ETTR, when and why

[pop]

Advertisement (gone after registration)

vor 11 Minuten schrieb pgk:

Boosting ISO and ETTR are opposing constraints.

The article above does not recommend boosting. It merely describes a technique which is applicable to any ISO setting.

ETTR is not overexposing. It's counteracting the underexposure routinely applied by the built-in exposure meters. It's not overexposed if no relevant part of the image is blown.

[pgk]

10 minutes ago, pop said:

The article above does not recommend boosting. It merely describes a technique which is applicable to any ISO setting.

ETTR is not overexposing. It's counteracting the underexposure routinely applied by the built-in exposure meters. It's not overexposed if no relevant part of the image is blown.

But the technique is contradictory and exposing for the highlights (an old technique from film days) is only relevant if you can viably control mid-tones and shadow detail.

To me ETTR would be valid only if it had significant advantages. Having tried it I have found that it does have disadvantages and I have yet to see any noticeable advantages. Its a theoretical solution to a marginal problem which has, at best, minimal yield. Despite this it is still touted as a useful technique and I really don't know why because I have yet to see an image which is noticeably better by using ETTR as opposed to a 'considered' exposure (which may not be about placing a mid-tone but still may not drift towards ETTR).

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[pop]

vor 4 Minuten schrieb pgk:

But the technique is contradictory and exposing for the highlights (an old technique from film days) is only relevant if you can viably control mid-tones and shadow detail.

To me ETTR would be valid only if it had significant advantages. Having tried it I have found that it does have disadvantages and I have yet to see any noticeable advantages. Its a theoretical solution to a marginal problem which has, at best, minimal yield. Despite this it is still touted as a useful technique and I really don't know why because I have yet to see an image which is noticeably better by using ETTR as opposed to a 'considered' exposure (which may not be about placing a mid-tone but still may not drift towards ETTR).

For every practical purpose, the digital sensor can be considered a linear device with an upper and a lower cutting off point. It becomes, by definition, non-linear at both cutting points. It also adds noise, presumably at a constant level.

If the medium is linear and controlling mid-tones and shadows becomes difficult when applying ETTR, then you are obviously accustomed to ETTL, or your software can not handle frames that are exposed to the right. If so, it's broken and needs to be fixed.

The topic is well understood in the domain auf audio recording. You set the recording level as high as you possibly can, but not so high that the loudest part of your record becomes distorted by the nonlinear range of your medium (or your devices). Turn the recording level higher and you will get nasty distortions. Turn the recording level lower and the noise produced by your medium or your set will become too noticeable.

The procedure outlined by ETTR consist of two "forces".

One: choose you exposure such that highlights are blown in none of the channels. For me, this is clearly the more important part. Clipping one or two of the color channels will produce very unpleasant color changes, clipping all of them will produce featureless white areas. Clearly, avoiding these artefacts is of prime importance for many applications, and I have yet to find a camera which reliably avoids them.

Two: Turn up your exposure as much as you can, as long as rule number one is adhered to. This has the obvious advantage that it improves the signal/noise ratio which, in turn, improves shadow detail.

Obviously, if there is not enough light to raise the exposure to the desired level, it all becomes moot. At times, changing the sensitivity of the sensor might help, and we're back to rule one and two above.

You mentioned before that you preferred exposing your frames "correctly". By what standard or measure do you decide which is the "correct" exposure?

[pgk]

6 minutes ago, pop said:

For every practical purpose.....

You mentioned before that you preferred exposing your frames "correctly". By what standard or measure do you decide which is the "correct" exposure?

Firstly the linearity ends before the cut-off points - hence the colour artefacts in overexposed highlights which can be frustratingly obvious. The point at which channels are 'blown' is usually after colour artefacts have crept in.

Correct exposure is quite simply the one which yields the results that you are after. Its not fixed but it doesn't have to pander to any mantra of mid-tone accuracy, ETTR or whatever. So if I need to use a specific exposure, assuming that its not too over/underexposed then I will do so. The image is more important than using a specific technique to ensure some theoretical 'quality' advantage. My point with ETTR is that whilst it might theoretically provide a technically better RAW image file, in practice I have yet to see an image which was significantly better because the photographer specifically used ETTR to take it. But ETTR is not the same as exposing for highlights which is valid when its valid and not a mantra. 

[pop]

vor 18 Minuten schrieb pgk:

Firstly the linearity ends before the cut-off points - hence the colour artefacts in overexposed highlights which can be frustratingly obvious.

Unfortunately, a sensor with a 3color CFA has three cut-off points, one for each color. What you describe is when one or two but not three of the color components have been clipped. It's avoided by setting the exposure to a value which avoids clipping in any color channel. Hence, you expose for the red, green and blue highlights, particularly when following ETTR.

vor 18 Minuten schrieb pgk:

Correct exposure is quite simply the one which yields the results that you are after.

So, if one applies ETTR and happens to produce the result he was after, the exposure is correct?

vor 18 Minuten schrieb pgk:

But ETTR is not the same as exposing for highlights which is valid when its valid and not a mantra. 

How is ETTR not the same as exposing for highlights? You call one a mantra but not the other. Is it just the term which gets your goad? I can quite understand that we oldsters call it "exposing for the highlights", having grown up without any exposure histograms, while the youngsters who are more familiar with LCDs and histograms than dials with hands rather refer to the bright side of the histogram.

[SrMi]

5 hours ago, pgk said:

Starts with a fallacy. You could not use a higher than base ISO if your aim is to utilise the maximum dynamic range. If ETTR works for you that is fine, but trying to convince anyone that its a solution to utilising maximum dynamic range is far too simplistic.

I am not following. As I see it, the point of this article is:
If you want to reach maximum DR measured at any ISO, you should saturate the sensor (without overexposing any channel). If you underexpose, with the plan to fix it in the post, you should be aware that you lose DR.

Exposing optimally is less expensive than buying a new camera to obtain better DR.

 

[Chaemono]

Advertisement (gone after registration)

With modern sensors, pushed shadows noise/banding is not an issue at base ISO, so there one is better off erring to the left. It only becomes an issue as one bumps up the ISO. Proper exposure as far to the right as possible then becomes more important. However, it’s a very narrow exposure range even with MF sensors before a channel is blown as shown in #19.

Dual, triple, and quadruple native ISO sensors would be nice. Then one can always err to the left, IMO.

[pgk]

54 minutes ago, SrMi said:

If you want to reach maximum DR measured at any ISO, you should saturate the sensor (without overexposing any channel). If you underexpose, with the plan to fix it in the post, you should be aware that you lose DR.

But by increasing the ISO you are reducing the DR that can be used. By then compensating by ETTR you are overexposing for the ISO - in effect using a lower ISO (except that there is inow camera processing taking place). IF you really want to use ETTR and think that it will help increase 'quality' then surely it is only a 'valid' technique at base ISO? Otherwise its a game of chasing your tail.

[pgk]

2 hours ago, pop said:

How is ETTR not the same as exposing for highlights?

Depends on the highlights you are exposing for. See your point about the rgb channels. This was not a concern on film and isn't a concern now - unless you seriously don't want to clip any channel. ETTR is a mantra because its highly formulaic and requires a specific technique to be applied with little beneficial result. We won't agree but I see ETTR as a solution to a problem which is marginal at best. In the real world I think that it has little benefit and as I have already said, many drawbacks (precise exposure, usable only at base ISO or of dubious relevance, increased post processing, and in my experience, tonal discrepancies which I suspect are often due to existent in-camera processing).

[SrMi]

15 minutes ago, pgk said:

But by increasing the ISO you are reducing the DR that can be used. By then compensating by ETTR you are overexposing for the ISO - in effect using a lower ISO (except that there is inow camera processing taking place). IF you really want to use ETTR and think that it will help increase 'quality' then surely it is only a 'valid' technique at base ISO? Otherwise its a game of chasing your tail.

Why are you talking about increasing ISO? Yes, the best results are achieved at base ISO and saturated sensor. Nobody in this thread suggests otherwise. There must be a misunderstanding somewhere.

 

[pgk]

7 minutes ago, SrMi said:

Why are you talking about increasing ISO? 

Post 57 - yours.

[SrMi]

16 minutes ago, pgk said:

Post 57 - yours.

You either misunderstood the article or my excerpt. 

[Chaemono]

Am 25.9.2020 um 20:17 schrieb adan:

...

So ETTR was a technique for "the rest of us" to get the widest tonal range out of jpgs. It still works with raw - it just is not as important now.

+1. I thought it mattered for RAW until I took some test shots. 😂