Alt + Exposure control with ACR 7.1 limited

[bjanes]

With previous versions (PV2010 and earlier) of ACR one could set the highlights by holding down the ALT key and adjusting the exposure slider until  the screen preview showed no clipping. With PV2012 auto white point has been introduced and the preview shows no clipping even though the raw file is massively clipped.

Here is an example. With this image the ACR histogram shows clipping of the red and green channels.



This is confirmed by looking at the histogram of the raw file in Rawdigger.



However, using the Alt + exposure, white, or highlight slider shows no no significant clipping.



ACR really should incorporate a raw histogram or some other means of evaluating clipping in the raw file so that one can easily determine if ETTR has been carried too far.

Regards,

Bill

[madmanchan]

The Exposure control in PV 2012 should not be used to set the highlights.  That's a mistake -- a carryover method from PV 2003/2010 which should be abandoned for PV 2012.

To set the highlight/white clipping point in PV 2012, you need to use the Whites slider.  You can use Alt/Option+Whites to get the visualization of where clipping happens.

The ACR histogram you show in your top screenshot does not actually show clipping of the red/green channels.  The histogram, which shows rendered output (not input), is very close to clipping, but not exactly clipping.  There is a very small "toe" to the right.  So, the values are actually just shy of clipping.

[bjanes]

The Exposure control in PV 2012 should not be used to set the highlights.  That's a mistake -- a carryover method from PV 2003/2010 which should be abandoned for PV 2012.

To set the highlight/white clipping point in PV 2012, you need to use the Whites slider.  You can use Alt/Option+Whites to get the visualization of where clipping happens.

The ACR histogram you show in your top screenshot does not actually show clipping of the red/green channels.  The histogram, which shows rendered output (not input), is very close to clipping, but not exactly clipping.  There is a very small "toe" to the right.  So, the values are actually just shy of clipping.

Eric,

Thanks for your observations. Since the green and red channels are clipped in the raw file the image is overexposed, and the overexposure is about 0.3 EV as judged by an exposure of 0.3 EV less where the green channels are just short of clipping. My problem is that the automatic white point setting in PV2012 gives no indication of clipping in the raw file. For practitioners of ETTR it is important to know when clipping in the raw file has occurred and it would be helpful to have a raw histogram so that one would know when the raw channels are clipped.

Since the image is overexposed by 0.3 EV, all tones are shifted to the right. Setting the white point with the white slider would move the white point but the other tones would still be too high. I would think that one should use -0.3 EV exposure in this case. However, this is complicated by the hot tone curve that PV2012 uses for my Nikon D3 with the Adobe Standard profile. A gray card exposure gives a 12 bit raw value of 504, which corresponds to a gamma 2.2 value of 98. This is what is expected, since the camera allows 0.5 EV of highlight headroom. However, rendering into Adobe RGB gives a pixel value of 162. Therefore, one would have to use a negative exposure value of less than -0.3 EV.

Since the automatic white point rendering in PV2012 results in no clipping, the alt+white or alt+exposure preview is of no use in adjusting the exposure. I understand that one should set the midtones with exposure with PV2012 and then the white point. How would you suggest proceeding with PV2012?

Regards,

Bill

[RFPhotography]

Use Whites to set the white point then use the other controls to adjust the rest of the tones.  It's a different approach from previous versions but useless it's not.

If you don't like the tone curve that Adobe has built into its profile, then it seems the simple answer is to create your own.

[bjanes]

Use Whites to set the white point then use the other controls to adjust the rest of the tones.  It's a different approach from previous versions but useless it's not.

If you don't like the tone curve that Adobe has built into its profile, then it seems the simple answer is to create your own.

That is not the best approach. I would recommend setting the mid-tones with exposure first. Exposure affects the whole image. If you use other adjustments first, then you will likely have to revise them after adjusting exposure. Charles Cramer agrees.

As Eric Chan pointed out previously, the alt+exposure method is not appropriate for setting the highlights with PV2012. The alt+white does work does work and I should revise my original statement. The question of baseline exposure offsets has been previously discussed. Unless one uses a special program to look at the raw file, it is difficult to determine what offset one should use. IMHO 18% saturation in the raw file should yield a rendered pixel value of 100 in an 1.8 gamma space such as ProPhotoRGB. With my camera an exposure adjustment of -0.8 EV is necessary. To the uninitiated, this leads to complaints that the camera "overexposes".

Bill

[RFPhotography]

I wasn't suggesting an order of operations.  Merely that the Exposure slider works differently with the 2012 PV from 2010/2003.  Whatever order you decide to use them in, the mindset needs to change in terms of what the various adjustments do to the image.  You were trying to use the wrong tool for the job you wanted to do.  That's the point.  I could have said 'and' rather than 'then' which would have perhaps been clearer.  But I didn't.  Oh well.

[Peter_DL]

... You were trying to use the wrong tool for the job you wanted to do.  That's the point.

So which tool would we have to use to detect true-Raw-clipping by means of ACR ?
for example in order to determine if ETTR has been driven a bit too far.
THAT was the point here.

Peter

--

[digitaldog]

So which tool would we have to use to detect true-Raw-clipping by means of ACR ?

Detect or correct? There’s a difference. Exposure might allow you to detect clipping but what Eric and others are saying is it is the wrong tool to alter said clipping. That is the job of the White slider in PV2012.

If you can pony up a mere $25 for George Jardine’s new video’s on LR4 and PV2012, video #4 is worth the price of admission alone in seeing how all the new tools interact.

http://mulita.com/blog/?page_id=724

[RFPhotography]

Peter, the histogram.  And the clipping tools that are provided in LR/ACR.  Why wouldn't those be what should be used?

Eric notes that the histogram shows the rendered output rather than the raw input but it really shouldn't make a difference.  Whether the histogram shows absolutely raw, unaltered data or data that's adjusted based on the gamma encoding of a particular profile and the tone curve that Adobe puts into the camera profiles, you're still looking for clipping.  The highlight protection that's now embedded in the 2012PV takes care of some of the 'apparent' clipping that ETTR may show.  So all that need be done is use the tools available to adjust for any additional apparent clipping.  Turn on the highlight clipping warning on the histogram.  It will show the same thing as using Alt/Opt.  Using Alt/Opt on any of the sliders will show the apparent clipping depending on which end of the image you're looking at.  Exposure/Whites/Highlights will show it at the top end.  Shadows/Blacks will show it at the bottom end.  Same as the clipping warnings being turned on in the histogram.  It's a matter of what tool should be used to correct it. 

[bjanes]

The highlight protection that's now embedded in the 2012PV takes care of some of the 'apparent' clipping that ETTR may show.  So all that need be done is use the tools available to adjust for any additional apparent clipping.  Turn on the highlight clipping warning on the histogram.  It will show the same thing as using Alt/Opt.  Using Alt/Opt on any of the sliders will show the apparent clipping depending on which end of the image you're looking at.  Exposure/Whites/Highlights will show it at the top end.  Shadows/Blacks will show it at the bottom end.  Same as the clipping warnings being turned on in the histogram.  It's a matter of what tool should be used to correct it. 

I did some tests with a Stouffer wedge overexposed by 2/3 EV as judged by examining the raw files with Rawdigger. The green channels of the properly exposed shots were just short of clipping and the overexposed shot received 2/3 EV more exposure. Opening the raw file in ACR 7.1 with PV2012 and the highlight clipping indicator enabled showed that that step one was clipped. Step 2 appeared intact rather than blown, likely because of the highlight protection. The Alt+exp method showed the same results. The steps of the wedge are in 0.3 EV decrements. Step 8 corresponds to mid-gray which should have a pixel value of 100 in ProPhotoRGB. With the ACR defaults, mid gray is far to light as shown.



Note that white balance is off as indicated by the spikes of each step are out of alignment. Performing white balance with the eyedropper changes the clipping indicator, and no clipping is now apparent. The same applies to the Alt+white method. I think this is a bug and hope Eric is following this thread. In this situation, the clipping indicator and Alt+Exposure methods fail.



Correction for the clipped highlights depends on the image. If the shot is overexposed by 0.67 EV, all tones are lightened by this amount. One should use the exposure control to decrease exposure by the same amount. This provides a linear correction except for the highlights, which are afforded highlight protection. The decreased exposure darkens the highlights and a positive Whites adjustment is then needed. If one attempts to remove the clipping with the Whites control, the midtones are left too light. With this image a negative Whites adjustment of -37 is needed. To reproduce the appearance of the wedge, it is best first to set the midtones with exposure and then the white point.

Shown below is a graph of exposure adjustments. The highlights are rolled off smoothly to afford highlight protection and the midtones and shadows are decreased linearly as expected.



The whites control affects the near whites and has a relatively limited range.



If you want to determine if ETTR is carried out too far, it is best to look directly at the raw file with Rawdigger or a similar tool. The clipping indicator or Alt+exposure gives a reasonable result if the white balance bug is avoided.

If one is dealing with a high dynamic range subject and the midtones of the image are properly exposed but the highlights are blown, one should use the Whites control to bring the highlights down and leave exposure unchanged.

Bill

[bjanes]

Detect or correct? There’s a difference. Exposure might allow you to detect clipping but what Eric and others are saying is it is the wrong tool to alter said clipping. That is the job of the White slider in PV2012.

This is not necessarily true. Clipping due to global overexposure is best dealt with the Exposure control. See my post below. Eric has confirmed this in a previous exchange involving PV2010 and the same principle applies to PVw2012.

Bill

[RFPhotography]

Clipping due to global overexposure is best dealt with the Exposure control. Eric has confirmed this in a previous exchange involving PV2010 and the same principle applies to PVw2012.

Bill

No, it doesn't.  Eric has confirmed this.  The article you linked previously is in conflict with the portion I've bolded above.  The 2012 PV requires a different approach.  It's true that the Exposure control is a linear adjustment but it's more concentrated in the midtones.  That's what the article you linked earlier confirms.  Your graph confirms the same thing.  Your statement that it's best to set the white point then make an exposure adjustment conflicts with a post you made in this thread previously and the article you linked earlier. 

[digitaldog]

No, it doesn't.  Eric has confirmed this.  The article you linked previously is in conflict with the portion I've bolded above.  The 2012 PV requires a different approach. 

That’s my reading of Eric’s post (and George Jardine’s fine video’s on PV2012).

[bjanes]

No, it doesn't.  Eric has confirmed this.  The article you linked previously is in conflict with the portion I've bolded above.  The 2012 PV requires a different approach.  It's true that the Exposure control is a linear adjustment but it's more concentrated in the midtones.  That's what the article you linked earlier confirms.  Your graph confirms the same thing.  Your statement that it's best to set the white point then make an exposure adjustment conflicts with a post you made in this thread previously and the article you linked earlier. 

If you were able to read the graph I posted regarding the behavior of Exposure in PV2012 you would see that the adjustment is not concentrated in the midtones but rather is linear except for the highlights which are rolled off gently rather than abruptly clipped. Exposure is used to set the midtones, but the effect is not concentrated in the midtones as you suggest. PV2012 requires a different approach for setting the white point, since exposure rolls off rather than clipping the highlights as the graph shows. What is different between PV2010 and PV2012 is that exposure can no longer be used to set the white point since it does not clip. It is still the tool of choice to correct global overexposure. How would you correct for 2/3 EV overexposure?

Where did I say to set the white point and then adjust exposure? It is you who is inconsistent.

Bob:

"Use Whites to set the white point then use the other controls to adjust the rest of the tones.  It's a different approach from previous versions but useless it's not."

Bill:

"That is not the best approach. I would recommend setting the mid-tones with exposure first. Exposure affects the whole image. If you use other adjustments first, then you will likely have to revise them after adjusting exposure. Charles Cramer agrees. "

"To reproduce the appearance of the wedge, it is best first to set the midtones with exposure and then the white point."

[RFPhotography]

OK, I'm done.  You've gone off into ad hominem-land yet again. 

[JeffKohn]

While I appreciate the improvements to content-aware adjustments (highlight/shadow recovery), it seems to me like the controls in PV 2012 are actually less convenient to use if you knew what you were doing with the old process...

[Peter_DL]

So which tool would we have to use to detect true-Raw-clipping by means of ACR ?
for example in order to determine if ETTR has been driven a bit too far.

Detect or correct?

So far I was referring to "detect" only.

Just think about a series of EV-bracketed camera-exposures.
Which one is the correctly ETTR’d shot and which one already goes too far by clipping Raw channels ?

Peter

--

[digitaldog]

If you were able to read the graph I posted regarding the behavior of Exposure in PV2012 you would see that the adjustment is not concentrated in the midtones but rather is linear except for the highlights which are rolled off gently rather than abruptly clipped. Exposure is used to set the midtones, but the effect is not concentrated in the midtones as you suggest.

To some degree yes.

Plus Exposure moves everything up (Histogram to the right), and expands shadows (more contrast) and compresses (highlights less contrast). Exposure flattens (less contrast) and results in less saturation of highlight areas. IOW, there is more to all this than just moving the tones in one direction and they are not moved the same so I’m not sure about the use of linearity above. Yes, there is a roll off of highlights and a compression of contrast. Exposure isn't useful for flat and what appears as under exposed images. The result is just a brighter but flat image. In fact, Exposure flattens the highlights more.

Applying Exposure to a 21 step wedge as George shows in his video is quite useful. Here’s one I mocked up.

[bjanes]

OK, I'm done.  You've gone off into ad hominem-land yet again. 

ad hominem, really? My reply was based on logic and experimental data in so far as possible. However, it was necessary to point out that you did not know how to interpret a simple graph where the exposure curves are parallel except for the highlights. Also I thought I should point out that you misquoted me. If you have nothing further to contribute, it is best to withdraw from the discussion.

Bill

[bjanes]

To some degree yes.

Plus Exposure moves everything up (Histogram to the right), and expands shadows (more contrast) and compresses (highlights less contrast). Exposure flattens (less contrast) and results in less saturation of highlight areas. IOW, there is more to all this than just moving the tones in one direction and they are not moved the same so I’m not sure about the use of linearity above. Yes, there is a roll off of highlights and a compression of contrast. Exposure isn't useful for flat and what appears as under exposed images. The result is just a brighter but flat image. In fact, Exposure flattens the highlights more.

Applying Exposure to a 21 step wedge as George shows in his video is quite useful. Here’s one I mocked up.

Yes, evaluation of a step wedge is quite useful. I used a 41 step wedge but evaluated only 30 steps. One has to be careful about evaluating expansion and compression of tones with the ACR/LR histograms since AFAIK the x-axis is linear and represents values in a gamma encoded space. With a log base 2 x-axis the steps are uniform and express f/stops as shown in this histogram:



In a gamma 2.2 histogram (the gamma used for LR histograms), the dark tones appear compressed and the lighter tones appear expanded as shown in these histograms from Iris:



If one applies positive exposure in ACR/LR, the darker tones are moved to the right where they appear to be expanded because of the gamma encoding. To evaluate compression and expansion of tones one should use a log histogram. Guillermo Luijk's Histogrammar is one such utility that works with gamma encoded files.

Similar information can be obtained by examining the characteristic curve of the rendering with Imatest. Step 8 represents the mid gray and an exposure of -1.65 EV is necessary to place mid gray at a pixel value of 98 in ProPhotoRGB as shown here:



Here are the results of such an analysis with various exposure adjustments in ACR 7.01 using PV2012. The curve for the nominal -1.65 EV exposure is linear until about 1.5 stops from saturation, at which point highlight compression is appied using a sigmoidal contrast curve . If one applies a higher exposure value, the highlights are smoothly rolled off as Eric has stated in his posts. The curves are parallel at lower exposure values, showing that the response is linear (the tones are simply lifted by a uniform amount) and there is no shadow expansion. The shadows appear expanded in the linear histogram since they are moved to an area of the histogram where equal steps are farther apart.

I evaluated another image (03) which is overexposed by 0.67 EV. Using an exposure of -2.30, recovery is successful and the curve is superimposed on the normally exposed image, showing that negative exposure is the proper tool for dealing with globally overexposed images.



Regards,

Bill