Camera Raw rendering vs Photoshop

[efreet_cs]

Hello,

My blacks in Camera Raw are crushed, while in photoshop not.

If i open a raw in Camera Raw, and open it in Photoshop without changing settings , the blacks in Camera Raw were crushed, and in photoshop, it's normal.

When i look the raw in bridge, blacks are fines too.

Why my camera raw crushes the black ? (and some red tones too in the strong shadows) ?

[JJMack]

I have trouble understanding what your mean.  Where  do you see the colors different.  Photoshop can not process RAW files it uses the ACR Plug-in to convert the RAW data into and RGB image that can be opened as a smart object layer or a raster layer.

[efreet_cs]

1) I open photoshop

2) I open a raw (and it launch Camera Raw so)

3) I can see the blacks crushed in camera raw. I open my I do "open" / "open as smart object"

4) In Photoshop, Black are normal, not clipped

5) I open the Raw in Bridge

6) I press space-barre to load the image in full screen (the raw loads to it's full quality), and black are normal not clipped.

7) I right click (still in Bridge) and "open in camera raw"?

8) black are clipped in Camera Raw

So, if i open a raw in Camera Raw, from Bridge, or from Photoshop, blacks are clipped.

[JJMack]

There does seem to be a differences in color on my machine too.  The colors in ACR seems more vivid or saturated.  But then I'm bit colorblind my displays are not the best and only calibrated via test web images use for calibration.  I'm not too obsessed with color as others and can always adjust colors to my liking.   In my mind many are too obsessed with colors and calibration I rather concentrate on printing image I like.

[efreet_cs]

Yes this is exactly this problem.
I can't understand why is there a difference, or why Camera Raw can't display correctly the good colors.

I tried on windows 10 calibrated with i1Profiler and macOS Sierra with colornavigator (same computer, bootcamp) and the two OS get the same issue.

If camera raw is not able to give a normal rendering, how is it possible to call it "professional tool" ?

[D Fosse]

Any difference between ACR and Photoshop means the monitor profile is bad. They should always display identically.

Even though they both end up in the same monitor profile, the source profiles are different, so the actual conversions are different. This means one may fail and the other work.

Rerun your calibration software to make a new profile. Make sure the profile policies are set to matrix-based (not table-based/LUT), and version 2 (not version 4). Both v4 and LUT are known to cause problems in some situations.

If that's not it, uncheck GPU from both apps' preferences. It's the same basic problem from a different angle: the GPU-resident color management can't handle the profile properly. Unchecking shifts display color management back to the CPU, which is more robust.

[efreet_cs]

my calibration is sRGB gamma 2.2 D65 v2 matrice in i1Profiler on windows 10

my calibration is sRGB gamma 2.2 D65 v2 gamma curve in colornavigator

the problem still occurs

[D Fosse · Accepted solution]

OK, then uncheck GPU from both apps' preferences and try again.

As long as the conversion from source/document profile into a valid monitor profile is correctly executed, there cannot be a difference between them. They should both display, simply, correctly.

So that conversion isn't performed correctly. With the GPU enabled, the conversion is done by the video driver. Off, it's done by the CPU. The latter is the traditional way, and is usually more reliable.

A potential source for errors here is that the source profile in ACR is linear TRC ProPhoto, in Photoshop it's sRGB/Adobe RGB/1.8 ProPhoto. So while they should both end up identically in monitor color space, the fact is that we have two different conversions, two different calculations. A problem may show up in only one of them.

[D Fosse]

Oh, one more thing - is this a dual display setup? In that case make sure that the right profile is really used for each display. ACR is known to occasionally use the wrong profile (for as of yet unknown reasons).

[efreet_cs]

Hello,

I finally found the problem :

While putting rendering intent to "relative" (default option in ps color settings ?), i get strange lights, and red in the shadows while opening some raw (raw adobe98, PS srgb), and with "perceptual" rendering intent option, all si finally fine !

normal lights, normal colors.

I tried a lot of calibrations with different mode etc... Always the same, all the calibration (v4.2, v2.2, gamma, lut, curve etc) works fine with "perceptual" in PS. But are wrong with "relative".

[D Fosse]

Well, that just has to be an obscure bug. The fact is that standard working color spaces don't even support perceptual intent. Whatever you set in Photoshop, all you ever get is relative colorimetric.

You should by definition not get any differences between those two.

I think you need to show in detail what it is you're doing here.

[D Fosse]

benjaminvingrief - is this on Mac OS Sierra?

Lightroom in pre-6/CC versions exhibit exactly this behavior in Sierra, and possibly it could be the same thing. Some API or framework has been changed in Sierra, without backwards compatibility, so that color management in these earlier versions ends up broken. The only option for users is to upgrade Lightroom.

Maybe flipping the rendering intent switch forces the app to use the Adobe CMM instead of the OS one. Something similar has been seen by going to soft proof in Lightroom.

This is a Mac-specific issue, so if you're on Windows this is not applicable.

I'd still like to see screenshots and more details (what calibrator, what settings, etc.)

[Stephen_A_Marsh]

While putting rendering intent to "relative" (default option in ps color settings ?), i get strange lights, and red in the shadows while opening some raw (raw adobe98, PS srgb), and with "perceptual" rendering intent option, all si finally fine !

normal lights, normal colors.

Do you have “black point compensation” (BPC) turned on when using relative colorimetric intent?

[D Fosse]

BPC should always be on. Without it you just get a flat cut-off at a certain value, and all original shadow separation below that point is lost.

In any case Photoshop color settings shouldn't matter here. PS color settings deal with policies for document profiles and conversions - but here we're talking about ACR display. That's a different operation altogether.

[Stephen_A_Marsh]

There were two screenshots – with different black behaviour, one ACR and the other Photoshop. As perceptual pretty much maps the same way in blacks as BPC on I was just checking if this could be part of the issue.

[magic carpet]

I don't use ACR often, but I had to check this out.

Surprisingly I had the same/similar bug using ACR.

In my case actually the whole color was displayed differently in ACR vs. opened in Photoshop after processing the raw to a smart object.

Re-opening the raw in ACR and changing form the "full screen mode" to the "not full screen mode", closing ACR and re-opening the raw in ACR (in stays in "non full screen mode") fixed it for me. Colors are now the same in ACR and PS, just how it should be, no matter what gpu settings you use ore if you use perceptive or relative mode.

Very strange thing.

However, this thread caught my attention because I currently have the very same problem with Hasselblad Phocus (latest version 3.1.5):

Blacks are displayed much deeper in Phocus as they are displayed in PS, where the rendering of the black is correct. It always stays like this no matter what I do. It even stays like this, if I use ONLY a MBP as single display. I know that we are on an adobe forum - but is anybody else experiencing this issue that blacks are deeper in Phocus then they are in PS (latest version 3.1.5) ?

system info:

El Capitan 10.11.6, PS CC 2017.1

MBP2014 hooked to a CG277 (calibrated using latest version of color navigator)

[efreet_cs]

Just to mention it :

I had made several posts about problem concerning "crushed black" with eizo CX271 connected to an iMac 27p. Problem in MacOSX but not in Windows 10 Bootcamp on the same computer... That's why i'm running Windows 10 under my iMac to get a "normal" black etc. Problem still here after more than one year, but this another thing than this post.

[magic carpet]

I just found the solution to my "different blacks in PS and phocus thing" over at the hasselblad forum:

here is the link to the topic at the hasselblad forum: crushed blacks/gamma – big problem

"Yes, I had this problem and it is solved by un-checking the box marked 'reflect black levels in tone curve' or something like that when you make your Eizo profile in Colornavigator.

It's an option in the advanced options, just before you place the sensor on the monitor to start the calibration. Sorry i can't be more specific, only on my laptop at the moment and the Eizo is in my studio."

[D Fosse]

There's a history of crushed blacks with ColorNavigator in MacOS. I do believe there's some sort of bug in ColorNavigator that only surfaces on Mac. Maybe they both have bugs that trigger each other.

There's a long thread in the Lightroom forum concerning Yosemite, and IIRC most if not all affected users had Eizos. And going full screen has triggered the same problem on and off for some time.

I have tried to find out what the "reflect black level in tone curve"-checkbox in CN does. The manual is less than helpful, it just states that "the black level is reflected in the tone curve", as if repeating it makes it any clearer. Speak loudly and they'll get it...

It can't be standard black point compensation, firstly because that shouldn't be an option, it should just always be on. Secondly because that option is only available for LUT profiles. With matrix profiles it's off and grayed out.

These particular problems have never been seen on Windows AFAIK (I haven't seen it, and I have looked). However, LUT profiles from CN have been problematic in some other more subtle ways, such as exaggerating the well-published ProPhoto shadow color banding issue in Photoshop (with GPU on). But there the problem already existed, it was just made worse.

In general, I would just advise everybody using ColorNavigator to set it to make matrix profiles (gamma value, as it's called there). That seems to be much safer.

[magic carpet]

Thank you for confirming this strange bug.

I'm glad that there is a workaround to get it fixed.

However, one must admit that macOS isn't the "it just works" experience anymore it used to be and I'm sure this bug has to do something with macOS and the interaction between colornavigator and macOS.

I researched the "reflect black level in tone curve" thing as well, but couldn't find an answer what it actually means and does.

[D Fosse]

What separates Mac and Windows here is that Mac OS performs all actual color management transforms on OS level. It's called by the application, but executed by the OS.

Windows, by contrast, leaves it all to the application and steps back. It makes profiles available as applications request them, but doesn't do anything with them. The application converts, with application code.

The Mac model is fine as long as it works, but it introduces an extra level where bugs can happen. And they do.

So whatever ColorNavigator bug is there, it apparently doesn't get to do much damage in Windows, but on Mac it's enough to trip the whole thing up.

Mac/Eizo users might want to see if there's an Eizo forum for ColorNavigator, and see what feedback they get there. I can't do it, since I'm on Windows and can't reproduce it.

[D Fosse]

Stephen_A_Marsh  wrote

There were two screenshots – with different black behaviour, one ACR and the other Photoshop. As perceptual pretty much maps the same way in blacks as BPC on I was just checking if this could be part of the issue.

Yes, sorry, I misread your post a bit. Lately I've seen some people saying you should turn BPC off, which is wrong. I jumped to conclusions.

[efreet_cs]

yes