Precise Digital Exposure

[fdisilvestro]

Hi,

I'm not sure if I understood correctly the article in the home page, but "Precise Digital Exposure" using the rendered RGB values in Capture one? seriously?

[digitaldog]

In terms of the rendering, can we equate that with ‘development’? If so, can we separate exposure from development?
That said, I glanced at the piece and it’s very, very possible I didn’t understand correctly either.

[Jim Metzger]

 Lee Varis has had a book out since June 2010 called "Mastering Exposure and the Zone System for Digital Photographers". It is still available. There was a companion webcast at some point.

I found the application of traditional (film) photo technique to digital capture to be very enlightening.

[Guillermo Luijk]

"My publication [2011] is, as far as I am aware, first to apply the Zone System principles and the spot metering correct for digital, simply because I did so according to the basis Adams wrote of"

I am sorry to say that article is nothing new. Two years before that, in 2009, I wrote a whole article about ETTR and spot metering, using real RAW data and plotting log RAW histograms to find out how much highlight headroom spot metering provided in several real world situations with Canon DSLR cameras:

ETTR WITH SPOT METERING




And two years before that, in 2007, I developed a piece of software to calculate and plot the Zone System of a scene from a linear image of it (it can be applied to RAW data if used properly):

HISTOGRAMMAR



Regards

[bjanes]

Hi,

I'm not sure if I understood correctly the article in the home page, but "Precise Digital Exposure" using the rendered RGB values in Capture one? seriously?

One can apply a tone curve in C1 or ACR to simulate the shoulder of a film exposure, but the fact is that the digital sensor is linear. The author talks about the shoulder of his sensor, but in my experience with the Nikon D800e, the sensor is almost perfectly linear right up to clipping, which occurs abruptly. I think this is typical for most other digital sensors. In the linear scale, clipping is 2.5 stops from mid-gray (rather than the 4 EV values from Zone V to Zone IX, as the author correctly states). Many years ago, Bruce Fraser suggested that one could meter the highlights where it is necessary to have some image detail (which would expose as mid-gray in the raw file) and increase exposure by 2.5 stops above the metered value. This might be extended to 3 or 3.5 stops if you want to chance it with highlight recovery in the raw developer. This is nothing new.

I agree that one should not base exposure on rendered values, but rather on raw values as determined by RawDigger or a similar tool. I do have the Pentax digital meter that the author (and Adams) prefer, but find its use in the field too cumbersome. Exactly which highlight should be metered? In most cases, I rely on the camera histogram and blinking highlights, and allow slight highlight clipping since the histogram is conservative. With the extended dynamic range of the D800 and other modern sensors, precise ETTR is not necessary in most cases. One can get good results with less than optimal ETTR. In challenging situations, it is easier to bracket rather than fiddle around with "precise" exposure values. This gives possibility of HDR, if necessary.

This would be a good opportunity for others to state how they handle exposure in various situations.

Bill

[fdisilvestro]

I agree that one should not base exposure on rendered values, but rather on raw values as determined by RawDigger or a similar tool.

Exactly!

[hjulenissen]

I agree that one should not base exposure on rendered values, but rather on raw values as determined by RawDigger or a similar tool.

Or (to beat a dead horse):


Audio people get to read out levels right after the A/D converter, presented in several ways optimized for (among other things) tuning input sensitivity. How incredibly hard can it be for the camera manufacturers to do something similar? Surely, they must have a few people with basic signal processing skills and the ability to implement a simple gui component?

I am not sure what to make of the article. It seems "stuck" on the notion of mid-grey that may have served film well for 100+ years, but I don't see how relevant it is today.

-h

[bernie west]

I'm waiting for the article to be translated into English ( ) before I comment too much on it.  Basically I've got no idea what he is saying, but I'm interested to see how he thinks using the zone system can apply to digital.  I used to be heavily involved back in the day in the film vs digital war, and there was a thread on photo.net with some guy trying to apply the zone system to digital.  His theory was kooky, but that's all I can remember, and I can't seem to find the thread any more.  One thread I can't find, and one article I can't parse.    But I have had some fun rereading some of the ETTR threads.  It is amazing how so few people understood how the digital sensor worked (and I mean in principle, let alone specifics), and SO many people were resistant to the ideas of digital sensitometry.  Thank god most of that film vs digital bollocks is over.  Rereading the threads again, I am amazed that I and people like Andrew Rodney had the will power to keep explaining it over and over and over again...

[MarkL]

Film and digital are similar in that they record light but that is about where the similarities end. I don't really understand tortured theories to fit a system designed around film into digital. The process should match the technology.

[Anders_HK]

Apologies if what I explain seem confusing. After learning this method I have found it simple and useful as a tool for precise spot metering exposure as per the exact capture latitude of my digital back. In principle it is same as spot metering for different film was (which likewise were different media/technology). Thus it is a specialised exposure tool for when you need to spot meter for preciseness and for and based on the digital technology.

---

The attached is a characteristic curve for my Leaf AFi-II 12, made by shooting test shots of an 18% grey card in constant light, which was made same way as Ansel Adams explained in his book The Negative of making test images one step apart. I have also included curves for extreme adjusted highlight and shadow compensations and together with extreme negative shadow exposure.

What do the curves bring? These tell me the changes in rendering of tonal values when I develop from defaults to extreme adjustments in post. This enables, at time for spot metering of a scene, to visualise precise the exposure latitude compared to the scene. Also for making a choice for a purpose made transfer zone into shoulder.

Linear response? Please note there is actually a slight shoulder in highlight end at defaults for my digital back (I assume due settings/processing in Capture One).

[Anders_HK]

I'm not sure if I understood correctly the article in the home page, but "Precise Digital Exposure" using the rendered RGB values in Capture one? seriously?

In terms of the rendering, can we equate that with ‘development’? If so, can we separate exposure from development?
That said, I glanced at the piece and it’s very, very possible I didn’t understand correctly either.

I suggest to think in terms of, that a very basis of the zone system is as a tool for visualising by spot metering a scene before snapping a shot, thus for pre visualising what our captured information will be (to me at defaults in Capture One thru extreme adjustments in post). Same as for film, spot metering is for scenes that require such preciseness, or when we desire to work by the method.


"For correctness this all dates back to what Ansel Adams wrote relating to the Zone System and how to determine spot metering for black and white negative film, Polaroid Land Print and more. Many have misunderstood the Zone System and believe it to merely consist of how Adams applied it for black and white negative film. My publication is, as far as I am aware, first to apply the Zone System principles and the spot metering correct for digital, simply because I did so according to the basis Adams wrote of. I thus tested the spectral response (sensitivity to light) for the particular sensor in my digital back and in detail evaluated the RGB values in my RAW converter of choice, Capture One."

As far as I know the above is correct, because - as stated - I went from the basis of what Adams wrote of for negative film, slide film, polaroid land prints, and applied it on basis of the precise response of my digital sensor. Please refer to my paper and Adams' book The Negative.

All photos in the article were spot metered. An example is in the tea house, which I spot metered from outside for sky and through a window for skin, thereafter went inside and within seconds made my exposure in a sudden appearance in front of the subjects.

Thank you for reading.

Anders

[bernie west]

I'm still not able to parse what you are saying very well, but I don't understand the point of spot-metering to visualise your exposure before you expose it.  You can pretty easily create a unitary white balance via Guillermo's work years ago and then your jpg histogram on the back of the camera will display a very close approximation of what is clipping in the raw data.  It's been years since I was involved in all that stuff, so I can't actually remember the specifics of it, particularly regarding whether the uni white balance takes account for the wb multipliers of greater than one in the green and red channels.  From memory, it did, I think, as the results it gave were really very close to the actual sensor clipping point.  So, using a uni white balance, you can just adjust your exposure with EC in camera till you get the highlights just about clipping.

[bjanes]

Apologies if what I explain seem confusing. After learning this method I have found it simple and useful as a tool for precise spot metering exposure as per the exact capture latitude of my digital back. In principle it is same as spot metering for different film was (which likewise were different media/technology). Thus it is a specialised exposure tool for when you need to spot meter for preciseness and for and based on the digital technology.

---

The attached is a characteristic curve for my Leaf AFi-II 12, made by shooting test shots of an 18% grey card in constant light, which was made same way as Ansel Adams explained in his book The Negative of making test images one step apart. I have also included curves for extreme adjusted highlight and shadow compensations and together with extreme negative shadow exposure.

What do the curves bring? These tell me the changes in rendering of tonal values when I develop from defaults to extreme adjustments in post. This enables, at time for spot metering of a scene, to visualise precise the exposure latitude compared to the scene. Also for making a choice for a purpose made transfer zone into shoulder.

Linear response? Please note there is actually a slight shoulder in highlight end at defaults for my digital back (I assume due settings/processing in Capture One).

You are examining the rendered file, not the raw values. The shoulder and toe are added from the rendering and are not in the raw file. In terms of Ansel Adams, the raw file is your negative and the rendered file represents the manipulations added in printing. Here are results from the Nikon D800e showing a linear curve.

Bill

[fdisilvestro]

In terms of Ansel Adams, the raw file is your negative and the rendered file represents the manipulations added in printing.

+1

Rawdigger is an excellent tool to analyse raw files and build the response curve of a sensor

[digitaldog]

In terms of Ansel Adams, the raw file is your negative and the rendered file represents the manipulations added in printing.

Development of neg? I asked earlier but haven’t heard back about that.
I’d agree, using a tool to examine the raw itself reduces any other process and tells us more truthful information about just exposure. I wish we had such tool on the damn camera. But in terms of the article and the rendered values that were not defined in highlights, isn’t that fair game since rendering is (I believe) akin to the development of the neg/raw?

[fdisilvestro]

Development of neg? I asked earlier but haven’t heard back about that.
I’d agree, using a tool to examine the raw itself reduces any other process and tells us more truthful information about just exposure. I wish we had such tool on the damn camera. But in terms of the article and the rendered values that were not defined in highlights, isn’t that fair game since rendering is (I believe) akin to the development of the neg/raw?

You could look at it in that way, but how do you know that your rendering is optimal? It would look as if you don't have control on the rendering, which is not the case.

With a negative (film) once you develop it, there is not much you can do. With Raw you can go back and re-render.

[MatthewCromer]

I've found that ETTR techniques usually result in subtle and undesirable color shifts in the finished result. I'd rather have better colors and tonalities and a bit more noise.

[bernie west]

Development of neg? I asked earlier but haven’t heard back about that.
I’d agree, using a tool to examine the raw itself reduces any other process and tells us more truthful information about just exposure. I wish we had such tool on the damn camera.

Do any camera makers make an SDK available to Joe Public to fiddle around with?  Even if you couldn't update the firmware because they want to keep it proprietary, you could perhaps create a phone app that could wirelessly tether with your camera and give you a raw histogram on your phone. 

[digitaldog]

You could look at it in that way, but how do you know that your rendering is optimal?

I’d consider that somewhat (or highly) subjective. I’d prefer not to blow out highlights but if that’s the intent of the image creator, so be it. Blocking up shadows is fine.

With a negative (film) once you develop it, there is not much you can do.

To a degree yes, but you now have to make a print. So perhaps it’s fair to say rendering is part development (normalizing ETTR) and part subjective (making ‘the print’).

[Rhossydd]

since rendering is (I believe) akin to the development of the neg/raw?

I don't think this is correct or a useful analogy.
With film 'development' was a process that could be manipulated to change the characteristics of the latent image, but on a 'once only' basis after exposure(capture). In a digital system, there's little other than choosing an ISO setting that can change the characteristics of the medium once you've chosen the camera(sensor) to use.
With digital, once that RAW image is saved, that's it, it's complete and invariable.
People refer to converting raw files as 'development' or rendering, but really the process is more akin to printing in analogue terms. You have a constant master raw file that can be output in various ways dependant on skill and the technology to render the data.

I just fail to understand why people still bother with such out dated concepts as the zone system. It was designed in the days when taking multi different exposures was impossible/time consuming/expensive, when there were post exposure possibilities that needed consideration and there was no immediate feedback of results. That just isn't the case now, you can cover all exposure variables in a swift bracketed burst that leaves far more options at negligible/no cost.