Luminous Landscape Forum
Raw & Post Processing, Printing => Adobe Lightroom Q&A => Topic started by: bns on June 29, 2013, 03:50:09 am
-
A nice feature in LR5 is the possibility to 'measure' colors also in Lab values. For me that works more intuitively then RGB values while making color adjustments. It is a pitty however that while softproofing only RGB values are available. Or am I missing something?
Boudewijn Swanenburg
-
I agree that LAB color should be on the list for LR6. Dan Margulis's book on LAB color in Photoshop had some useful techniques that I have not been able to replicate in LR.
-
LAB color readout is in Lr5 - right click the histogram in Develop. I must say, I hope that's the end of it.
John
-
LAB color readout is in Lr5 - right click the histogram in Develop.....
Yes, but once one enables the softproofing mode, one is back to RGB values.
-
I agree that LAB color should be on the list for LR6. Dan Margulis's book on LAB color in Photoshop had some useful techniques that I have not been able to replicate in LR.
If you're talking about changing the working space to Lab and exposing Lab editing to the user, it would be a complete change in philosophy for Lr, which tries to make the actual, under the covers, working space something the user doesn't have to think about.
Or maybe you mean something else. When you talk about Dan's book, it makes me think you want to edit Lab values. I guess you could do that and keep the linear ProPhotoRGB working space, but the trend in Lr controls is towards more visual editing and more proprietary algorithms.
Jim
-
Yes, but once one enables the softproofing mode, one is back to RGB values.
Why on earth would you need Lab values in soft proof mode? You have it without when editing the master image which will go (where?). In Soft Proof you have on screen simulation AND actual output RGB values. Having Lab values isn't necessary or useful. Having the actual output values are.
To much of a love feast with Lab values, they can be useful to a degree, but not here.
-
Ever look at a "Paint By Numbers" painting?
Every damn one of them looks like crap!
Why on earth would you want to edit according to the numbers? Do you know that LAB is not a uniformly perceptual color model meaning on some colors you can be off by 5 points in either channels and hardly see a substantial change while others can look grossly off by just 1-2 points.
Do an online search on "Color Constancy" and see where else going by the numbers can screw things up visually.
You have a color managed preview that was brilliantly engineered into the software and hardware so we wouldn't have to rely on exact numbers.
I'ld suggest you rely on that since you paid good money for it.
-
I thought that by being able to compare Lab values both before and after the softproof preview would make it easier to distinguish between required adjustments in hue or saturation or lightness.
Might of course be wishful thinking on my part. Thanks for the advice to believe my eyes rather then my brain.
-
Might of course be wishful thinking on my part. Thanks for the advice to believe my eyes rather then my brain.
Don't feel bad, Lab has been way over sold. It was never designed for what the Lab proponents try to force it to do. Especially when they are told to convert into and out of Lab to 'fix' something they could do in RGB in the first place. It is a useful color model. It's real, real useful for defining color differences numerically. It is truly device independent but then RGB working space's are close enough and that's what is being used within LR. IF you are more comfortable viewing Lab values on the master image fine, but you can do that with an RGB working space values as well. When soft proofing, you are presumably done doing the heavy lifting and you now want to see a simulation of the output based on paper, ink, and contrast ratio and the effect of the rendering intent on the image in context. The numbers, RGB or Lab are not really useful, the idea of a soft proof is to allow you to see the output simulation before you make a print and if you wish, conduct very subtle corrections based on that view.
A good 90% if not all corrections from the god awful images illustrated in Dan's book, fixed using Lab could be fixed in RGB or better, just doing a better, less sloppy job of photography and raw processing (assuming any of those awful images were raw to begin with).
-
LAB color readout is in Lr5 - right click the histogram in Develop. I must say, I hope that's the end of it.
John
Reading LAB values is of little use to most people. Manipulating them can be very useful. That's what I would like to have.
-
I thought that by being able to compare Lab values both before and after the softproof preview would make it easier to distinguish between required adjustments in hue or saturation or lightness.
The computer needs the numbers because it's a dumb machine and needs to be told what is meant by color for every pixel on the screen. Its algorithms make adjustments to each pixel in relation to each other as a continuous tone image (photo) NOT INDIVIDUALLY which is what I meant by the "Paint By Numbers" analogy.
It maps each individual pixel so it mixes together to form a realistic image which involves complex color adjustments under the hood to trick your eyes into seeing the image as a whole and as intended. If you zero in on one color in the image you now change the perception relationship which influences your eyes to adapt. For example if you zero in on just the red patch of the Color Checker chart it will change color. It does for me.
That chart was meant to be used as a standard measure using Lab coordinates for a computer to understand and adjust the color tables to create a camera profile. Direct examination of the chart can have Lab numbers substantially off where a human won't see a difference. It's just a guide for our eyes but a road map for the computer.
Graphic designers who create swaths of single color graphics in their page layouts to be viewed individually utilize Lab numbers just like a paint mixer at Home Depot uses to mix house paint to match existing paint. It's understood that single color is to be viewed without any other continuous tone elements such as a photo to change the perception of that one color.
Are you printing a graphic that must match a Pantone number CMYK can't reproduce? Then you need the Lab number equivalent to get the closest match possible. Photos are a completely different animal.
-
And Dan's Lab edit processes take into account his "prepress color reproduction" experiences over the years of recognizing which image will work best using complicated Lab workflows. I cut my digital imaging teeth on his complicated channel mixing and Lab edit techniques back in '98 when I was teaching myself this new technology seeing there were no schools back then.
Yes you can get dazzling results using his processes but he's working from a mindset his students haven't been able to develop because as humans we first look...respond...change with edits as long as we know what that tool does to that certain part of an image.
The more complicated it becomes between "look and change" the slower the results. The slower the results, the faster the eyes adapt. The faster the eyes adapt the more the image has been changed to an outcome that hadn't had time to enter the mind at the outset because the user didn't understand what the tools were doing. That one process won't necessarily work on every image which will eventually make the user a slave at the computer because they've become enthralled with the process and not the results.
That's why folks fall in love with the process Dan puts forth but fail to properly see the image as others are seeing which often looks off.
You have to edit quicker than your eyes can adapt and change your perception of the image.
-
Reading LAB values is of little use to most people. Manipulating them can be very useful. That's what I would like to have.
Will never happen in ACR/LR. It's an RGB engine.
-
Will never happen in ACR/LR. It's an RGB engine.
Well, ACR 8.1 allows you to set Lab as the working space–Thomas even invented an Lab histogram...course, the processing adjustments aren't in Lab :~) But you can open a raw image directly into Photoshop as an Lab color space.
-
Well, ACR 8.1 allows you to set Lab as the working space–Thomas even invented an Lab histogram...course, the processing adjustments aren't in Lab :~) But you can open a raw image directly into Photoshop as an Lab color space.
Are you privy to the reasons why the engineers included Lab color space in ACR 8.1?
-
Are you privy to the reasons why the engineers included Lab color space in ACR 8.1?
Yes...Thomas decided that ACR should be able to process into any color space including Lab and CMYK and provide readouts and a soft proof of the output space–note ACR doesn't have a before/after capability (yet). Lightroom however, will still be limited to processing into RGB only color spaces.
-
Well, ACR 8.1 allows you to set Lab as the working space–Thomas even invented an Lab histogram...course, the processing adjustments aren't in Lab :~) But you can open a raw image directly into Photoshop as an Lab color space.
During testing, when I saw this, I converted a few hundred color patches using ACR and the same in Photoshop, ran em though ColorThink and the results were absolutely identical, dE 0 for every patch. So Jeff, do you suppose ACR hands this off to Photoshop to convert?
-
Will never happen in ACR/LR. It's an RGB engine.
Thank goodness for that! Modern output modalities are also mostly RGB driven, and Lab is not absolutely perceptually uniform and it introduces color shifts during conversion.
Cheers,
Bart
-
Are you privy to the reasons why the engineers included Lab color space in ACR 8.1?
they included it in ACR long time ago (using excursions to LAB for various purposes, wrote Eric Chan on May 24, 2011 in ColorSync mailing list = "ACR does use L*a*b* for some internal color difference estimates, e.g., for auto-calculated masks")... just ACR 8.1 now allows more than 4 output color spaces (including LAB).
-
So Jeff, do you suppose ACR hands this off to Photoshop to convert?
ACR had all necessary Lab math in it long ago
-
ACR had all necessary Lab math in it long ago
Except the previewed edits are driven by ACR's color engine, not Lab's. What goes on under the hood mathematically is for the code warriors to be concerned about, not photographers.
All I know is charcoal neutral or dead looking shadows, dreary cool and warm tertiary colors next to sappy saturated key fills are not a desirable look as I've seen in other Raw converters that use Lab as a color engine.
-
So Jeff, do you suppose ACR hands this off to Photoshop to convert?
Nope...but since Thomas wrote ACE I would expect both ACR and Photoshop to produce the same results with color transforms...in fact, I would be surprised if there were different results!
The thing I was hoping for (which hasn't happened "yet) is the ability to import DNG files directly into Illustrator and InDesign. You can do that now by setting the Workflow Options setting to CMYK and then open as a raw Smart Object into a TIFF or PSD CMYK document and then place that in InDesign (I've tested this and it works). You can go back to the Smart Object and modify the ACR settings and update them into InDesign. So, the ability to output in any color space is, intriguing...but bleeding edge. None of which is on the table for Lightroom...
-
Except the previewed edits are driven by ACR's color engine, not Lab's.
and what is "Lab's" color engine, if you don't mind ?
What goes on under the hood mathematically is for the code warriors to be concerned about, not photographers.
as I've seen in other Raw converters that use Lab as a color engine.
such as ?
-
and what is "Lab's" color engine, if you don't mind ?
What goes on under the hood mathematically is for the code warriors to be concerned about, not photographers.
such as ?
Here ya' go. A comparison demonstrating the two color engines on the same Raw PEF file below. I applied the same color temp settings and attempted to get each rendering to match as much as possible.
Note the lack of color (Charcoal neutral) in the shadows of of the leaves and the differences in actual dark colored leaves. I can use Adobe Standard color profile to get a similar effect but even still there's more depth because that profile still injects some kind of color even though it has a neutralizing effect throughout the color tables of the image.
This has a lot to do with psycho-optical effects similar to psycho-accoustics involved with human perception. This is typical color grading I see in a lot of Dan Margulies' final Lab processes posted in forums by those that process images in Lab. Those charcoal shadows tend to make inkjets render shadows with a lot of black ink which looks like crap next to vibrant colors on a print.
Just a note:
I prefer not to mention the Raw converter used for the Lab engine version because I respect the guy's efforts in creating a Raw converter with so many other features and preview renderings mainly having to do with sharpening and noise reduction appearance. He also keeps up religiously with updates and getting back with emails. I just like ACR's workflow simplicity and previews better.
-
I prefer not to mention the Raw converter used for the Lab engine version because I respect the guy's efforts in creating a Raw converter with so many other features and preview renderings mainly having to do with sharpening and noise reduction appearance. He also keeps up religiously with updates and getting back with emails. I just like ACR's workflow simplicity and previews better.
if you can't name a software then you shall probably refrain from using that as an argument and you shall refrain from generalizing on top of that... in addition what you comparison demonstrates exactly ? some unknown conversion (parameters wise) using unknown software ? there was one PhD here who compared iphone (?) jpg w/ results of the raw conversion from MFDB trying to prove some point...
-
if you can't name a software then you shall probably refrain from using that as an argument and you shall refrain from generalizing on top of that... in addition what you comparison demonstrates exactly ? some unknown conversion (parameters wise) using unknown software ? there was one PhD here who compared iphone (?) jpg w/ results of the raw conversion from MFDB trying to prove some point...
It's Iridient Developer and I was told by the developer he uses Lab as its color engine. This look is not confined to just this Raw converter but in Lab edits generally speaking as I indicated above.
Now, I'll go on to tell you why I'm right about editing in Lab.
The crux of the matter with Lab was mentioned several times here at LuLa about how well its saturation edits behave "better" than applying in an RGB space, the behavior of which stems from the fact Lab applies saturation by degrees according to how close a color is to absolute gray=(R=G=B/0=a*,0=b*). That's a problem with regards to the adaptive nature of human perception and how we view a scene in accordance with the actual definition of the function of a saturation boost.
What is happening when saturation is being increased?
It's not defined by science as "Make my picture look pretty" slider. Its function is to mimic the spectral reflectance characteristics of full spectrum light on any given object lit by it that we "FEEL" (from memory) is missing in the image. Shadows as defined by lab and a spectro technically exhibit less spectral reflectance and are closer to gray thus should get less saturation.
But we humans don't look just at shadows when we view an entire scene. I have looked just at shadows and some are quite neutral, some are bluish, some are greenish, etc. But they change color in relation to when I view the entire scene because the surrounding other spectral reflectance driven colors induce the adaptive effect.
Also saturation levels especially bumped up high also induce the adaptive effect into seeing less saturation scanning individual areas of an image. Lab does not calculate for this effect under the hood. It was built defining color one color at a time just like a machine (spectro) defines color which is how a machine can only understand color BY THE NUMBERS, just like a digital sensor defines color by measuring voltage variation of charged pixel cells to define gray luminance for each RGGB combination.
Lab only cares about the numbers when increasing saturation when the data is farthest away from gray not taking into account how a human sees the entire scene which is greatly influenced by adaption by both saturation levels and overall color in the scene.
Like I indicated with the Color Checker Chart when a human zeros in on just one color patch the adaptive effect kicks in and changes the color appearance by comparison to looking at the entire chart as a whole object viewing all the patches at once.
Shadows can be many colors some R=G=B and some not. Some may look neutral but read R>G>B and vice versa. Lab doesn't care what you see. If it's gray by the numbers it doesn't increase saturation. If it is colored it will increase saturation whether it needs it or not. See the example below and manipulation of the appearance of shadows in relation to its numbers.
Which one looks neutral? Which one will increase in saturation equally in relation with the rest of the image if applied either in Lab or in an RGB space?
-
The crux of the matter with Lab was mentioned several times here at LuLa about how well its saturation edits behave "better" than applying in an RGB space, the behavior of which stems from the fact Lab applies saturation by degrees according to how close a color is to absolute gray=(R=G=B/0=a*,0=b*). That's a problem with regards to the adaptive nature of human perception and how we view a scene in accordance with the actual definition of the function of a saturation boost.
Hi Tim,
Lab was never designed as an editing space, but more as a colorspace that allowed to judge neutral colors. For Editing, something like CIECAM 2002 is much better, and RawTherapee offers just that as one of the Tonemapping functions. It is very effective for believable saturation and contrast adjustments and it can be adjusted for different viewing (brightness) conditions.
Cheers,
Bart
-
It's Iridient Developer and I was told by the developer he uses Lab as its color engine.
you mix using a specific color space for postdemosaick postprocessing and how a specific software is implementing for example "saturation"... why do you think Adobe's profiles have hue twists ( http://dcptool.sourceforge.net/Hue%20Twists.html ) ?
-
For Editing, something like CIECAM 2002 is much better, and RawTherapee offers just that as one of the Tonemapping functions.
but it is not an internal working colorspace there - it is just a space used for a specific tool... it is unlike RPP using upLab internally post demosaick, while exposing Lab readouts in its UI and exporting image in Lab (as an option, in addition to RGB colorspaces or RGB w/o a "colorspace")... RPP is said to use Munsell inside for its next release.
-
you mix using a specific color space for postdemosaick postprocessing and how a specific software is implementing for example "saturation"... why do you think Adobe's profiles have hue twists ( http://dcptool.sourceforge.net/Hue%20Twists.html ) ?
Yeah, I already know about hue twists in Adobe profiles.
What's that got to do with what I said about editing in Lab? Or did you understand about that Gray=White Balance=doesn't get saturated?
Perceived neutrals in an RGB color space are detached from color temp appearance when applying saturation as it should be according to human perception which in nature there's really no such thing as R=G=B/a*=0,b*=0 as well as it having any connection to an illuminant. But Lab stays strictly by the numbers anyway.
We define neutral by only one illuminant=D50 in Lab. Do you know how many hues there are to white and gray in nature?
Don't know how to make it any clearer to you.
BTW when I brought up this subject to Brian Griffith, the developer of Iridient Developer, he agreed with me and said it was the reason he included RGB curves as a way to correct for it. ACR includes Split Tone but I only use that for color effects OR...wait for it...HUE TWISTS! OMG!
-
Yeah, I already know about hue twists in Adobe profiles.
so what do you think they need that ?
-
so what do you think they need that ?
I guess you'ld have to ask them.
But my guess based strictly on observing Golden Hour lit highlights that are orange-ish with a regular custom DNG profile but turn a dull reddish brown yellow hue switching to Adobe Standard may have something to do with clipping. At least that's what happens when I switch to Adobe Standard profile and I don't know if that one includes a hue twist.
This hue twist thing hasn't really affected any of my images or at least I don't notice it when it happens. If I get funky hue twists to highlights I just use Split Tone or the HSL hue slider or Color Temp slider. Geez! So many tools to fix one little thing. Why would the Adobe engineers do such a thing? It's scandalous I tell you!
-
so what do you think they need that ?
Doesn't present a problem for many of us. What's that got to do with Tim's comments on Lab?
-
Hi Tim,
Lab was never designed as an editing space, but more as a colorspace that allowed to judge neutral colors. For Editing, something like CIECAM 2002 is much better, and RawTherapee offers just that as one of the Tonemapping functions. It is very effective for believable saturation and contrast adjustments and it can be adjusted for different viewing (brightness) conditions.
Cheers,
Bart
Thanks for the suggestion, Bart, but I'm quite happy with the way things are working with my current ProPhotoRGB workflow with ACR and now Lightroom 4 when I get the chance to dive into learning that app and sort out its catalog/library messing around with my existing xmp edited files in Bridge.
Vlad,
Is this the hue twist you're referring to that I described about Adobe Standard profile?
-
Doesn't present a problem for many of us. What's that got to do with Tim's comments on Lab?
because he was effectively stating that great RGB engine does not need external crutches, his Lab issues apparently does not present a problem for many of others...
-
...his Lab issues apparently does not present a problem for many of others...
Problems with color perception are quite subjective, aren't they?
In art there's no accounting for taste as I saw first hand at portraits of pale complected children with rosy cheeks shot outdoors and processed in Lab space. The highlights of their skin looked normal but the modeled shadows transitioning to the side of their face away from the light looked like they were dusted or airbrushed in charcoal.
That's what I see in a lot of finished images in forums that discuss the wonders of processing in Lab space. Every darn one of the contributors said the Lab processed images look so much better than when they were processed in RGB.
That sounds like a religion to me. But again, there's no accounting for taste.
-
That's what I see in a lot of finished images in forums that discuss the wonders of processing in Lab space. Every darn one of the contributors said the Lab processed images look so much better than when they were processed in RGB.
That sounds like a religion to me. But again, there's no accounting for taste.
I don't say that, I say I saw many images that I wish were mine from LR, ACR, Iridient, SilkyPix, RPP, DPP, CaptureNX, you name it (even OOC JPG, when sufficiently downsized for visual consumption)... and I saw the same share of "bad" images from the same converters...
-
I don't say that, I say I saw many images that I wish were mine from LR, ACR, Iridient, SilkyPix, RPP, DPP, CaptureNX, you name it (even OOC JPG, when sufficiently downsized for visual consumption)... and I saw the same share of "bad" images from the same converters...
But how could you attribute the color space edited in as the influence on why you wished those images were yours without a comparison?
Each image is going to present some kind of non-normalcy from the outset or else why would they need editing. What I'm saying is edits in Lab to get the image to look normal always have that non-normalcy I just describe above.
And what non-normalcy could you spot that constitutes "bad" images. Any visual descriptions?
-
Do you know where I just love the way the Lab color model works?
When I eyeball calibrated my first HDtv.
After I neutralize and apply a warmer white balance to my B&W movies after first setting the Saturation slider to zero all the way so all channels are in B&W, I can increase the Saturation slider to a level that makes color content look great but doesn't increase the saturation of the warmer white balance of my B&W movies.
B&W movies don't exist in nature so they aren't part of the human color perception model, but it sure works for calibrating a machine. That's when I love the way Lab works...on a machine.