Expose to right, it is as simple as

[John R Smith]

Nope, likely Hassy works same way as does Leica S2, Leaf and Phase One and derives the histogram from the raw data. See here http://www.luminous-landscape.com/forum/index.php?topic=49859.0

As Eric has said, NR reports that Leica derive the histogram from the JPEG preview. As far as my own CFV back goes, I have no firm data, but I notice all the time that the histogram and the firmware warnings report over exposure when the image is just fine in LR.

John

[NikoJorj]

Schewe's point is important.

I fully agree, and was trying to oversimplify a bit (or put in a better way, foresee the point where even Canon sensors will have too much DR - may not happen that soon).

To further my post #36 above there are in essence two means to meter precisely for ETTR without trial and error. Michael’s is for automation. The other is using a spot meter when knowing (having predetermined) the sensor’s precise limits of DR and simply knowing how many stops above and below mid tone it is capable to capture.

And you may add a method where some raw based blinking highlights are displayed while on live view, and you decide what to burn (that's close to trial and error but you don't waste a shoot, and it's not really slower than the spot method).

PS just to say the obvious : for zone system reasoning, keep in mind that there is a hard edge between Zone 0 (burnt, no detail at all) and Zone I (where the textures are the most detailed) and a very broad transition from Zone VI to zone X with the noise gradually kicking in, in contrary to a film shoulder and foot which are both gradual.
In the digital case the optimal zone is near the highlight edge (let's call that Zone I), whereas with film the optimal zone is in the middle of the linear part of the curve (which should be more or less around Zone V).

[stamper]

No, it won't make *my* photography any better -- I concede that it may make yours better. When I'm shooting, I don't have time to think about that stuff, because I'm thinking about too many other things. If all the engineers and Michael get together and decide Michael's article wasn't accurate, I'd be extremely interested in knowing that. What I then want from them is a recipe, or a prescription, or rules-of-thumb, for what is roughly, probably, maybe, the best practice under a set of given conditions, in and out of trees and hot sunlight, into heavily shaded doorways, through windows into the street, and from the street through windows to the inside. I don't expect the rules to give me perfect exposures, or even the best possible under the conditions, I just want them to be very good. I use automatic settings and autofocus a lot, because sometimes that's the best I can do with a reasonable chance of success. When possible, I'd like to go to the best fast manual set-up I can get, to see if I can kick up the quality a notch. I use rough ETTR rules some of the time, using ISOs of 160-800 (RAW) on several Panasonic m4/3 cameras with zoom lenses and a Pentax K5 with pancake primes and exposing to the right, as had been recommended here, and trying to get back as much as I can in Lightroom. That has actually seemed to me to work, though my shots fail a lot because of camera shake, poor framing, inaccurate focus and so on. If the engineers here would like to give me a set of rules-of-thumb (which is what I've done with Michael's articles -- I've boiled them down to rules) I would be very interested in seeing them, and grateful to anyone who'd provide them. That's about the most I can work with. I really don't have time for analysis or calculation, or even a lot of rules. Five rules would be about as many as I can handle; any more than that, and they'd slow me down too much.
 
JC 

I think if you have read and fully digested what has been written in this thread then there should be enough information for you to work out a method to suit your needs. It looks like you are a man in a hurry, more so than most photographers. I have certainly refined useful information from this thread. The forum at times is full of threads which imo isn't about photography. One recent one was about credit cards and the i phone. This thread is one of the best in a long while with respect to photography and Michael should be thanked for it.

[stamper]

PS just to say the obvious : for zone system reasoning, keep in mind that there is a hard edge between Zone 0 (burnt, no detail at all) and Zone I (where the textures are the most detailed) and a very broad transition from Zone VI to zone X with the noise gradually kicking in, in contrary to a film shoulder and foot which are both gradual.
In the digital case the optimal zone is near the highlight edge (let's call that Zone I), whereas with film the optimal zone is in the middle of the linear part of the curve (which should be more or less around Zone V).
   
Unquote

There isn't a perfect solution to everything. This shouldn't stop anyone from using the method. Using the digital zone where it is most useful and using what Michael has written in other circumstances is possibly a good work flow which I personally will pursue.

[Rob C]

"No, it won't make *my* photography any better -- I concede that it may make yours better. When I'm shooting, I don't have time to think about that stuff, because I'm thinking about too many other things. If all the engineers and Michael get together and decide Michael's article wasn't accurate, I'd be extremely interested in knowing that. What I then want from them is a recipe, or a prescription, or rules-of-thumb, for what is roughly, probably, maybe, the best practice under a set of given conditions, in and out of trees and hot sunlight, into heavily shaded doorways, through windows into the street, and from the street through windows to the inside. I don't expect the rules to give me perfect exposures, or even the best possible under the conditions, I just want them to be very good. I use automatic settings and autofocus a lot, because sometimes that's the best I can do with a reasonable chance of success. That's about the most I can work with. I really don't have time for analysis or calculation, or even a lot of rules. Five rules would be about as many as I can handle; any more than that, and they'd slow me down too much.  JC "

 
My edited cut from John’s post, above, really sums up the practicalities of photography as a thing you do, and not as some esoteric process that, unless you work in a studio on still life or/and with standardised lighting, generally happens more quickly than you think – in Ferris Beulers’s words: you could miss it.

My simple solution – with Nikon – is Matrix. Trust it.

In the case of something like a very narrow slice of the frame, say an outside scene shot from well within a darkish room, you can even apply the emergency film 'sunny sixteen' maxim of using the ISO speed as shutter setting and setting the aperture at around 11/16, but, of course, you use whatever combination of that value suits the image. If I can find it, here’s an example. It was easy with film transparencies – you used the incident light meter or, short of access to the same lighting source as on the subject, you could spot meter a white face and open up a stop-and-a-half. Done.

Rob C

[ErikKaffehr]

Hi,

The zone system was mainly used for negtive to positive process. Exposure was for shadow which is zone 1. The zone system adjusts film development as to obtain a complete tonal scale stretching from Zone 1 to Zone IX. Mapping of tone was done in printing.

With digital we don't have a film a tone curve, the sensor response is essentially linear, that is, the "curve" is almost a straiht line. So what we do is to make the straight line as long as possible (that is exposing to the right) and we bend it to our needs in development.

The photography of Ansel Adams built on intimate knowledge of all components of the process:

Choosing emulsion for subject
Choosing exposure for planned development
Choosing developer and development method
Choosing paper and developer for printing

Pretty much engineering...

Best regards
Erik

PS just to say the obvious : for zone system reasoning, keep in mind that there is a hard edge between Zone 0 (burnt, no detail at all) and Zone I (where the textures are the most detailed) and a very broad transition from Zone VI to zone X with the noise gradually kicking in, in contrary to a film shoulder and foot which are both gradual.
In the digital case the optimal zone is near the highlight edge (let's call that Zone I), whereas with film the optimal zone is in the middle of the linear part of the curve (which should be more or less around Zone V).
   
Unquote

There isn't a perfect solution to everything. This shouldn't stop anyone from using the method. Using the digital zone where it is most useful and using what Michael has written in other circumstances is possibly a good work flow which I personally will pursue.

[Anders_HK]

Yes, the zone system was simplified (read often applied) to B&W and in tad different way to slide film and polaroid. Albeit the interesting is that the fundamental zone system in its basics applies same to any media, also digital and ETTR.

Engineering? Same as Ansel but extension thereto is what we do from capture of image thru processing RAW thru print...  Things have changed, yet they really have not

[ednazarko]

I'm not sure that the participants on this forum, or even those who just drop by to read the features, represent the greater real world of camera users, which is why I think we're not likely to see an ETTR implementation any time soon.  (The one exception I could see may be in top of the line bodies, but even there, there point and shooters probably make up the majority of buyers.)  As long as the world of RAW shooters consists mostly of people who believe RAW means you don't have to pay any attention at all to camera settings, least of all exposure, you can just wrench a lovely image out of the file - ETTR bias in a camera will produce nothing but loud howling on the digital camera forums.  In workshops I've assisted, I can't remember one person who knew that the histograms were from JPG/sRGB representation of the data captured. There are times when I wanted to tape over the LCD...  I know there's a pervasive belief that doing your white balance in RAW saves everything -I like to show people files of jazz musicians shot in clubs with gelled stage lights - depending on the perversity of the person who picked the gels, sometimes there are channels completely devoid of information to work with.

I remember early digital point and shoots and DSLR bodies... on a couple of them, there was not such a huge bias exposure bias towards the dark side, and digital forums erupted in anger about blown highlights.  The generalized prescription became "set your meter to -EV.7, always".  Funny how that's hung on.  The dynamic range of the sensors was a lot smaller then - I always felt like I was shooting transparency film then, where now I feel like I'm shooting negative film.  In Ye Olde Filmic Tymes I remember having to shift mental gears on the few occasions when I shot negative film instead of transparencies.  My metering (and meter, marked up with pens) was biased towards maximum tolerable highlight loss because I usually shot transparencies, but when I had negative film in the camera, I had to adjust my thinking considerably to make sure I got enough info on the film.

My general solution is that I still carry an incident light meter, and I use it often in situations where I know there's a huge spread from brightest to darkest. There are very, very few situations where the incident reading results in a bad file, and I've learned what those situations look like... and a stop up and down bracket almost always get me home.

[digitaldog]

"No, it won't make *my* photography any better -- I concede that it may make yours better. When I'm shooting, I don't have time to think about that stuff, because I'm thinking about too many other things.

So the exposure of the image doesn’t enter the picture at all?

My simple solution – with Nikon – is Matrix. Trust it.

Well then you can use that simple solution AND expose properly for raw (ETTR) if your camera has something as simple as exposure compensation! It will not make your pictures better, agreed. It will make the data better, if that’s important to you (and I’ll freely admit, its not for everyone).

[digitaldog]

As long as the world of RAW shooters consists mostly of people who believe RAW means you don't have to pay any attention at all to camera settings, least of all exposure, you can just wrench a lovely image out of the file - ETTR bias in a camera will produce nothing but loud howling on the digital camera forums. 

I don’t believe that’s generally perceived by educated photographers who understand what raw provides. They know only exposure and ISO (and none of the other JPEG settings) affect the raw data. 

In workshops I've assisted, I can't remember one person who knew that the histograms were from JPG/sRGB representation of the data captured.

When shooting raw? That’s sad and I suggest not going to such workshops if that isn’t made abundantly clear to the audience. Note, the Histogram could represent an Adobe RGB (1998) representation of the JPEG data, not that helps the raw shooter.

There are times when I wanted to tape over the LCD... 

Lets not throw the baby out with the bath water. Its darn useful. Just tape over the Histogram (I keep mine off).

My general solution is that I still carry an incident light meter, and I use it often in situations where I know there's a huge spread from brightest to darkest. There are very, very few situations where the incident reading results in a bad file, and I've learned what those situations look like... and a stop up and down bracket almost always get me home.

I’d agree an incident meter tells us something different and often less of a lie than a reflective meter assuming one doesn’t fully understand the lie the reflective meter can provide. But using either, its important to understand that the meters are still thinking about film with a classic H&D curve, not linear encoded raw data and as such, even with the most accurate incident meter, setting the camera this way will not implement ETTR by itself. One still has to understand where and how highlights clip based on the raw data and the raw converter.

[michael]

I'm not sure that the participants on this forum, or even those who just drop by to read the features, represent the greater real world of camera users, which is why I think we're not likely to see an ETTR implementation any time soon.  (The one exception I could see may be in top of the line bodies, but even there, there point and shooters probably make up the majority of buyers.) 

Sorry, but I can't agree. This site gets about 1.3 million unique readers each month. Unlike many photo sites that are much bigger (DPReview and similar) this is a site about photography, not just reviews of the latest cameras. Technical articles, like the one on Optimizing Exposure, get literally hundreds of thousands of readers over the course of a year. This shows the level of interest.

Most casual photographers shoot JPGs and post to sites like Flicker. But there is a huge world-wide audience of people who care about the technology of photography and optimizing image quality. Manufacturers ignore them at their peril.

Michael

[Nick Rains]

Most casual photographers shoot JPGs and post to sites like Flicker. But there is a huge world-wide audience of people who care about the technology of photography and optimizing image quality. Manufacturers ignore them at their peril.

Michael

It's all about maximising the quality of your data capture, and also capturing all the data that is in the scene. If you have a scene that is too contrasty for a single capture, shoot more than one frame at different exposures. If the contrast of the scene will fit in the camera's DR then, if you want to squeeze the last drops of tonal quality out of your camera, then you need to put the data into the region of the camera's range which has the highest number of tonal steps available. ie ETTR. Even if the scene is of modest contrast (if I am working off a tripod) I will often shoot a second exposure at plus 2 just to capture better shadow detail. Then, if shadow detail is critical to the quality of my final image, I have a reserve of data I can dip into. Pixels are free, why not take that second shot, just in case you need its data? Its the work of 5 seconds to do and if you don't need it, well nevermind.

In the field, you do the best you can. If you have the time, shoot carefully using LiveView and watch your histograms. If you don't have time, trust your meter and learn how it behaves in different lighting situations.

[dreed]

Yep, in Camera to Print and Screen...

In terms of arrival date for this tutorial, what sort of expectations should we have? CY2012? CY2011?

[Richowens]

     1 week

[fdisilvestro]

The visual aid I'd like to have in the camera is similar to the blinking blown out highlight present today in most cameras, but implemented in a different way:

1.- based on Raw values
2.- Both highlights & shadows (maybe different color for each)
3.- User configurable thresholds

The user configurable threshold would be useful especially for the shadows, since it could be adjusted to the acceptable noise level (I.E. on a 0-255 scale I'd like to be warned about any value below 7).

This will help prevent important parts of the image getting down in the shadows while protecting the highlights.

[Rob C]

So the exposure of the image doesn’t enter the picture at all?

Well then you can use that simple solution AND expose properly for raw (ETTR) if your camera has something as simple as exposure compensation! It will not make your pictures better, agreed. It will make the data better, if that’s important to you (and I’ll freely admit, its not for everyone).

A bit of a mixed up response: the part of the post credited to me that instigated your first sentence was a quotation from John, not a direct one from myself! As for making my pictures better, that wouldn't demand an exposure function, it would require a mind transplant.

Regardless of who wrote what, I don't quite see how you can extrapolate what was written into a suggestion/interpretation that anyone was saying "exposure doesn't enter the picture at all"! Of course it does; that's why the plug for Matrix metering. It measures exposure and, without further alteration at shooting stage, gives pretty reliable results. I refer here to shooting where you don't have the luxury of either static subjects or plenty of repeat time, which I expect is closer to most realities than the heavy, dusty Gitzo and stepped exposures. At any rate, it sure is closer to mine and, from what I read, John's. No opinion caters perfectly for all - mine is jut what fits my current world.

Rob C

[digitaldog]

Of course it does; that's why the plug for Matrix metering. It measures exposure and, without further alteration at shooting stage, gives pretty reliable results.

Yes but not based on ETTR (not without having a specific idea how much more exposure can be used prior to blowing out highlights you don’t want blown out). Incident meter, reflective meter, matrix, all are only as good as the person setting them for ideal exposure (in this context raw). Simply setting the camera ISO (or meter ISO) and taking a reading doesn’t take ETTR into account. In fact you’ve got to test the sensor and the raw converter before you know how far you can expose to the right. The meters have no idea of this. Much as the ISO written on a box of film was, without any testing of exposure and development, simply a starting point. It gave pretty reliable results but hardly ideal results.

[bjanes]

Furthermore, it is also common knowledge that a sensor captures color-depths of 12 to 16 bit. So it makes no sense to show exposure in a heavy reduced range of 8-bit by first creating a jpg.
Robert

Robert,

Linear encoding as with current raw formats is very inefficient since it uses most of its bits in recording redundant levels towards the bright end of the image and too few bits for the shadow areas. Gamma encoding such as with JPEGs redistributes bits towards the shadows where the are needed. Norman Koren gives a good explanation on his web site. A gamma 2.2 8 bit image can represent the visible levels in a 12 bit linear file but barely. For true HDR, a linear 16 bit encoding is entirely inadequate. Gamma encoding extends the DR, but for real HDR one must go to log or floating point encoding. See Greg Ward.

Regards,

Bill

[fdisilvestro]

In theory, in linear encoding there is a direct relationship between the minimum number of bits needed for a given DR, so for 12 f/stops DR you need at least 12 bits (the inverse is not true, you will not increase the DR by increasing the number of bits, because noise will be a limiting factor)

With gamma encoding, this relation changes to minimum number of bits = gamma x DR, so with 8 bits and gamma 2.2 you could represent 8 x 2.2 = 17.6 f/stops DR from an original scene, more than your 12 or 14 original linear RAW.

Gamma encoding is not used only in JPegs, it is used also in Tiffs

This of course is theoretical, since if you don't have information to begin with, you will not have it at the end, but it is important to know that you cannot compare directly them (12-14 bits linear RAW vs 8 bit Jpeg / Tiff).

I'm a believer of editing in linear 16 bits, keep RAW as long as possible, but once you're done with your editing, a 2.2 Gamma encoded 8 bit TIFF is more than adequate for most uses (except maybe high end printing)

[bjanes]

With gamma encoding, this relation changes to minimum number of bits = gamma x DR, so with 8 bits and gamma 2.2 you could represent 8 x 2.2 = 17.6 f/stops DR from an original scene, more than your 12 or 14 original linear RAW.

I'm a believer of editing in linear 16 bits, keep RAW as long as possible, but once you're done with your editing, a 2.2 Gamma encoded 8 bit TIFF is more than adequate for most uses (except maybe high end printing)

I don't know where you get your 8 bit gamma 2.2 DR of 17.6 f/stops. In his post on encoding, Greg Ward gives the DR of 8 bit sRGB at 1.6 orders of magnitude. 10^1.6 = 39.81 or 5.32 f/stops.

Microsoft has proposed a scRGB standard for HDR. The first uses 12 bits per channel and a gamma of 2.2 with a linear ramp for the deep shadows and has a DR of 3.2 orders of magnitude or 10.6 stops. The other uses 16 bits per channel and a linear ramp, giving a DR of 3.5 orders of magnitude or 11.6 f/stops. See Table 1. This is for a cutoff error of 5% at the lower end of the ramp and differs from the oft quoted 1 f/stop per bit, where the error would be larger. He says that the eye can often notice differences of 2% in the shadows.

With these considerations in mind, I would not recommend 8 bit sRGB for current digital cameras with good DR, although it is suitable for most current printers. However, high DR monitors are becoming available and more than 8 bits are needed for them.

Regards,

Bill