TriX Loc: Raleigh, NC

Once you have the lens wide open and minimized the shutter speed, you can’t get more light to the sensor, But you CAN raise the ISO which either increases the gain of the amp between the sensor and the A/D (utilizing more of the DR of the A/D) or multiplying the output of the A/D by a constant. I get the arguement, but it ignores those two methods of increasing the effective ISO.

selmslie Loc: Fernandina Beach, FL, USA

They were already blown in the JPG you started from. You can’t un-blow them.

tomcat

True, don’t you reach a point where the ever increasing ISO values are increasing the noise as you keep trying to get ETTR ? I keep bumping up to 20,000 range quite often after I’ve maxed out aperture and shutter speed

BebuLamar

If you can't get more light (intensity and or duration) then you can't reduce noise with ETTR. Boosting the ISO increases noise and then you reduce it back by turning the exposure slider in post and you gain nothing in term of noise level but you will have to do the PP. And since ETTR only effective when using base ISO and modern cameras at base ISO are virtually noiseless so there is really no need for ETTR.

selmslie Loc: Fernandina Beach, FL, USA

You get the highest DR an S/N at base ISO because you’re forced to increase the exposure.

The improvement you get from ETTR is also because you increase the exposure, usually by lowering the shutter speed. But simply lowering the ISO also causes you to lower the shutter speed.

So you get the same benefit by lowering the ISO as you do from ETTR. But once you reach base ISO, your only remaining option is ETTR.

TriX Loc: Raleigh, NC

Here’s what’s missing from the arguement - let me provide an example (and remember that amplifiers do not necessarily amplify both the signal and noise equally - it depends on the type of noise and whether it originates in the sensor or the amplifier).

You encounter a subject where you are at base ISO, and even with the shutter speed as slow as possible without motion blur, and the lens wide open, you are say 3 stops underexposed. You record the image on a 14 bit A/D. There are two sources of noise - that from the sensor and that from the amplifier between the sensor and the A/D. The noise from the sensor is constant regardless of the amplifier gain, but the noise from the amplifier, AS A PERCENTAGE of the total amplifier gain, is not linear - it is a larger percentage of the signal at low gains, but smaller as the gain is increased. Since you are 3 stops underexposed with zero amplifier gain and noise (base ISO) you are using only 11 bits of the A/D or values of 0-2047 whereas the A/D actually has a range of 0-16,383. In PP, you multiply all the values by 8, which reproduces the entire 14 bit range, BUT you still have only have 11 bits of resolution. Every 8th point in the 14 bit array is a real value, but all the values in between are interpolated - you can’t increase resolution by multiplying by a constant. And, as you mention, you are multiplying the sensor noise by 8 as well - you have accomplished nothing in terms of DR or S/N, just brightened the image and increased the posibility of posterization or banding.

Now instead, you turn up the ISO (to 800) by increasing the amplifier gain on the sensor’s analog signal by a factor of 8x. Now you are using the full 14 bit resolution of the A/D - each data point of the 16,383 points is real data, not interpolated. You have amplified the sensor noise by 8x, but the noise contributed by the amplifier, since it is constant, is a smaller percentage of the total signal as the gain is increased.

Now you can call this ETTR or not - it’s a matter of semantics, BUT the advantage, both in terms of resolution, dynamic range and S/N are real. Multiplying a digital value after the A/D in post is not the same as increasing the amplifier gain prior to the A/D. And if you doubt this, try both and take a look at the histogram of the output two methods - you’ll see lots of missing points (discontinuities) in the image shot underexposed at base ISO and “fixed” in post. The same as always exposing at base ISO and bringing up in post with “ISO invariant” cameras - it’s never an advantage compared to exposing “correctly”. That is my counter argument to the idea that ETTR is only useful at base ISO. Correctly exposing and using the max dynamic range of the A/D is always preferable up to the max gain of the amplifier.

selmslie Loc: Fernandina Beach, FL, USA

It’s not the ISO that causes the noise. It’s the lack of signal - not enough exposure for the available light.

BebuLamar

You missed my point. Let assume that the amplifier contributes no noise of its own. If you can't give the sensor any more exposure then it's fine to increase the ISO until the image looks right (SOOC). Doing so is better than if you use a lower ISO and have to boost the exposure in post (multiplying the values after the A/D) which results in lost of quantization bit (1 bit per stop) and is not desirable. However if you continue to increase the ISO (ETTR) until your brightest part of the image is close to the clipping point but the image would look too bright. You will then reduce the exposure in post to make the image looks right then you do not gain anything.
Now if with the same scene but brighter so even with base ISO you can increase the exposure until the brightest part of the image is near clipping then reduce the exposure in post to make the image looks right then you do reduce the noise further.

selmslie Loc: Fernandina Beach, FL, USA

Of course, there are scenes that are too dark to capture at base ISO. But increasing the ISO does not amount to using ETTR. It’s just a matter of using an ISO appropriate to the scene’s light level.

Gene51 Loc: Yonkers, NY, now in LSD (LowerSlowerDelaware)

I am surprised at your response.

1. Low contrast scenes also benefit from ETTR, particularly if there is a lot of shadow in the scene - ETTR, will improve signal to noise and clean up the shadows considerably.
2. ETTR is perfect for wide dynamic range - by training yourself to look for the highlights and with your knowledge of your camera, using a spot reading of those highlights where you want to retain detail, and adding the correct amount of additional exposure (or placing that part of the scene in the correct zone, if using the zone system), is exactly how it works -

1. ETTR makes you aware of your camera's tolerance for overexposure, and with practice there is zero, well maybe 1% risk of an unrecoverable miscalculation
2. It depends on how you recover the highlights - Clarity, Highlight Recovery, White Point, Contrast, and to some degree, Dehaze - can all work to help recover highlights. The other approach is to render the image as best you can in raw conversion, then open the image in Photoshop, check out which channel(s) are blown, and do channel replacement to apply the unblown channel's values to the blown channel(s). It's a bit tedious, but it works.
3. Adjusting the exposure slider in Lightroom, DXO PhotoLab and Capture One is not equivalent to an exposur "shift" as you suggest. The best way for you to understand this is to open an image and do just that. The entire histogram does not "shift" - it is re-contoured.
4. ETTR works best at base ISO, but is perfectly valid at higher ISO. The point being that you are using as much exposure as possible before blowing the highlights. This works at ISO 100 just as well as it does at 16,000 - as long as you are familiar with your camera's limits. It's not any more complicated than that.

Wow, I generally agree with a lot of what you write - but you missed the boat on 100% of this post. Drop the theory, pick up the camera, and check it out. I have been shooting like this for 52 years - with negative film it was ETTL, with color reversal and digital, ETTR. Works 100% of the time. No guesswork.

TriX Loc: Raleigh, NC

Actually, if I read you correctly (and I will re-read again to make sure), I think we’re on the same page. I think I would summarize my comments as that you’re always best using the maximum DR of the A/D, regardless of what you do in post or what you call the process. Can we agree on that, or am I misunderstanding you?

E.L.. Shapiro Loc: Ottawa, Ontario Canada

As an old time and long time Zone System guy, back in the film era, I have no difficulty taking the time to perfect my exposure techniques to accommodate dynamic range issues. I great deal of my professional work is carried out in the studio or on location under well controlled conditions. Usually there is ample time to refer to the histogram or the indicator lights in the viewfinder and make certain that the highlight details are attended to. Many of my images will end up on adverting billboards, menu screens on fast food establishments or print advertisements. I photograph food and beverage layouts, a bit of fashion, lots of products and white bridal gowns gowns. In all theses cases, blown out or grayed down whites or jet black vacant shadows are unacceptable as is excessive noise. So...I will shoot to the right, minimize ISO settings, bracket exposures and process my RAW files to maxamize range and quality. Many years of exposing color transparency films involved similar challenges and issues so many good habits helped in my transition to digital techniques.

On the other side of the story is shooting under conditions where the is time limitations and somewhat unfavorable shooting conditions. Folks here have alluded to shooting sports events in low light, being forced to employ higher ISO settings, having to freeze motion with higher shutter speeds and acheive a modicum of depth of field. Of course there is no opportunity to apply precise settings so some compromises are reacquired. In the case of sports coverage the most important aspect of a successful image is viewer impact and certain aspects of picture quality may have to be sacrificed. Excessive grain or "noise" is never an issue with a tack on action image captured a its peak with great story telling components and dramatic expression on the athletes. If the middle tomes and highligh are sufficient, oftentimes the shadows will with take care of themselves or go black but not ot the detriment of he total image.

Another technique or approach I use is to excerpt certain aspects of a controlled technique or system and apply it to difficult or time limited situations. I have had assignments to cover sports events in venues lighted for television (easy and fun)- so I could utilize my ETTR method, pre-set things and shoot away. Some more modern arenas are not lighted for TV but the existing light is even and somewhat color correctable with a custom white balance and again some pre settings will work. Older venues for amateur sports, however, can be "lighting hell- low light, uneven mixed sources with sodium vapor lamps in various states of disrepair etc. and of course flash restrictions. All bets are off so you pump up the ISO, make do with an integrated exposure readings and an autonomic white balance and "praise the Lord and pass the ammunition".

As I have alluded to in othere conversation about dynamic range is my comparison to the film era. In the olden days we coud choose from a wide variety of films with different characteristic curves. God only know all the processing techniques and formulas that were designed to control and alter the dynamic range of theses materials. There was a plethora of developers, there were infinite formulas and variables, of course the Zone System, pre-soaking, intentional under and overexposure and accompanying processing accommodations. Even if thethe "ideal" negative was produced there were more options- different kinds of enlarger ligh sources, many contrast grades of printing papers, and again countless processing formulas and methods. There were also many color film choices as to speed, saturation and palette but altered processing had its limitations in that certain gremlins like color-crossover, de-saturation and other issues would set in. Problem is, in digital photography, even the finest top-of-the-line cameras don't offer interchangeable sensors or circuitry- we can't change the characteristics beyond certain adjustments. So...as uneconomical as it is, I have found that if a camera can not deliver the DR I need under most circumstances, I just have to go to a better camera- larger sensor and improved circuitry etc. There is just so much I can squeeze out of some of the older models- there is not enough time or windows of opportunity wide enough to maximize everything every time. One technique can not serve every job.

selmslie Loc: Fernandina Beach, FL, USA

A low contrast scene with a lot of shadows sounds like a contradiction. That sounds more like a scene with a wide DR and I call that high contrast.

I think that's exactly what I have been saying. On the other hand, if the DR is not wide, ETTR is less than perfect to the point of providing no benefit at all.

What has made me perfectly confident in knowing the camera's tolerance for overexposure has been the testing I did with the blinkies and RawDigger. ETTR had little to do with it.

Prevention is is better than curing. I know how to avoid blowing any channel by watching the blinkies.

That's not quite correct. Moving the exposure slider in Capture One, Capture NX2 or Picture Window Pro changes the gain just like changing the ISO setting on the camera. It's likely the same for other programs. But moving the Brightness, Highlight or Shadow sliders alter the tone curve disproportionately.

I have found that the blinkies are the simplest way to predict where the raw file is going to blow out. I have only tested from ISO 100 through 800 because I hardly ever go above that level.

Better yet, on my A7 II, I can see the blinkies (Zebra highlight warnings) before I trip the shutter. That beats the trial and error method I have to use with my Df and D610.

I'm not doing any more testing for my own benefit. It's just to illustrate the principles to others.

tomcat

Sorry, I knew that...I've just gone brain dead over this whole ETTR deal and I totally forget that this is accomplished at base ISO....So as we discussed earlier, I just don't have enough light to pull this off... But I have noticed that when I do bump up the ISO really high to around 52000, some of the girls looks like newborn speckled pups.

TriX Loc: Raleigh, NC

Q

Just to clarify the difference between adjusting the gain in-camera and in PP. I can’t speak for all cameras, but the ones I know (especially Canon) use a combination of amplifier gain in the amplifier between the sensor and A/D (HW gain) and multiplying the digital output of the A/D in SW by a constant, which is essentially what you do when you adjust an exposure slider in PP. in general, amplifier gain is used between base ISO and ISO 1000-1200, and SW multiplication above that. You can actually see the HW amplifier steps (the gain adjustment isn’t continuous) in the DR curve of many Canon bodies below ISO 1000-1200, and it’s smooth after that as you would expect. There is an excellent technical paper on the subject: http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p2.html

One interesting test is to expose a subject “correctly”, say at ISO 800, and then underexpose by some number of stops and then raise the exposure in post so it is of equal brightness, then look at the histograms from your PP ap, and the difference will be glaringly obvious. I’ve done exactly this and will post the link to the thread if anyone’s interested.