I agree that analysis tools help both photographers and developers.
I will try converting to Adobe DNG meanwhile.
It was D50, 3 of CC24 averaged (last batch), 2 SG and 2 DC. Device used was Gretag Spectrolino.
Okay, good. That will produce the best results. What illuminant was used - D65? D50?
Interesting you mentioned 'better rawnalyse'. Our feeling is that raw analysis tools are at least as important as raw converters. They help both photographers and developers. You can expect various tools there, not just a raw histogram.
Have you checked Rawnalyze with D40 files converted to DNG?
These were produced from spectral data.
Were these produced from spectral data, from measured or published LAB values, or from RGB data in a known colorspace and whitepoint? What is the illuminant used?
Thanks for the explanation. If people are somehow reading "The simple lesson to be learned from this is to bias your exposures so that the histogram is snugged up to the right" to mean "bias your exposure so the histogram is snugged u to the left" then they are (sorry to be blunt) clueless, espacially given the two histograms labelled
"Histogram to the Right
For Maximim S/N Ratio"
which visually and clearly indicate exposing to the right.
On the other hand, his glib "not to the point that the highlights are blown" is inaccurate. Firstly, some hilights are always blown in a real world scene; its a question of how badly. Secondly, ETTR trades of more blown hilights for less shadow noise. His article should make that more clear.
You may want to look at many underexposed images and ask those who took them why images are so underexposed. Too often you will be referenced to that unfortunate phrase.
The problem is with how people understand what is suggested. Part of it is that using just the LCD on the back of the camera it is very difficult to recognise those highlights that should be left clipped. Part of it is that camera settings affect the point where the highlights start to blink. If the Reichmann's article would go deeper into the practice and warn inexperienced photographers of the possible misreadings and problems many would avoid a lot of disappointing results.
I do not think I need to quote Reichmann article again to show how definite he is when it comes to protecting highlights. That phrasing made a context of itself.
You can find a more accurate formula, like "Expose to place as much data within this linear-encoded RAW image without losing highlight values you wish to reproduce", suggested by Andrew Rodney. He continues, "ETTR presents a few problems, one being that the LCD camera preview, including the histogram and clipping indicators, isn t based on the linear RAW data. Instead, this preview is based on the rendered gamma-corrected JPEG your camera is set to produce, even if you don t save that JPEG and only shoot a RAW file! If your goal is to produce the best possible exposure for RAW, using the ETTR technique, the feedback on the LCD could steer you in the wrong direction."
Still, the major practical problem to many is to recognise the highlights that they want to reproduce from the highlights that can't be reproduced without unacceptable underexposure - not because those are specular, but just because the total range of brightnesses found in the scene is too high. Worse, some even do not know the problem exists at all.
Zone method is the easiest way to recognise the range of brightness and to identify the region that will allow the proper metering for the scene.
Referring to bits of smoke and mirrors, I can't see why you are trying to say the factual mistake Reichmann made is irrelevant. That mistake is a very grave and consequential one, still influencing too many to be considered minor.
Simply quoting that one sentence from the Reichmann article is out of context with the rest of the article and the concept of ETTR.
Read the entire Reichmann article in context and the concept involes pushing things to the right, not to the left; and this should be done for scenes and camera settings where it makes sense to do so.
Why not post the original image used to produce the data for this article and then everyone will be in a better position to make a judgement on the value of the information it conveys. At this stage, I have serious doubts because the article above seems to be aimed at proving the incorrectness of a single, poorly written statement within the context of a different concept.
To me it is easy to blindly protect highlights causing underxposure, even when those highlights are for example specular and should be blown out if a scene is capture properly. Simply shifting things left to protect specular highlights is not an application of the concept of ETTR. So why not post the image so that we can make an assessment of the value of the information.
Overall, nice site and some useful information here, but some of it seems like a little bit of smoke and mirrors.
It was not our label; it is the most common understanding of the subject we are running across over and over again. We were referring to the article http://www.luminous-landscape.com/tutorials/expose-right.shtml , where they say: "The simple lesson to be learned from this is to bias your exposures so that the histogram is snugged up to the right, but not to the point that the highlights are blown. This can usually be seen by the flashing alert on most camera review screens. Just back off so that the flashing stops." The article was written 5 years ago and for all that time since many do understand it exactly as "no flashing of highlights is allowed". Our standpoint is that the article desperately needs a revision.
Thank you for your comments, they will sure help readers to realise that incorrect application of ETTR rule defeats the purpose an is in fact practising the direct opposite, ETTL.
The practice of underexposing to prevent any possible hilight from being clipped is not exposing to the right. Its exposing to the left. It will produce a dark raw image which is then lightened in post (increasing the noise). It does, as you say, give more hilight headroom.
Exposing to the right is deliberate overexposure, to move the midtones into lighter regios where the linear sensor count makes better use of the available bit depth. This results in a light raw image which must be darkened in post (decreasing the noise). It does, as you say, give even less hilight headroom.
I agree with the main point of the article; the available dynamic range of the sensor is insufficient. A given exposure will have lost information from the scene, and this will always be the case until we have non-linear sensors with greater bit depth.
I also agree that using RGB histograms derived from a white balanced image (or worse, a single luminance histogram; or even worse, a single green histogram) does not allow for accurate judgment of how well the 0 - 4095 range has been used for each channel and whether severe under exposure has happened on one of them, or whether one or more channels have clipped. But that also won't change until we have cameras with open source firmware or, at least, the ability to accept on-camera plugins.
So really, all that I am disagreeing with is that you labelled the 'underexposure' method ETTR - which is confusing, and the real ETTR method has different defects than the ones you describe.
Lastly, interesting site and keep up the good work!
Exposing to the right in its extreme, in its blind application, that is while trying to keep all and every highlight from clipping, is exposing to the left when it comes to midtones and shadows.
The purpose of the article is to point out that with current metering there is at most 3 stops from the metering point to highlights; and to help realize that if the range of brightness found in the scene is higher then 3 stops from midpoint to highlights trying to keep highlights necessitates underexposure of midtones and shadows. The price for that and the latitude (that is, by how much one can underexpose still getting acceptable image quality) depends on the camera, ISO, light, colour of subjects in the scene, raw converter in use and some other factors.
I would say with current top cameras, especially in 14-bit mode, underexposure of 1.5 stops at base ISO is relatively safe for outdoor scenes. This allows to have close to 4.5 stops headroom in highlights.
It is also necessary to take into account that it is very difficult to judge the amount of underexposure using histograms on the back of the camera. In our opinion metering and trusting the meter is easier and more predictable for now.
This article argues correctly (here and in other articles on this site) that the single channel, white-balanced histogram is insufficient to detect hilight clipping or channel underexposure.
However, the table in this article is (as Peter points out) not comparing to ETTR but rather ETTL! Which will of course give the noise problems you mention. Underexposure is exposing to the left.
> Riechmann has expressed that poorly in his article
Not only poorly; but extremely misleading. His article causes a lot of headaches for pre-press people and for photographers. One part that is missing is about the range of the brightnesses in the original scene for ETTR to be useful. Nothing wrong with using ETTR given one knows how far shadows can be pushed and how to use exposure compensation with ETTR. We think Zone System gives a very good guideline when it comes to compensating ETTR. Never try to get details in Zone X. The limit to the upper Zone that can be captured is posed by the number of stops shadows can be pushed to become midtones.
See one of the next articles, http://www.libraw.org/articles/peace-in-lights.html
Just two quick comments:
1. Firstly the recording of the scene with the sole purpose of protecting the brightest highlight is a departure from the original concept of ETTR, although Riechmann has expressed that poorly in his article on the subject where it was first discussed. The intention of ETTR is to shift everything to the right, even though the original article is written with the quote "...bias your exposures so that the histogram is snugged up to the right, but not to the point that the highlights are blown." The intent however comes through clearly in the rest of the article where he says for example "...when you look at the RAW file in your favourite RAW processing software... ...the image will likely appear to be too light." and "...This will accomplish a number of things. The first is that it will maximize the signal to noise ratio. The second is that it will minimize the posterization and noise that potentially occurs in the darker regions of the image." So the whole purpose is to shift things to the right, not simply to blindly shift things to the left just to protect all highlights, including specular highlights, which should rightly be blown out.
2. ETTR is not a technique to use at any ISO other than the base or lowest ISO. If you are working at higher ISO, ETTR effectively lowers the ISO anyway and the best way to improve the image would be to shift the ISO rather than use ETTR. So exposing for the subject at higher ISO should be self-evident and ETTR should be ignored. It would be ridiculous to use ETTR and high ISO. That defies common sense. That is also touched on by Reichmann with this statement "...Also be aware though that by doing this you are in fact effectively lowering the ISO used to capture the image, requiring slower shutter speeds and/or larger apertures. If you are hand-holding the camera, or shooting moving objects, the tradeoff may not be worth the reduced noise level.".
So it would be good to see the scene that was shot to come up with these figures and to analyse it, because you shouldn't be shifting the subject left with ETTR, it should be going the opposite way.
Roger W. Hicks writes:
The darkest black achievable with a top-flight black and white paper, optimally processed, reflects about 1/200 as much light as the pure, paper-base white: a brightness range of 200:1. Few prints will achieve this: a more realistic brightness range is between 100:1 and 125:1. What is more, there is a difference between the total range -- the purest white and deepest black -- and the dynamic range, within which there is texture and detail. This is very unlikely to exceed 100:1, and may well be more like 64:1.
Numbers and diagrams are results of experimenting, and even contain a certain experimental error of about 0.5%..1% (see for example how 18% grey translates into L=49.5 instead of being exact L=50). The text is also based on our experience in coding raw converters. All the conclusions you can easily verify in direct experiments - a camera with a spotmeter, a grey card or a sheet of white paper that does not contain optical brightner, a source of light like a flashgun, a wide dynamic range scene like a landscape in a sunny day, and Rawnalyze software is all that is needed.
Centred exposure is an exposure that places the subject into one of three zones that result in most detailed print. In other words, it is the exposure that is centred on the proper exposure of the subject rather on the preserving extreme highlights. You may want to read the complimentary article Spot-Metering: Reading Adams in Reverse Direction for the suggestions on exposure compensation.
But sometimes, figures and samples speaks more than numbers and diagrams.
Do you have any example to show ?
And if I understand, when you speaks about centered exposure, you mean relying to in camera lightmeter without compensation ?
Thanks in advance
We've choosen another option
- public version of LibRaw is GPLed as includes GPLed code
- we'll offer personalized licenses to any developer (company or person) of specific software. Of course, personalized LibRaw will not contain GPLed Coffin's code.
It can be LGPLed with a simple caveat for the GPLed Foveon module.
By the way, I wrote about Libraw before you did it, here :
LibRaw uses GPLed Foveon decoder from dcraw.