Blacks in Lightroon 4 look different in Photoshop CS6

[AoxoA]

Created another profile.  Restarted the computer.  Restarted again zaping the p-ram.  Didn't help.

Ran a NEF through the paces and the result is the same as the Tiff and Jpeg.

Sounds like I need another calibration device,  I do have some old Spyder2Pro software.  I'll give that a try before I head to Amazon.

[rubencarmona]

spyder2 is indeed difficult, as it doesn't support wide gamut. I'm using a Spyder4 since 1 year, works great too... you need a new colorimeter to hardware-calibrate your wide gamut screen.

[Tim Lookingbill]

Saw the video to the end and I'm at a loss to figuring out the cause as well.

It hasn't been asked (or maybe it has and I missed it) but when did this start? What's changed from when it did work?

For me after ten years hardware calibrating at least 3 displays this is the first I've seen this happen.

Have Apple's Digital Colormeter open and sample the lighter blacks to the darker one while checking Photoshop's RGB Info Palette to see how way off the black point is from actual data. Maybe the data will read the same between the two which will definitely point to a profile issue.

[digitaldog]

Have Apple's Digital Colormeter open and sample the lighter blacks to the darker one while checking Photoshop's RGB Info Palette to see how way off the black point is from actual data.

How would ADC do anything helpful?

[Simon J.A. Simpson]

I have done some tests using Lightroom 5.7 and Photoshop CC 2014.2.1 (Mac OSX 10.7.5) and have been able to replicate the problem.

I do not believe the problem has anything to do with the monitor profile.

I used a variety of files, including a test file designed to test the clipping on displays (monitors and projectors).  Some files were RAWs but the test file is a PSD (Photoshop) file in the ProPhoto colour space.

With all the files a slight lightening of extreme blacks is visible in Photoshop relative to Lightroom.  Using Apple’s DigitalColor Meter to read the tones sent to the screen confirms a slight raising of the extreme blacks (2–3 points on the 0–256 scale).

So, it appears that it is the values being sent from Lightroom and Photoshop that are at variance and which give rise to visible difference in the extreme blacks.  If my understanding is correct then if it were a problem with the display profile, which the operating system uses to correct the display globally for all applications, then Photoshop and Lightroom would show no difference between them.  However, since Photoshop gives a different result from both Lightroom and Preview (see the OP’s video) then it would appear that the ‘fault’, if any, lies with Photoshop and the way it is handling the image data.

As far as the OP, AoxoA, is concerned it is most unfortunate that this shows so clearly in his image.  The good news is that at least Photoshop is not clipping the data so the situation is retrievable with some adjustments in Curves or Levels.  I would also suggest that, although irritating (and perhaps not up to the standard we should expect from Adobe) it is the final destination and how those blacks are displayed in a print, or by a monitor or projector, that is most important and the changes brought about by these devices will probably eclipse the small difference introduced by Photoshop.

[digitaldog]

I have done some tests using Lightroom 5.7 and Photoshop CC 2014.2.1 (Mac OSX 10.7.5) and have been able to replicate the problem.

Can you perhaps provide a file and exact steps you used? I just tried LR6CC and Photoshop CC2014, Edit in PS using ACR 9. Viewing very dark tones in both appear identical on this end. Viewing the image in Develop at 1:1, rendered image in ProPhoto RGB in PS at 100%. They match.

I do not believe the problem has anything to do with the monitor profile.

In theory that makes sense unless the issue were affecting just the preview, not the actual numbers.

[D Fosse]

(...) the display profile, which the operating system uses to correct the display globally for all applications

This is a very, very common misunderstanding. No, the display profile is not used globally, and not by the OS. That's the calibration LUT - something else entirely.

The profile is a description of the display in a certain state - usually the calibrated state. It uses many more parameters and has a much higher precision level than the calibration curves. It is used by applications, not the OS. The OS just makes it available for the application.

Only color managed apps use the profile, and in a perfectly standard profile conversion: source profile > destination profile. And even though the destination is the same display profile for both Lightroom and Photoshop, the source is very different. So the conversion is different, and thus a bad profile can show up differently between apps.

The basis for this confusion between profile and LUT is probably that the LUT is usually stored inside the profile. This is just for convenience. The two are functionally unrelated and perform different functions.

[Simon J.A. Simpson]

Can you perhaps provide a file and exact steps you used? I just tried LR6CC and Photoshop CC2014, Edit in PS using ACR 9. Viewing very dark tones in both appear identical on this end. Viewing the image in Develop at 1:1, rendered image in ProPhoto RGB in PS at 100%.

Andrew, did you measure the values using the Apple DigitalColor Meter ?  Appearances can be deceptive (See below).

[Simon J.A. Simpson]

This is a very, very common misunderstanding. No, the display profile is not used globally, and not by the OS. That's the calibration LUT - something else entirely.

The profile is a description of the display in a certain state - usually the calibrated state. It uses many more parameters and has a much higher precision level than the calibration curves. It is used by applications, not the OS. The OS just makes it available for the application.

Only color managed apps use the profile, and in a perfectly standard profile conversion: source profile > destination profile. And even though the destination is the same display profile for both Lightroom and Photoshop, the source is very different. So the conversion is different, and thus a bad profile can show up differently between apps.

The basis for this confusion between profile and LUT is probably that the LUT is usually stored inside the profile. This is just for convenience. The two are functionally unrelated and perform different functions.

I stand corrected.

I have tried two other profiles and taken measurements using the Apple DigitalColor Meter.  One profile was the Apple supplied profile and one was generated by a different version of the ColorMunki software (see my other posts on a related issue on the LuLa Colour Management Forum).

The two profiles, visually, compress the extreme blacks to the extent that I would not rely on them for working on images.  Measuring the tones generated by each, using the DigitalColor Meter, supports the assertions by our more knowledgable brethren that each application uses the profile independently.  However, with these profiles the dark tones measured in Photoshop are darker than Lightroom, a reversal of the OP’s problem (and that generated by my other display profile).  Although this effect is not visually discernible because the extreme blacks are so compressed.

[Simon J.A. Simpson]

Can you perhaps provide a file and exact steps you used?

I would be delighted to provide a file but it is very large to upload.
You can download a small version from here:
Shadow and Highlight Compression Test SMALL (ProPhoto RGB)

You can make your own by creating a series of rectangles of steps of increasing tones from black to grey and from white to grey.  I use a reference circle in the middle of each rectangle of either black or white to make quick visual reference to see whether the display/projector is clipping.  The central square contains a black (zero) surround with an inner square of slightly lighter black (I think if memory serves this was at about “7”).  For the purposes of visually testing a display/projector for clipping the precise values are not important although they need to be stepped fairly close together at black and white ends of the tonal scale to be of use.  Neither are the precise values important to test the differences in ‘blacks’ rendering between Lightroom and Photoshop, it is the differences that matter.

The methodology is simple.  Open the test image in Develop in Lightroom and the same in Photoshop.  Use Apple's DigitalColor Meter to measure the same blacks in each.  Write down the numbers and compare.  QED.

Andrew, it would be great if you could give this a try and report back with some results.  We might be able to get enough information to isolate the problem.

[digitaldog]

Andrew, did you measure the values using the Apple DigitalColor Meter ?  Appearances can be deceptive (See below).

An unnecessary and not reliable product. As you can see below, using PS to examine a screen capture of the two side by side, they appear identical and measure identically. 3x3 sample, 22/22/22.

[Simon J.A. Simpson]

An unnecessary and not reliable product.

In your opinion.

I too get the same result using your methodology, but a different result using Apple’s DigitalColor Meter.  All this proves is that a difference can be measured using the DigitalColor Meter but not using a screenshot.  I question whether measuring a screenshot (and indeed using the DigitalColor Meter) captures the data actually sent to the screen*.  Also, the screenshot opens in Photoshop with the display profile attached.  Might this not have an effect on the results ?  In addition I have not measured any differences between Lightroom and Photoshop at RGB 22, as in your example above, but at values of between 0 and 10.  So I do not believe your test would show the differences observed and measured.  The use of a standard test image with values between zero and 15 would help here.

Andrew, if you say that you have not (yet) observed this phenomon then we need to try and establish why.  What we need to ascertain is whether it is the profiles causing the problem, or the way Photoshop/Lightroom send their data to the display, or some other as yet unidentified reason.  I do not think we have established this with any degree of reliability yet.

* tests done on ColorMunki profiles demonstrate this to be the case – the only reliable way to measure the effect of a profile is to actually measure the display itself – data sent to the screen is not necessarily what the screen is displaying.

[digitaldog]

I too get the same result using your methodology, but a different result using Apple’s DigitalColor Meter. 

I get the same in both. And ADC doesn't do anything useful and can't be set on two points at the same time, the screen capture using PS can. They appear the same, they measure the same. There's nothing going wrong on this end.

I question whether measuring a screenshot (and indeed using the DigitalColor Meter) captures the data actually sent to the screen*.

It's actually more reliable but let's not go there, the point is, the two image previews are identical.

Also, the screenshot opens in Photoshop with the display profile attached.

It can, but ADC has to use a single profile too so it's moot.

In addition I have not measured any differences between Lightroom and Photoshop at RGB 22, as in your example above, but at values of between 0 and 10. 

The Lstar value is 3, the RGB value will differ based on the RGB working space!

So I do not believe your test would show the differences observed and measured.

That's AOK with me, because there are no differences here between the two. You and the OP can attempt to figure out why either your systems or methodology differ. 

[digitaldog]

AR: An unnecessary and not reliable product.

In your opinion.

Explain this:




The ADC isn't measuring anything. It is taking two or three bits of information:

1. The color that an application is actually outputting to a pixel. i.e. an RGB level.
2. The colorspace that the app says should be used for that pixel for ColorSync to correctly display it. It defaults to sRGB if the app doesn't specify.
3. The ICC profile associated with the display as seen in the screen capture above.

Then given those bits of information, it can calculate via ColorSync what that particular pixel should be if you were to measure it with an external device, and all of the color transforms, profiles etc. are correct.

So it is sometimes useful if you really know what you are doing, to make sure that all your profiles are in order if you can measure patches with an external sensor.

It doesn't really do anything to validate anything without an external sensor, other than to verify numbers are what they should be.

It's "interesting" to see the various calculations that it can do, but there are a lot of assumptions and pitfalls along the way. For one, an application can bypass ColorSync and use it's own CMS.

[Tim Lookingbill]

I do remember this issue discussed before a while back and nothing was ever done about it nor was the cause determined.

Just did my own experiment and found the problem or the feature in LR4.4. You gotta' watch for it though.

I loaded a Raw image in LR4.4/Mac OS 10.6.8 with a known black level of around RGB-3,3,3 according to PS info palette in ProPhotoRGB working space zoomed in at 100% in Develop module. I then turned on Soft Proof where I have ProPhotoRGB as the output space (not a printer space) which gives me the same RGB numbers as Photoshop (instead of % readouts). I saw the 3,3,3 near black level actually go darker and now match how it appears in Photoshop.

Apple's Preview shows a slightly lighter black level exporting the Raw to tiff out of LR4 I'm guessing because whatever LR is doing with Soft Proofing turned on and set to a simple matrix profile such as ProPhotoRGB Preview doesn't implement quite exactly with regard to mapping near black. When I turn Soft Proof off in LR4, the black level lightens very noticeably. Maybe this is how LR4 renders black point in its 1.0 gamma ProPhotoRGB internal space since Soft Proof is turned off and there is no defined output space. I remember something similar happening in Photoshop converting to a gamma 1 simple matrix profile several years ago.

So if Soft Proof was left on or off while editing the image upon first opening in LR4's Develop module, black levels will need to be defined according to Soft Proof.

I don't know if this is a feature or a bug in LR4, now for sure this is just another reason why I edit in the less complex world of CS5 ACR 6.7.

My monitor profile is made by Colormunki Display.

[Simon J.A. Simpson]

I get the same in both. And ADC doesn't do anything useful and can't be set on two points at the same time, the screen capture using PS can. They appear the same, they measure the same. There's nothing going wrong on this end.

I know.  This is the point, why are you not seeing this whilst others are ?  What differences between your and our software/hardware/set-up account for this ?

Explain this:

You explained it very well, and so thank you.  However, I am unable to reproduce the differences you demonstrate on a system running OSX 10.7.5 with ADCM 4.4.  ‘Measurements’ from ADCM are completely congruent with the the Eye Dropper tool in Photoshop – in all the variations you show.

Nevertheless, this is a red herring and of no use to the OP who needs the help of someone with your knowledge and expertise to help him resolve his problem.  So I will now retire gracefully from this discussion.

[digitaldog]

Nevertheless, this is a red herring and of no use to the OP who needs the help of someone with your knowledge and expertise to help him resolve his problem. 

In your opinion. 

[Simon J.A. Simpson]

In your opinion. 

Touché !!!
 

[digitaldog]

I thought you were going retire gracefully.
Sorry if pointing out the holes in your testing methodology upset you.

[Tim Lookingbill]

Putting aside the ADC debate what is the cause of Soft Proof making black level previews change switching to ProPhotoRGB as a Soft Proof output profile in LR4.4?

That pretty much addresses the OP's issue.

I suggested the use of the ADC because it helps me to show the rate of change that is being applied to the video card frame buffer (x/y color table for every pixel on the screen) which the ADC reflects and derives its numbers. It's not meant as an accurate indicator of actual color according to a hardware sensor meter. Besides we're only addressing black level luminance not color. I use ADC to make sure the subtle visual differences of black level changes in LR4 turning off/on Soft Proof isn't caused by adaptive optical influences of the surrounding GUI.

And the changing black level in LR by turning off/on Soft Proof using a matrix profile shows up even in "Fill" zoom view, not just 1:1 previews.