One can argue that different methods produce different results and sometimes some results are better than other results depending on the images and what is needed to adjustment. But to argue one method is antique and "modern" methods are better (presumably because they are newer) is pretty naive and is letting a bit of an anti "something" to show through...PS/ACR/LR do things a certain way...if you like the results, keep learning how to use them better.
Hi Jeff,
I find that (learning how to use better, when it's wrong to begin with) a very unproductive approach, which avoids any chance of improvement of something that (according to many) does not work well at all. You seem to argue that when something (sort of) worked in the past, that newer insights will not produce better results, without even doing an honest verification of that hypothesis. I could reverse the argument and claim that it is that attitude that is pretty naive.
If you don't like the results Bart, feel free to use something else.
I do, but should I therefore stop trying to educate people about the way that their favorite applications could be improved? If they are unaware, will they ask for better tools? Or are you suggesting that users of an application should not ask for more than they get handed out, and act more like sheeple.
But calling somebody a fanboy is really kinda juvenile in any reasonable debate or discussion.
You are missing the point.
The Germans have a nice saying that pretty much sums it up for me: "Was sich liebt, das neckt sich". An English version could be something like: "the quarrel of lovers is the renewal of love".
It is
because I care about Photoshop as my main image processing tool, that I am critical about it's deficiencies. That is something that fanboys cannot understand, hence I'll let those whom the shoe fits wear it.
You have yet to prove your point regardless of the hand waving you've been doing :~)
I'll give it another try, although a true fanboy will not be convinced by any reasonable argument (regardless of the presentation), it's behavior that's also called "cognitive dissonance". I'm an optimist, so I believe there is hope for those who are willing, therefore another attempt.
Attached there is a ZIP archive with 3 images.
a) One image as downloaded from BabelColor.com is the source of this experiment, it's called "ColorChecker_sRGB_from_Avg_16bit.tif". I took that image because it is easily displayed even on displays with modest capabilities, and it has several common/relevant colors for investigation, like skin tones, and some tones we can find in nature/landscapes. There are other colors not in the chart that are more critical, like those of flowers. So the effects demonstrated in the ColorChecker, can become even more visible in other subject matter.
b) An image called "ColorChecker_sRGB_from_Avg_16bit_ACR
Contrast+000_ProPhotoRGB.tif" is a simple import of that file in ACR, and exported as a ProPhoto RGB version of the file, the default working space of many users. No changes were made, no sharpening, no noise reduction, nothing. This will serve as the baseline to compare against. So any hidden changes to the file data would be present in the file, just to make sure we will compare apples to apples.
c) An image called "ColorChecker_sRGB_from_Avg_16bit_ACR
Contrast+100_ProPhotoRGB.tif" after only applying a +100 Contrast adjustment, nothing else was changed.
The images 'b' and 'c' were saved without any further alterations as they came out of Photoshop's ACR plugin, so they are as you could create them from the original image 'a' yourself, but I've spared you the exercise and thus it is also possible for non-Photoshop users to follow the proceedings or for those who do not want to fiddle with their settings and are afraid to change their current setup.
Now, let's have a look at e.g. the first (Row 1, Column 1) , "Dark skin" patch.
After the ACR conversion to ProPhotoRGB, image 'b' now reads in 8-bit colors:
RGB=[81, 68, 55], Lab=[38, 13, 14], HSB=[30°, 33%, 32%]
After the ACR Contrast boost and conversion to ProPhotoRGB, image 'c' now reads in 8-bit colors:
RGB=[79, 60, 40], Lab=[36, 19, 23], HSB=[31°, 50%, 31%]
Because it is a darkish color, the contrast adjustment made it darker still, as intended. But it also increased the saturation, just look at the coordinates in Lab, _a and especially the _b values increased significantly, or the HSB coordinates which even show a slight Hue change, a significant Saturation increase by 17% (whereas a 1% difference is usually visible in a direct comparison).
I know that the +100 Contrast adjustment in ACR is extreme, but that's just to make it easier to see what also happens at lower settings, albeit obviously with lower amounts.
Let's have a look at e.g. the second (Row 1, Column 2) , "Light skin" patch.
After the ACR conversion to ProPhotoRGB, image 'b' now reads in 8-bit colors:
RGB=[157, 136, 113], Lab=[66, 16, 18], HSB=[30°, 28%, 62%]
After the ACR Contrast boost and conversion to ProPhotoRGB, image 'c' now reads in 8-bit colors:
RGB=[191, 162, 133], Lab=[75, 20, 22], HSB=[31°, 31%, 75%]
Again, we not only see an intended change in brightness, but also an additional increase in Saturation, and it's expected when we change brightness channels with the same amount in the R/G/B channels in RGB space instead of doing it in the correct manner. Because Luma and Chroma are not decoupled in RGB colorspace, a change to one will affect the other in the wrong way.
One more color patch, relevant to landscape photographers, the third (Row 1, Column 3) , "Blue Sky" patch.
After the ACR conversion to ProPhotoRGB, image 'b' now reads in 8-bit colors:
RGB=[95, 102, 134], Lab=[50, -4, -22], HSB=[230°, 29%, 53%]
After the ACR Contrast boost and conversion to ProPhotoRGB, image 'c' now reads in 8-bit colors:
RGB=[104, 114, 163], Lab=[55, -6, -31], HSB=[229°, 36%, 64%]
The results speak for themselves, Saturation is again exaggerated compared to the intended change in Brightness.
The amount of the Saturation change will vary a lot, especially with medium and lower Brightnesses and with already rather Saturated colors (one or two channels with a low contribution) like those in the second row of the Colorchecker. I'll leave the analysis of those for the readers, but if such colors are part of the scene, they will suffer even more than the more neutral colors.
I could continue, but I assume that the pattern is clear by now.
Now, the naysayers might argue that that exaggeration of Saturation with increased contrast is more pleasing, well I disagree. There is a reason why many others raise questions about the saturation changes, like the OP, or programmers who devise perceptually better methods that produce more natural/pleasing results.
Well arguing about taste is pretty useless, but maybe we can quickly look at how nature intended things to be.
Again, take the first patch, "Dark skin": RGB=[81, 68, 55], Lab=[38, 13, 14], HSB=[30°, 33%, 32%]
When we plug the RGB values into Bruce Lindbloom's excellent
CIE Color calculator (see attachment), and change the RGB model to ProPhoto RGB, then the Lab values come out at the exact same values (after rounding to integer numbers). So we've verified that Photoshop uses the same (correct) color coordinate calculations.
When we now change the luminance 'Y' value in xyY coordinates from 0.102501 to say 0.090082, we'll get about the same drop in the L channel as ACR produced ( Lab=[38, 13, 14] -> Lab=[36, 19, 23] ), L=38 becomes L=36. However, instead of boosting Saturation like ACR does, according to Bruce's calculator the Saturation should have been reduced a bit (in line with the reduced luminance),
the color should have become RGB=[75, 63, 51], Lab=[36, 12, 13] to look as nature intended it.
So we may conclude that ACR produces Saturation values that are not how natural Contrast changes would affect color/saturation. Some find that acceptable, others like me don't. Even if it were true, after research results would be published for peer review, that people prefer those unnatural color changes (just like some like overprocessed HDR images), it would be much better if that was left to the users to begin with.
Because first oversaturating, and then desaturating in a different way, will not be as accurate, nor will it be easy (due to how color vision works). We as humans are not that good at absolute color (due to preconditioning and changes due to ambient luminance levels), but better at color differences (almost simultaneous observation). When we are preconditioned with an oversaturated version, it becomes harder to restore to natural color. In Photoediting, it is usually best to start with a neutral basis (only adjusting exposure and contrast), and then alter colorbalance / local color / Saturation to better match a creative intent.
Cheers,
Bart