Precise Digital Exposure

[Guillermo Luijk]

As soon as any big adjustment are done in post this seems to happen. The file may have a better SNR but the raw converter algorithms are not perfect. It is a big like making big EQ changes in audio.

Correcting exposure needs no algorithm.

If 1000 photons produce RAW level=500, then 2000 photons (+1EV in exposure) will produce RAW level=1000

So correcting RAW exposure is simply scaling all its RAW numbers by a constant factor. e.g. 1000/2 recovers 500. When color shifts take place is surely because some data got clipped in the RAW file or wrong post processing technique was applied.

This is a HDR mix of a 3-shot bracketing obtained at 2EV intervals (each gray color indicates the source RAW file). The resulting composite is color seamless. Exposure correction by -2EV and -4EV was just a matter of multiplying numbers by a constant value.





Regards!

[digitaldog]

did he still states that ICC profiles are not scene-referred ?

You should take this up with the ICC (you know about them right?). From their white paper #17, Using ICC profiles with digital camera images:

Also, ICC color management workflows generally assume that the colorimetry expressed in the PCS is of a [color-rendered] picture, and not of a scene. There is currently no mechanism to indicate that the colorimetry represented in the PCS by a camera profile is relative scene colorimetry. Even if there were, use of the PCS to contain relative scene colorimetry is not fully compatible with current ICC workflows, which assume color rendering has been performed. This distinction is especially important with respect to highlight reproduction. Many scenes contain highlights that are brighter than the tone in the scene that is reproduced as white in a picture. An important part of the color rendering process is selection of the tone in the scene that is considered "edge of white", and graceful compression of brighter tones to fit on the reproduction medium (between the "edge of white" tone and the medium white).

Maybe they will invite you to be a member of the ICC photography work group committee* and you can write your own piece on the subject for their site.

An ICC working group* has been formed to attempt to address issues with the use of ICC profiles in digital photography applications, but at present progress is difficult. Even if improved characterization targets (such as narrow-band emissive targets) and profiling tools are introduced, colorimetric intents will still be illumination specific, and perceptual intents will optimally be scene-specific. Some argue that scene-to-picture color rendering should be restricted to in- camera processing and camera raw processing applications, and correction of color rendering deficiencies limited to image editing applications.

[bernie west]

Correcting exposure needs no algorithm.

If 1000 photons produce RAW level=500, then 2000 photons (+1EV in exposure) will produce RAW level=1000

So correcting RAW exposure is simply scaling all its RAW numbers by a constant factor. e.g. 1000/2 recovers 500. When color shifts take place is surely because some data got clipped in the RAW file or wrong post processing technique was applied.

I think what MarkL was referring to (and I've wondered the same thing in the past) is pulling back exposure in LR, which is dependent on what demosaicing algorithm is used and what camera colour profile is used.  And then, as you say, you can get their attempt at rebuilt highlights if one or more channels has clipped.   Essentially, it seems as if the "exposure" slider in LR isn't a simple linear function.  Or perhaps it's linear over most of the range, but non-linear near the highlights.  I pretty much gave up on bothering with ETTR years ago due to the crappy renderings I was getting out of LR for images that were pushed right to the ETTR limit.  I got much better renderings detail and gradient wise with dcraw, but it was too much effort getting good colour due to the need to go through two different bits of software just for a half-decent initial rendering.  Then to photoshop if necessary after that.  Since pulling back just a tad from true ETTR, I've had much better results.  But if there is a fool proof way of normalising true ETTR images, then I'm all for learning it.

[BobD]

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.

Matthew,

Do you use a custom "camera profile" in your processing to assure that these "subtle and undesirable color shifts" are accurate to the scene's color? If not accurate then precise?

Bob 

[Internaut]

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.

This usually happens if one colour channel is blown.  I must admit, I don't fully understand the article, but the gist of it is, I think, being able to predict precisely how far you can push the exposure, from behind the camera, having metered for a given highlight (i.e. avoiding a problem we both encounter).  When I get time, I'll have a further read.  At the moment, my ETTR technique, with my smaller format cameras, is bracketing with matrix/pattern metering.  Modern sensors are good, so it's not unusual to find a +1 or +1.3 exposure with fully recoverable highlights, depending on the scene. 

Of course, the flip side of the debate is whether ETTR applies at all, with modern sensors, for most photographers and shooting situations.

[Torbjörn Tapani]

Of course, the flip side of the debate is whether ETTR applies at all, with modern sensors, for most photographers and shooting situations.

The way I see it, ETTR will always apply. Sure I know what you mean, we can recover shadows just fine on newer chips. But shot noise is a property of light itself. More exposure gives cleaner shadows, simple as what.

[Guillermo Luijk]

I think what MarkL was referring to (and I've wondered the same thing in the past) is pulling back exposure in LR, which is dependent on what demosaicing algorithm is used and what camera colour profile is used.  And then, as you say, you can get their attempt at rebuilt highlights if one or more channels has clipped.   Essentially, it seems as if the "exposure" slider in LR isn't a simple linear function.  Or perhaps it's linear over most of the range, but non-linear near the highlights.

It's possible but that would be a LR issue. In the example I posted the dark areas were pulled down by 4 stops (i.e. RAW linear values were divided by 16.0), and colour is perfect.

[digitaldog]

The way I see it, ETTR will always apply.

Yes, I agree, of course it will. ETTR is just optimal exposure. The exposure is either such it produces optimal data or it isn’t and there are degrees in which sub optimal data affects our work. Now what the comment might imply is that less than optimal exposure (ETTR) will produce results no one can see and that I suppose is possible. This is much like the use of editing in high bit (16-bit) because we know rounding errors could, possibly result in data loss that is visible at some point on some output devices. It might not. But why take the chance?

[BobD]

...perhaps it’s fair to say rendering is part development (normalizing ETTR) and part subjective (making ‘the print’).

Andrew,

I think there are parallels to B&W film processing and digital image processing. A while back, I posted on my blog what I think are 4-Phases of digital image processing and it’s parallel to black and white film. I think it fits this discussion well.

Here’s the link http://bobd.tv/lightroom/4-phases-of-processing-the-digital-image/
I’d like you to take a look at it and give me your thoughts.

BTW, I'm the guy that wrote the article on "The Optimum Digital Exposure" on LuLa. The one that Anders says “presents essentially nothing new".  Have you read my LuLa article? A lot of the concepts are from my book which, if your interested, I can send to you.

Bob DiNatale

[Rhossydd]

I posted on my blog what I think are 4-Phases of digital image processing and it’s parallel to black and white film. I think it fits this discussion well.

But it's wrong. See #reply 32 http://forum.luminous-landscape.com/index.php?topic=99565.msg815359#msg815359
Your idea of stage 1 'import' where you apply "Apply Camera Defaults" is NOT like film development because at any stage in the future those defaults can be modified and changed.

[AlterEgo]

Also, ICC color management workflows generally assume that the colorimetry expressed in the PCS is of a [color-rendered] picture, and not of a scene.

but we are not talking about PCS in camera (input) profiles in icc containers because dcp profiles are too directing color transform into some [color-rendered] picture, so you have absolutely the same situation... we are talking that the input when building such profiles is a scene-referred data

[MarkL]

I think what MarkL was referring to (and I've wondered the same thing in the past) is pulling back exposure in LR, which is dependent on what demosaicing algorithm is used and what camera colour profile is used.  And then, as you say, you can get their attempt at rebuilt highlights if one or more channels has clipped.   Essentially, it seems as if the "exposure" slider in LR isn't a simple linear function.  Or perhaps it's linear over most of the range, but non-linear near the highlights.  I pretty much gave up on bothering with ETTR years ago due to the crappy renderings I was getting out of LR for images that were pushed right to the ETTR limit.  I got much better renderings detail and gradient wise with dcraw, but it was too much effort getting good colour due to the need to go through two different bits of software just for a half-decent initial rendering.  Then to photoshop if necessary after that.  Since pulling back just a tad from true ETTR, I've had much better results.  But if there is a fool proof way of normalising true ETTR images, then I'm all for learning it.

Yes that's right. It might not need them but certainly in LR if you pull the exposure around you can see non-linear changes to the channels in the histogram.

[BobD]

Anders,

How did you arrive at your +2½ to 3 ½ stops for highlights? Trial and error? Experience?
“I was able to establish… key values to spot meter, where… the location of extreme highlight and shadow values with RGB values maintained or with recovery”

It doesn't need to be this complicated! Simply expose a gray card to +5 stops in 1/3 stop increments. This will produce 16 exposures. View these 16 exposures in your digital Raw Processing software and see which exposure reads 99% brightness - the Optimum White Point [OWP] for your system (meter/camera/software) combination.

My experience with all my cameras shows the [OWP] falls between +3_2/3 stops to +4_1/3 stop (arrived at by the exposure just before I reach LR's red “Highlights Clipping” warning). This is the Exposure Bias [EB] that needs to be applied to your spot meter reading of the brightest area in your scene. This will produce 99% brightness in your raw software- the [OWP].

I can appreciate you liking your Pentax spot meter… I had one also. However, I now use a Sekonic L-758. This is a wonderful tool because you can place the Exposer Bias [EB] you found in your testing into the L-758's ISO2 button. Now, you simply read the brightest part of the scene with your spot meter and press the ISO2 button… the L-758 displays the exposure settings needed to place the brightest part of the scene at the Optimum White Point [OWP] - 99% brightness in your software.

As far as applying the Zone System Process/Development routine to the digital process – this gets a little tricky. Digital images are so much different than photographic film images. A digital image is comprised of monochromatic mosaic linear B&W digital data. A photographic film image is a panchromatic emulsion with continuous tone densities that responds to non-linear (geometric progression) of light.

[BobD]

But it's wrong... Your idea o f stage 1 'import' where you apply "Apply Camera Defaults" is NOT like film development because at any stage in the future those defaults can be modified and changed.

Rhossydd

Yes. After Import there are more modification and changes with raw processing software. That is Phase2 “Global” and Phase3 “Selective”.

Phase2 “Global” in film/print processing is your base exposure set on your Gray Lab timer… “Minimum exposure for Maximum black”.
Phase3 “Selective” in film/print processing parallels “dodging & burning” done under the enlarger.
This is not even taking in to consideration that the digital tools use adaptive algorithms … neither linear nor geometric but adapting as needed based on the data it sees!

All these unique aspects of digital image processing mentioned above is the flaw in Anders' idea of trying to parallel digital processing to the Zone System.

[bjanes]

I think what MarkL was referring to (and I've wondered the same thing in the past) is pulling back exposure in LR, which is dependent on what demosaicing algorithm is used and what camera colour profile is used.  And then, as you say, you can get their attempt at rebuilt highlights if one or more channels has clipped.   Essentially, it seems as if the "exposure" slider in LR isn't a simple linear function.  Or perhaps it's linear over most of the range, but non-linear near the highlights.  I pretty much gave up on bothering with ETTR years ago due to the crappy renderings I was getting out of LR for images that were pushed right to the ETTR limit.  I got much better renderings detail and gradient wise with dcraw, but it was too much effort getting good colour due to the need to go through two different bits of software just for a half-decent initial rendering.  Then to photoshop if necessary after that.  Since pulling back just a tad from true ETTR, I've had much better results.  But if there is a fool proof way of normalising true ETTR images, then I'm all for learning it.

The behavior of the exposure slider in ACR/LR depends on which process version is in use. With the current process, PV2012, the exposure slider as well as all the other sliders in the basic panel are image adaptive. See the post by Eric Chan on the Adobe forums. Auto highlight recovery is always in use, which can hide overexposure in the raw file. With the earlier process version, PV2010, I think that the exposure slider was not image adaptive (that is was linear) unless highlight recovery is taking place.

The newer Adobe profiles have hue twists, which can cause problems when exposure and recovery are in use. See SandyMc's post here. These are introduced with profiles that have lookup tables in addition to the matrix math. Sandy's DCPtool an address this problem. The whole topic is complicated and beyond the scope of my expertise. However, in my experience the hue shifts are not problematic if there are no blown highlights.

Hopefully, some of forum heavies will enter into discussion.

Bill

[fdisilvestro]

in LR if you pull the exposure around you can see non-linear changes to the channels in the histogram.

That's expected because the scaling is performed on the raw channels, not the rendered RGB channels. Additionally the histogram representation in LR is not linear and the vertical axis is autoscaled.

You need a more reliable way to determine if there are color shifts or not

[AlterEgo]

That's expected because the scaling is performed on the raw channels, not the rendered RGB channels.

Adode code does not do exposure correction on raw channels, the data are already after demosaick and then after color transform...

[Anders_HK]

BTW, I'm the guy that wrote the article on "The Optimum Digital Exposure" on LuLa. The one that Anders says “presents essentially nothing new".  Have you read my LuLa article? A lot of the concepts are from my book which, if your interested, I can send to you.

My article - "A publication titled Perfect Exposure on my website dates from 2011 and was reviewed by Michael Reichman at the time. It points out both the simplified concept to upon one single spot meter reading determine exposure for precise location of the latitude of a sensor compared to a scene, and much more. 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."

[Anders_HK]

How did you arrive at your +2½ to 3 ½ stops for highlights? Trial and error? Experience?
“I was able to establish… key values to spot meter, where… the location of extreme highlight and shadow values with RGB values maintained or with recovery”
It doesn't need to be this complicated! Simply expose a gray card to +5 stops in 1/3 stop increments. This will produce 16 exposures. View these 16 exposures in your digital Raw Processing software and see which exposure reads 99% brightness - the Optimum White Point [OWP] for your system (meter/camera/software) combination.

Bob,

Precise Digital Exposure is not complicated, it is simple  . Yes, I followed the recommendation by Ansel Adams (his book The Negative) to expose at stop increments, and I applied by 1/3 increments to an 18% grey card for the full exposure range of my digital back. Included and next to the grey card I also had red, green, and blue paper cards. This is described in my paper.

Photographically speaking a digital sensor can be viewed no different than other photographic media such as was negative B&W film, slide film, negative colour film, Polaroid Land prints and more. We use aperture, shutter, film speed etc to control the exposure. Thereby we can test the spectral response (sensitivity to light) by same means as Ansel, which is what you also point out with the increments of stops in above. Therefore the principles of the Zone System apply to digital perfect well as they did to other media, however as already pointed out by Ansel they need an adaptation in how we apply them to a new media based on the characters of that media, which Ansel did to several media and which I have pointed out for digital in my article and paper. - A very essence, contrary to the belief of some posters in above, old and traditional methods of photography apply and they can be more precise for situations when we need or want so precise.

In my post above I attached the characteristic curve which I arrived at per increment photos of test cards and which is in my paper. Based on this I decided for which number of stops above mid tone that I was comfortable to spot meter before clipping of first channel and before clipping of last channel (which was based on what I viewed as viewable texture information, again similar to Adams did). I also do not want to risk clipping beyond these for readings with my spot meter, since I view important to spot meter just before these clippings occur. Because we have RGB sensors, if my understanding is correct the clipping may vary pending on sensitivity of the coloured filters over different sensors, and pending on any colour cast of the light. If you have looked into this please advise.

What I find advantage with the Pentax Spot Meter is the fact that I can with great simplicity view and visualise with a zone scale for the above highlight points and also for ease of inspecting/visualise also the shadow end. I am able to precise expose based upon one of, or two, highlight points, just before clipping of first and last channels, and can pending on if it is high dynamic scene also inspect at the shadow end. The disadvantage with the Pentax Spot Meter is that it is discontinued...

Anders

[Anders_HK]

AN ADDED EXPLANATION

In an attempt to explain more simple the effectiveness of the method of Precise Digital Exposure...

Having determined the following extreme points per 1/3 stops increment testing (textural maintained):
(A) Stops above 18% grey card just prior to clipping of first channel
(B) Stops above 18% grey card just prior to clipping of last channel
(C) Stops below 18% grey card just prior black point
(D) Same as C) with maximum negative exposure in post. My assumption is that deliberate negative exposure is only of interest to extend shadow end due benefits of ETTR to other data.

This brings a tool to determine a Precise Digital Exposure based upon the exposure range of my digital back compared to that of a scene. Hence my metering is simply a decision for determining the placing of the exposure range for a specific scene, and which in most simple application can be done upon one spot meter reading.

I can do so per:
1) Spot metering based on (A) for a brightest point in a scene - For simple situations this is all I need.
2) Spot metering based upon (A) and (B) - With sky and high dynamic range of sky this is what I can apply. Obvious, beyond clipping of one channel is not optimum data but what I find acceptable for sky etc. Per choice I can also decide exposure less than (B) for intention to later apply an extended shoulder in post to be able to create a more extended shoulder transition into highlights similar to film.
3) For a high dynamic range scene, inspect for (C) and (D). Further, pending on the scene at hand, I can check for what falls on -2 ~ -3 stops and upwards which is what for my digital back per my trial and error appear to maintain quality data when some require to be brought upwards in post.

Please refer to my paper for download, see link in article. Pointed out in it is more information and parallels referred to Ansel Adams principles.

Anders