Lightroom/Camera Raw: Camera Standard profile for A7R III is terribly off

  • 4
  • Problem
  • Updated 2 months ago
  • (Edited)
The Camera Standard profile Adobe has created for the A7R III adds a notable yellow cast to the image (unlike say Neutral or Vivid). Clearly not a match for the in-camera Standard profile when shooting jpg.Dear Adobe, can you please correct this?thanks
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes

Posted 10 months ago

  • 4
Photo of Rikk Flohr

Rikk Flohr, Champion

  • 1373 Posts
  • 335 Reply Likes
Some examples would help diagnose the issue.  Can you post screenshots? 

Do the exported versions of your pictures when using this profile still exhibit the yellow cast?
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
You can fix this! .dcp profiles are white balance agnostic and users must still apply WB.
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
@Rikk, will do when I have a sec

@Andrew I know I can still fix it changing the WB but if Adobe got the profile right in first instance I would save time...Canon ones are very good.  What is the origin of the yellow cast, is it color rendition or the profile is also applying it's own white balance on top of color rendition and tone? With the Canon profiles I never experiences any color cast/WB shift.
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
There is no 'right' here! You have a warm cast; fix it. The profiles don't handle this by design. Or roll your own profile for your specific sensor:
Everything you thought you wanted to know about DNG camera profiles:

All about In this 30 minute video, we’ll look into the creation and use of DNG camera profiles in three raw converters. The video covers:


What are DNG camera profiles, how do they differ from ICC camera profiles.

Misconceptions about DNG camera profiles.

Just when, and why do you need to build custom DNG camera profiles?

How to build custom DNG camera profiles using the X-rite Passport software.

The role of various illuminants on camera sensors and DNG camera profiles.

Dual Illuminant DNG camera profiles.

Examples of usage of DNG camera profiles in Lightroom, ACR, and Iridient Developer.


Low Rez (YouTube):

http://youtu.be/_fikTm8XIt4


High Rez (download):

http://www.digitaldog.net/files/DNG%20Camera%20profile%20video.mov

Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
I mean the other profiles (Adobe Standard, Sony Camera Neutral, Vivid etc) do not have the same cast, I assume it was a mistake on Adobe side when creating the profile rather than a  problem of my hardware?
Photo of Todd Shaner

Todd Shaner, Champion

  • 1091 Posts
  • 352 Reply Likes
I see no difference in the A7R III Camera Standard and Camera Neutral WB and no yellow tint on my 6500K  calibrated display. Images downloaded here:

http://www.imaging-resource.com/PRODS/sony-a7r-iii/sony-a7r-iiiTHMB.HTM

The embedded preview looks identical to the LR Camera Standard rendering. Ditto for the camera JPEG file at the above site. My guess is you have a monitor calibration issue download the files for your own comparison:

http://www.imaging-resource.com/PRODS/sony-a7r-iii/A7R3hSLI000100NR0.HTM

http://www.imaging-resource.com/PRODS/sony-a7r-iii/A7R3hSLI000100NR0.ARW.HTM

Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Todd, see my last message with the picture comparison below. My monitor is also calibrated to 6500k 
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
Don't make such assumptions. Do shoot say a color checker in daylight and provide  example here AFTER WB. 
(Edited)
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
would love to but I don't have a color checker. Will see what I can do but other people have noted the same yellow cast using that profile.

While you are here, can you take a look here? https://www.dpreview.com/forums/post/60635548
Thanks Andrew!
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
The DPR post is kind of meaningless. CCT values define a range (a large range) of possible colors and differing values can define the same color! 
http://digitaldog.net/files/22Thecolorofwhite.pdf
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
You are The expert on the subject and I only have limited knowledge of color science so i struggle to follow you. All in all shall I be happy with the hardware calibration and profiling I obtained vs my targets (D65 for whitepoint) or have to do it again?
Thanks
PS probably worth contuining the whitepoint discussion on dpr and leave this thread to the profile discussion
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
There is only one object that can produce D65 (an average of many measurements) and its 93 million miles from your display.
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
So the short answer is that I shall be happy with the result achieved by the calibration (despite the dispcal fail)? Would appreciate a straight, non technical answer, thanks
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
What calibration? You are working with generic .dcp profiles. For display, the values WILL vary.
(Edited)
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
:-) Difficult to get a straight answer from you. I am talking about the DPR post.
Shall I be happy with the results of that calibration (to D65) or not? If you can answer with a YES or NO would be great.
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
Watch the video! If you get a visual match to (whatever) that IS the correct number! If not, the calibration target is incorrect. Simple.
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
About display calibration targets:
Why are my prints too dark?
A video update to a written piece on subject from 2013

In this 24 minute video, I'll cover:

Are your prints really too dark?
Display calibration and WYSIWYG
Proper print viewing conditions
Trouble shooting to get a match
Avoiding kludges that don't solve the problem

High resolution: http://digitaldog.net/files/Why_are_m...
Low resolution: https://youtu.be/iS6sjZmxjY4
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
@Rikk

Going back to the main topic. Here is an example. 
Left is camera Standard, right is camera Neutral (also Adobe Standard looks similar)

You can see there is a yellow cast in camera Standard (eg color of the wall). Also, I was shooting in RAW + Jpg with Standard profile and in the jpg file there is no yellow cast.


Photo of Todd Shaner

Todd Shaner, Champion

  • 1091 Posts
  • 352 Reply Likes
Differences you see between the other profiles is probably due to higher color saturation with the Camera Standard profile. Do you have LR's WB panel set to 'As Shot?' If not that's what's causing the difference you're seeing between the camera JPEG and Camera Standard raw file rendering. If it is set to 'As Shot' please post a camera JPEG and ARW file pair that exhibit the issue to Dropbox and we can see what's happening on our systems. Thank you!
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
yes, as shot.

take this raw https://www.dropbox.com/s/fsmt09hr3pvouyv/2018_01_Ghana_637.dng?dl=0

apply Camera Standard and then compare with the related jpg
https://www.dropbox.com/s/mpgmgxstlpcfixx/DSC01411.JPG?dl=0

Thanks! 
Photo of Steve Sprengel

Steve Sprengel, Champion

  • 2656 Posts
  • 341 Reply Likes
Comparing the DNG and JPG images on DropBox, to me it looks like all the Camera xxxx profiles have stronger yellows.  This is NOT the same as saying there is a yellow color cast because neutrals still seem neutral so no overall cast where whites are yellow, but the yellows are stronger.  Since the image does not have a full range of colors it is difficult to tell if it is only yellows or various other colors--do you have any market scenes with a range of vibrant colors?

Here are some comparisons the camera JPG and camera DNG with no corrections in LR, using side-by-side Reference View (Shift-R) mode:

First, here is using the Adobe Standard profile, which has slightly muted colors and in my opinion even weaker yellows than the camera JPG:


Here is the Camera Standard comparison where the yellow on the wall at the top right is more intense:


Here is the somewhat closer result after messing with the Camera Calibration Hue sliders for a bit to make the Camera Standard DNG closer to the JPG, though it's difficult to say how well these adjustments would work with other images with more colors:


One final comment is that checking the JPG metadata in EXIF Tool, while most settings are Standard, it appears that Dynamic Range Optimizer was set to Auto and if the camera actually did some auto-fixing before creating the JPG this could make things slightly different than whatever Adobe used when they did their calibrations and can make matching a RAW to a JPG a moving target from one photo to the next:

(Edited)
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Thanks Steve, also had the same doubt about stronger yellows vs cast. But the issue seems with the profiles that Adobe has created. And Standard is the worst in terms of overboosting yellows. DRO should not affect color rendition of jpg, i think it affects tone curve slightly.

Is there a chance that Adobe will redo these camera profiles?

Will post scenes with more colors when i have a chance.
(Edited)
Photo of Todd Shaner

Todd Shaner, Champion

  • 1111 Posts
  • 360 Reply Likes
DRO should not affect color rendition of jpg, i think it affects tone curve slightly.
Are you sure about that? Try shooting with Dynamic Range Optimizer set to OFF and check the ARW file Camera Standard rendering to the Camera JPEG file. The Imaging Resource A7R III test images (shot under controlled conditions) match the camera JPEG file with Camera Standard profile, but 'Dynamic Range Optimizer is set to OFF. Your DNG file is also pulling some errors in EXIF tool related to the DNG file data also reported here:

https://forums.adobe.com/thread/1852381

It may be of no consequence, but I suggest checking it. Does the original ARW file render the same as the DNG file?
_____________________________________________________
---- ExifTool ----
ExifTool Version Number         : 10.25
Warning                         : Bad IDC_IFD SubDirectory start
Warning                         : Bad AdobeSR2 offset for MRWInfo
Warning                         : Error reading SR2 data
---- File ----
File Name                       : 2018_01_Ghana_637.dng
Dynamic Range Optimizer         : Auto
Dynamic Range Optimizer         : Standard
SR2 Sub IFD Offset              : 54016
SR2 Sub IFD Length              : 56958
SR2 Sub IFD Key                 : 0x44332211
____________________________________________________
---- ExifTool ----
ExifTool Version Number         : 10.25
---- File ----
File Name                       : A7R2hSLI000100NR0.ARW
Dynamic Range Optimizer         : Off
Dynamic Range Optimizer         : Off
SR2 Sub IFD Offset              : 50924
SR2 Sub IFD Length              : 56958
SR2 Sub IFD Key                 : 0x44332211
_____________________________________________________
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Todd, I am not sure about DRO but the manual says “By dividing the image into small areas, the product analyzes the contrast of light and shadow between the subject and the background, and creates an image with the optimal brightness and gradation.” So I would not expect to see different color saturation or rendition, but not sure. Will test it.

Interesting observation on the dng. Luckily I haven’t deleted the arw files yet (I usually keep only the dng) so i can test also that. You are saying that the same camera profile will give different results in dng and arw?
Photo of Jao van de Lagemaat

Jao van de Lagemaat

  • 175 Posts
  • 49 Reply Likes
It's highly likely that the dynamic range optimizer setting is the culprit. color perception is generally not intensity independent (except when you work in Lab color spaces) so this sort of dynamic optimization of local contrast and brightness really will mess with the perception of the color. If you shoot raw, you generally want to keep any dynamic range option turned off on your camera regardless of the camera maker.
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Jao do you know if DRO affects Raw files as well? And why the same file with Adobe Standard does not show the same problem?
Photo of Jao van de Lagemaat

Jao van de Lagemaat

  • 175 Posts
  • 49 Reply Likes
I don’t know about this camera but on canon and Nikon the comparable option will actually slightly underexpose the raw and write a flag in the raw file to compensate for this that only the camera maker’s software understands. This then leads to muted images in Lightroom that can be fixed by dialing in a bit of exposure compensation. The reason that camera profiles might react differently might be because they are somewhat “skewed” in that the perceived hue varies throughout its brightness range in order to try to reproduce the camera jpeg engine and Adobe standard perhaps has less of that. It is known that many jpeg engines and camera profiles do this.
Photo of Andrew Rodney

Andrew Rodney

  • 581 Posts
  • 99 Reply Likes
>>The Adobe  DNG converter is over-writing and/or truncating some of the Sony proprietary MakerNote fields in the DNG file.

At least one expert thinks otherwise and he's got a long history of being correct:
https://www.dpreview.com/forums/post/60650613
Photo of Todd Shaner

Todd Shaner, Champion

  • 1091 Posts
  • 352 Reply Likes
Yep, after working with the profiles in DPE it appears to be simply a color calibration issue. ;)
Photo of Steve Sprengel

Steve Sprengel, Champion

  • 2656 Posts
  • 341 Reply Likes
For fun, try taking a RAW+JPG photo of this color chart as displayed on your monitor, perhaps slightly out-of-focus to keep the moire patterns from the LCD dots from causing issues, and upload both the RAW+JPG so we can compare how things look with various color profiles.  Keep the DRO off.
https://www.atozdesigns.com/CMYK.jpg


Or if that has too many colors, try this one:
http://www.diem2.com/images/color-chart.jpg


I'd suggest opening them in LR and go to full-screen but zoom in or out to leave a bit of a margin of neutral gray around if you can, or black or white, to make the differences in colors more obvious and cut down on any distracting background in the photo.   Turn off the room lighting to keep screen reflections to a minimum.

White-balance the camera to a white screen on the monitor, first, so there isn't a cast from that.

The purpose of this exercise is to see if RAW and JPG look the same over a wide range of colors.
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Will do
Photo of Todd Shaner

Todd Shaner, Champion

  • 1111 Posts
  • 360 Reply Likes
So going by exclusion, if dng and arw look the same, and dro also does not explain the difference, then it is an adobe profile problem?
That would only exclude the DNG conversion and DRO as the cause. There still could be differences in other camera settings, camera defect, firmware difference that is causing the issue. Using the Imaging Resource A7R III test image files there is virtually NO difference in rendering between the camera JPEG and ARW file with Camera Standard profile. So by process of "exclusion" I can say with confidence there's something "different" with your camera settings or LR system rendering (installation, monitor profile, etc.).

Click on the below image to launch it an then right-click and select 'Save Image As.' Open the saved file in LR or PS and measure any of the color patches for comparison using the Eyedropper sampling tool. The only LR settings applied to the ARW file are 'Camera Standard' profile, 'As Shot' WB, and 'Exposure' = -0.12 to match the JPEG midtone reading.

(Edited)
Photo of Tiffany Fay J

Tiffany Fay J

  • 1 Post
  • 0 Reply Likes
Actually, the left image is more yellow in the sense that some of the pinks got stripped. Compare the left brownish color. The right one is much better color-wise. It looks more cardboard colored.
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Ok, I have done some of the home-works

1) Here is the arw file for the castle picture. When I apply Camera Standard in LR I get the same yellow boost I get with the dng, so I think we can exclude a dng conversion issue, agreed?
https://www.dropbox.com/s/qa0n3x0d7r4bq53/DSC01411.ARW?dl=0

2) Here are jpg and arw of a pic just taken with DRO off. When I apply Camera Standard to the raw the pic does not look bad but it is not a match to the jpg. I still see a notable yellow boost, most notable in the first floor of the building top left (eggshell paint). The pure white of the van, the red of the door and the blue of the car do not seem much affected, I don't see an overall yellow cast. Can we exclude it is a DRO issue?
https://www.dropbox.com/s/ulhb8xfrggi5xea/DSC02421.JPG?dl=0
https://www.dropbox.com/s/vcrbxiwsnwgbb70/DSC02421.ARW?dl=0


3) Here are the RAW and jpg of the color chart (DRO off) 
https://www.dropbox.com/s/lbcpuppcxna77tu/DSC02422.ARW?dl=0
https://www.dropbox.com/s/6eq3stnmfn723wr/DSC02422.JPG?dl=0
Not sure what conclusion to draw here.


It seems to me it is a yellow saturation issue of the profile created by Adobe, do you agree? I would not say it is an issue of the camera as the jpg look fine.
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Yeah, thanks Cameron, let me add that I know that a camera profile will never make my RAW to exactly match the jpg....but come closer though, like we are used with Canon. First Point.

Second point. The hypothesis Todd and I are making is that coming closer would be possible if only Adobe validated their Camera Standard profile against the jpg, while it looks like they tried to match the RAW+CS in Imaging Edge.

Cameron, would you reply to my earlier question, just in case you missed it, thanks.

"Because X-Rite's software doesn't generate the same forward matrix transforms as DNG Profile Editor or Lumariver (Which gives you the option to use ACR Matrices)"

"Thanks Cameron, meaning that DPE and Lumariver are superior?"
(Edited)
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Gents, someone senior from Adobe just "liked" my post ("Thanks, also my experience with 600D and 6D. ....") where I first hypothesised that Adobe targeted the wrong image for their profile. Looks like we are getting some attention hopefully. Would love to hear an official comment from them.
Photo of Andrew Rodney

Andrew Rodney

  • 581 Posts
  • 99 Reply Likes
Yes, the JPEG output's a desired rendering based on proprietary processing inside the camera. The raw can be rendered any way one desires within reason. 
I don't believe Adobe EVER guaranteed that their profiles will match the camera JPEG exactly nor should they. BTW, exactly would be within 1 or less deltaE; every pixel matches. 

A long time ago, in an ACR far, far away, there were no custom .dcp profiles and there was a way to 'calibrate' ACR's rendering, thought up by the late and great Bruce Fraser. He took a 24 patch Macbeth and shot a raw. Then, he adjusted the numerous sliders in the Calibration tab (which exist today because thankfully, Adobe rarely removes legacy features). The adjustments were such that each of the 24 patches produced Lab values after rendering that matched the Lab values of the Macbeth. A bit later on, a script was produced to do this kind of automatically as moving each slider and inspecting the 24 patches for targeted Lab values was a huge amount of work. The results in any case were 24 patches that colorimetrically matched; capture to scene! 

If you had a properly calibrated display, the MacBeth looked a lot like the actual target but keep in mind, the Cyan patch falls outside sRGB gamut and that's an 'issue' on sRGB gamut like displays. But you ended up with a capture we could state IS accurate! 

If you shoot lots of MacBeth's, you got accurate results, within reason. The illuminant still plays a role. White balance might too. Now you turn around and shoot something totally different from that Macbeth. Was it also accurate everywhere? Not always. Because digital cameras suffer what is know as metameric failure. We can go there but the point is, some colors under some illuminants under some conditions were not close to accurate. 

What some of you are doing is what we see above. You're not shooting a known color reference with known values in Lab. So when you view these images, your response to them being right is highly subjective. You may feel one rendering is 'too warm' but if I view that on a display that is run 'cooler' in white balance, I may not see that. Or dislike the rendering  as you did. That wouldn't be the case with the Macbeth! One can numerically see the Lab values which are published! Not at all the case with images shoot without such a reference using this tweaked profile (well tweaked calibration setup). 

Are Adobe's profiles perfect or close to that? No. But they try. Can they do better (we may see....). Is this mostly a subjective evaluation? So far yes. Will any profile fix the issues of metameric failure? Not at all. 

So we have tests without reference and we have assumptions that the results are an issue due to the profile. That's a huge leap to make. It may indeed be true but no colorimetric data has been provided that I've seen. I may have missed it. Can and should you consider making a custom profile and maybe tweaking it? I've said so too many times now. That again will not solve all the issues but it reduces some in some cases. 

Why are there a suite of profiles that differ? Because this is a subjective process. A scene with a prefect colorimetric match of a Macbeth in 25% of the image may look unacceptable in areas outside that 25%. Now what? 
Photo of Andrew Rodney

Andrew Rodney

  • 581 Posts
  • 99 Reply Likes
>>Sony Imaging Edge and Adobe both do not match the straight out of camera JPEG rendering.

Who (outside a marketing dept) guaranteed they would? To what average deltaE value?
What makes the camera JPEG so correct and special? It isn't. It's one rendering. If people love the camera JPEG rendering, they should just shoot JPEG! 
Photo of Cameron Rad

Cameron Rad

  • 75 Posts
  • 15 Reply Likes
I agree haha
Photo of Jao van de Lagemaat

Jao van de Lagemaat

  • 178 Posts
  • 51 Reply Likes
Camera standard is supposed to approximate the out-of-camera jpeg at all standard settings on the camera (no dynamic anything). If sony’s Cameras give different rendering for jpegs than when you run the raw through their own software I would have no clue what to try to approximate if I were Adobe. I would take the raw software as the reference normally.
Photo of Cameron Rad

Cameron Rad

  • 105 Posts
  • 27 Reply Likes
"Because X-Rite's software doesn't generate the same forward matrix transforms as DNG Profile Editor or Lumariver (Which gives you the option to use ACR Matrices)"

"Thanks Cameron, meaning that DPE and Lumariver are superior?"

No it just means it works differently. 
Photo of Mario Adario

Mario Adario

  • 85 Posts
  • 2 Reply Likes
Well now that they are reading this let's not give them an excuse not to at least try it! :-) 
Photo of Andrew Rodney

Andrew Rodney

  • 592 Posts
  • 102 Reply Likes
>>Well now that they are reading this let's not give them an excuse not to at least try it! :-) 

NO profile will fix metameric failure of a camera sensor.