Pages: [1]   Go Down

Author Topic: LAB Color -Dan Margulis  (Read 1657 times)

rmazzi

  • Newbie
  • *
  • Offline Offline
  • Posts: 37
LAB Color -Dan Margulis
« on: February 25, 2017, 05:22:12 PM »

My intellectual curiosity drew me to Photoshop LAB Color written by Dan Margulis.  The author warns that the topic is difficult.  I was able to understand the first chapter in concept and technique.  It was very useful and i have used it to improve a couple of images that have vexed me.  The exercise in chapter 2 has me stumped.   I am lost as i went deeper in the book and i was curious if there are others that have found the techniques useful.  Are there other resources that can help with me to understand some of the additional concepts so that i can apply them on other types of images?
Logged

LesPalenik

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1238
    • advantica blog
Re: LAB Color -Dan Margulis
« Reply #1 on: February 25, 2017, 05:45:40 PM »

I've been using a quick image conversion to LAB with slightly steeped curves and back to RGB for years, mainly to make the colors pop, but sometimes even changing some of the colors. It works for me. In most cases, this is a better technique than increasing the saturation of the colors.

Somewhat similar pop effect can be achieved also by using Topaz Clarity plugin. However, it is a different method and one which doesn't allow adjusting the curves.

TOPAZ CLARITY

scyth

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 653
Re: LAB Color -Dan Margulis
« Reply #2 on: February 25, 2017, 05:51:16 PM »

I've been using a quick image conversion to LAB with slightly steeped curves and back to RGB for years, mainly to make the colors pop, but sometimes even changing some of the colors. It works for me. In most cases, this is a better technique than increasing the saturation of the colors.

Somewhat similar pop effect can be achieved also by using Topaz Clarity plugin. However, it is a different method and one which doesn't allow adjusting the curves.

TOPAZ CLARITY

there is a curvemeister plug-in ( curvemeister.com )
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2120
Re: LAB Color -Dan Margulis
« Reply #3 on: February 25, 2017, 06:34:34 PM »

I am lost as i went deeper in the book and i was curious if there are others that have found the techniques useful. 

On scans of negatives of deep, almost cobalt blue pansies I found working in Lab space using curves and Hue/Sat tool in Photoshop very useful in forcing out tertiary hues (purples & violets) that make up hidden color detail in the shadows of the blue pansies while warming white balance of colors that are suppose to be warm at the same time adding cool cyan tints to fresh green leaves.

I don't find working in Lab more useful editing Raw files from digital cameras. Adobe Camera Raw I find the best and most efficient tool for doing the same type of color constancy corrections with color temp manipulations.

I never got into the channel masking and mixing I first saw in one of Dan's articles in a long forgotten digital imaging technology magazine back in '98. It was too involved with not that much pay off. But that was to correct for bad film scans which Lab is ideal for, but it does require a good eye at recognizing when color constancy looks off which is sometimes hard to catch due to long edits effect on the eye's ability to quickly adapt to changing color temps.

Just a reminder that Lab space was not designed as an RGB editing space which of course is the display you're editing through. There are non-uniformity issues in Lab in defining certain colors on color managed RGB display previews from Lab edit algorithms that can create bad artifacts and tonal transitions. Working in 16bit is a must.
« Last Edit: February 25, 2017, 06:44:46 PM by Tim Lookingbill »
Logged

CeeVee

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 70
Re: LAB Color -Dan Margulis
« Reply #4 on: February 25, 2017, 06:38:55 PM »

I've found that LAB is useful for a number of thing, but not a panacea. B&W conversions are one. But working on color is a real PITA.

Sent from my SAMSUNG-SM-N900A using Tapatalk

Logged

TonyW

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 375
Re: LAB Color -Dan Margulis
« Reply #5 on: February 26, 2017, 10:28:32 AM »

My intellectual curiosity drew me to Photoshop LAB Color written by Dan Margulis.  The author warns that the topic is difficult.  I was able to understand the first chapter in concept and technique.  It was very useful and i have used it to improve a couple of images that have vexed me.  The exercise in chapter 2 has me stumped.   I am lost as i went deeper in the book and i was curious if there are others that have found the techniques useful.  Are there other resources that can help with me to understand some of the additional concepts so that i can apply them on other types of images?
Lab can be very useful on occasion and perhaps accomplish some things which would be more difficult or time consuming in any RGB space (at least for me). 

On the other hand I have a feeling that sometimes its benefits have been oversold, nevertheless the areas I find most useful (agreeing with others that have posted before me)

1.  Making colours pop without introducing artefacts that may be apparent using other methods.  So called Picture Postcard workflow.  Useful for accentuating very subtle changes.

2.  Potential to repair and replace a damage channel (often blue) in an RGB file and reduce noise in the process.

3.  A nearly penalty free method to reduce/eliminate colour noise

4.  Making gross changes to colour e.g. Red car to green blue or any other shade.  Usually accomplished with little in the way of masking with just the use of Blend IF

5.  Finding a starting point mask that may be better than RGB or CMY

Generally I would forget about it (most of the time) for 8 bit images as the risk of introducing nasties is increased you losing around 30 levels from the 256 you have available.  Working in original 16 bit data the losses are likely to be any real concern.  I am sure there are exceptions in both cases

Quick example where it may be more useful than other modes.  Client gives you a T shirt to shoot or T shirt image.  Bad news he only has a green one available but has a range of colours he needs to show in an ad. and of course it is needed yesterday.  He is able to give you however a precise colour perhaps Pantone or colour swatch etc.

Some like to use a Gradient map in PS.  I invariably find that I can get a better looking result in Lab in a much shorter time than using Gradients.  Below is a quick play changing the colours of a green T shirt image.  No masking needed in this case for skin tones and jeans just used Blend If sliders

Click on the Thumbnail.  Image should change after a second or two



« Last Edit: February 26, 2017, 10:39:14 AM by TonyW »
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2120
Re: LAB Color -Dan Margulis
« Reply #6 on: February 26, 2017, 01:44:18 PM »

Generally I would forget about it (most of the time) for 8 bit images as the risk of introducing nasties is increased you losing around 30 levels from the 256 you have available.  Working in original 16 bit data the losses are likely to be any real concern.  I am sure there are exceptions in both cases

Just to clarify that the use of editing in 16bit is to provide more levels of bits that aren't in the original file similar to providing more handles or nodes for remapping tone and color for smoother behaving edits in previews. Since video cards preview in an 8bit pipeline the 16bit (extra handles) precision provides smoothness in the EDITS to prevent posterization/banding in wide swaths of gradations such as deep blue skies in the preview as well as the file itself.

Technically with DSLR Raw files there is no original 16bits, only the 12 to 14bit precision filtering that was used in the camera's Analog to Digital converter that sifted out more detail with less noise especially in shadows and in gradients. The preview in the Raw converter is in 8bits but as in the case of Adobe Camera Raw the 1' & 0's/or demosaicing stage (can't recall which right now) describing the pixels has been upconverted to 16bits (must be set in ACR by clicking on stated resolution below image frame) so the editing tools have more "handles" on 256 levels for each RGB in order to squeeze and stretch a huge amount of dynamic range and color gamut for smooth behaving 8bit previews.

That precise editing provided by the extra bits and resulting in the final preview are locked in once you save/convert the Raw file to tiff or jpeg.
Logged

leuallen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 333
Re: LAB Color -Dan Margulis
« Reply #7 on: February 26, 2017, 01:47:24 PM »

You might want to check out Lee Varis 10 channel workflow. It is based on Margulls lab work. He has some videos that explain things clearly. He also has a comprehensive tutorial which is not too expensive if you want to become more involved.
Logged

TonyW

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 375
Re: LAB Color -Dan Margulis
« Reply #8 on: February 26, 2017, 02:55:07 PM »

Just to clarify that the use of editing in 16bit is to provide more levels of bits that aren't in the original file similar to providing more handles or nodes for remapping tone and color for smoother behaving edits in previews. Since video cards preview in an 8bit pipeline the 16bit (extra handles) precision provides smoothness in the EDITS to prevent posterization/banding in wide swaths of gradations such as deep blue skies in the preview as well as the file itself.
Agreed to a degree except that if the original is 8 bit then converting this to 16 bit is pretty much a voodoo move which adds nothing.  Do not forget that some video cards e.g nVidia Quadro  and AMD FirePro are actually 10 bit and without using such it is entirely possible to observe posterisation/banding that is not actually part of the original data and therefore not seen in a print

Quote
Technically with DSLR Raw files there is no original 16bits, only the 12 to 14bit precision filtering that was used in the camera's Analog to Digital converter that sifted out more detail with less noise especially in shadows and in gradients. The preview in the Raw converter is in 8bits but as in the case of Adobe Camera Raw the 1' & 0's/or demosaicing stage (can't recall which right now) describing the pixels has been upconverted to 16bits (must be set in ACR by clicking on stated resolution below image frame) so the editing tools have more "handles" on 256 levels for each RGB in order to squeeze and stretch a huge amount of dynamic range and color gamut for smooth behaving 8bit previews.  That precise editing provided by the extra bits and resulting in the final preview are locked in once you save/convert the Raw file to tiff or jpeg.
Again do not forget some cameras output original 16 bit (Hasselblad?), but PS only deals in 15 bit precision i.e. 32,768 levels vs 16 bit @ 65,536 levels.  I am not aware of any evidence supporting the case that 16 bit would yield a significant increase in IQ although it might be argued that there may be a small increase in editing headroom before you break the image?

Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2120
Re: LAB Color -Dan Margulis
« Reply #9 on: February 26, 2017, 04:27:46 PM »

Quote
Agreed to a degree except that if the original is 8 bit then converting this to 16 bit is pretty much a voodoo move which adds nothing.

It adds precision points, levels, nodes, instruction paths (however it's defined within the software at the A/D converter or Raw converter) that more smoothly maps the edits to an 8 bit preview.

I agree adding bits by upconverting from 8 to 16bits in ACR does not add extra image information. It adds additional instructional paths for more precise drawing of the edit on the 8bit preview. An image doesn't exist unless it has a preview. If it needs a lot of editing then there needs to be more refined points of instruction to make the preview not show artifacts like banding in gradients. Bad preview, bad image. No preview, no image.

The 16bit of the Hasselblad means the A/D converter used much more precision at converting the sensor's voltage readings off each sensor cell. It's just filtered numbers representing various levels of luminance for each RGB pixel site. The A/D converter is doing what we do in 16bit edits in ACR but it's happening farther up the process pipeline before ACR gets hold of it. Bits are about precision in determining what is meaningful detail captured vs garbage noise in what quality of data goes to the Raw converter or gets converted to tiff or jpeg.
« Last Edit: February 26, 2017, 04:31:58 PM by Tim Lookingbill »
Logged

rmazzi

  • Newbie
  • *
  • Offline Offline
  • Posts: 37
Re: LAB Color -Dan Margulis
« Reply #10 on: February 26, 2017, 08:08:13 PM »

Thank you all for your comments.  The T Shirt demo was quite interesting and I did not realize that is a real issue for commercial and advertising photographers at times.  The comments suggest that advances in other tools and utilities may be a more convenient substitute for some of the LAB techniques.  As a hobbyist, i can chose to avoid some of the more complex challenges for now.  I have found real value in the technique that solved the canyon conundrum, which is not to difficult to apply.
Logged

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384
Re: LAB Color -Dan Margulis
« Reply #11 on: March 01, 2017, 11:31:38 AM »

Thank you all for your comments.  The T Shirt demo was quite interesting and I did not realize that is a real issue for commercial and advertising photographers at times.

A while back, maybe 15 years, I toured a studio that did work for Target. They claimed it was sometimes much faster to stage and shoot one item against white, and use similar techniques to do all of the other color variations than it was to setup and shoot each variation.
Logged
Pages: [1]   Go Up