blur requirements without clipping important highlights.
Of course in good light where the SNR will be naturally very high it is not as big a concern as it is in low to very low light where maximising the exposure as described earlier will minimise visible noise and your noise reduction app will thank you for it.
I agree with both of you and I think that gets ETTR in about the right and current perspective. I'd add one more but it's marginal. What I intend doing with the image is a factor. If I think the shot might be something I want to do as a large print, and I'm talking about A1 size, I'd be more inclined to use ETTR as Dan has described.
I was replying to @ArvoJ, who wrote: "Sure usual exposure stack gives almost similar results." which I assumed to mean a stack of equal exposures, from context. My reply gave one good reason why the three exposures approach avoids one potential artifact of the equal-exposures stack method, which is fixed pattern noise, which can make it necessary to subtract black frames with some sensors. Some sensors work with some methods better than others. One day, several months ago, I was looking at how noisy a web camera was in indoor low evening light, and I decided to write code that kept taking webcam frames and added them together in a buffer with floating point precision, and the screen showed the summed image until you hit the spacebar, and it started over again. I wrote the code and ran it, and lo and behold, the noise fell away, but also, as it fell away, all the fixed pattern noise (which looked like a plastic fake straw basket) started to get clearer. So, I was reminded of the potential of fixed pattern noise and what needs to be done about it if it is strong enough to be visible after much stacking.
Your 3 exposure method covering 4 stops, of course, gives great "bang for the buck", because it avoids visible fixed pattern noise, and gets the target minimum SNR to happen very far below the highlight clipping point, with just 3 images, and does not require a sensor with great single-exposure DR, with a small number of frames.
The point of optimizing exposure for a scene via multiple frames can be not just to get a certain amount of DR, per se, but to reduce SNR at all levels, with a priority on the darkest areas of interest. When a camera is said to have high DR at base ISO, that tells us nothing about about SNR of middle grey or white. Quantum efficiency hasn't made any major headway in years, so there is a long-standing plateau of SNR for any given "percent grey" tone at base ISO, despite a wide range of dynamic ranges in sensors made since that plateau was reached. So, ETTR and/or multiple stacked exposures is still useful for reducing SNR, even with "high DR" cameras.
The latest Olympus m43 sensor has basically the same PDR at Photons2Photos at its ISO 320 as the FF Canon RP has at ISO 100! They may have similar measured SNR somewhere around 9 stops below raw saturation at those ISOs, but the RP has much higher SNR in the brighter tones that are very low in relative levels of read noise. DR is a poor general "handle" for full SNR curves as it only provides one point from which to extrapolate a curve. It only tells you about noise near the bottom, as measured from the top, and nothing about the bulk of the tonal range within the DR.
Quite so. The ISO Standard doesn't get even that near to the bottom ... it measures SNR at 1% of saturated exposure and then extrapolates downward to guess the level at which SNR=1.
Yes, noise that appears randomly reduces relative to signal, but noise that is fixed the same in every frame gets clearer as the otherwise-masking random noise disappears. You may need black frame subtraction to avoid fixed noise getting clearer.
I don't think the concept of "noiseless pixels" is useful to understanding noise. Some values are just closer to the mean, but its not like they avoided chance; chance just landed them on the mean, and the final values are often the sum of multiple chance sub-noises.
Roll one die many times, and you get a fairly even distribution of 1 to 6. Roll 2 die, and you get a sum of 2 to 12, but no longer equally distributed, with a sum of 7 happening most, and sums of 2 and 12 happening least, but every sum of 7 is 1+6, 2+5, or 3+4, with no 7 involved, except in the sum. Are the rolls that summed to 7 chanceless (noiseless)?
No need to apologize - it was incorrect of him to assume that "exposure stack" meant all at equal exposure settings, just as a focus stack does not normally mean all at the same focus settings.
