RetouchPRO

Go Back   RetouchPRO > Technique > Photo Restoration
Register Blogs FAQ Site Nav Search Today's Posts Mark Forums Read


Photo Restoration Repairing damaged photos

CIE LAB anyone?

Reply
 
Thread Tools
  #11  
Old 09-17-2005, 12:44 AM
Duv's Avatar
Duv Duv is offline
Senior Member
 
Join Date: Oct 2003
Location: Nanaimo, British Columbia
Posts: 1,213
OK. So I've read up to page 75 for the third time..had a little success with removing color cast. This image though I'm having trouble. No problem correcting in the other spaces but can the LAB pros post a correction including curves.

Thanks
Dave
Attached Images
File Type: jpg PinkGirlOrig.jpg (78.1 KB, 31 views)
Reply With Quote top
  #12  
Old 09-18-2005, 10:15 AM
edgework edgework is offline
Senior Member
 
Join Date: Jul 2004
Location: Brooklyn
Posts: 494
Quote:
Originally Posted by Duv
OK. So I've read up to page 75 for the third time..had a little success with removing color cast. This image though I'm having trouble. No problem correcting in the other spaces but can the LAB pros post a correction including curves.

Thanks
Dave
This one is tricky, but not impossible. Keep in mind that the lab techniques referenced, all deriving from Dan Margulis' approach, are based on his personal preference for avoiding at all costs local corrections until the very end of the process. An image this badly damaged probably is worth the trouble to just go ahead and cut masks and apply corrections locally.

However, lab does present some options. The three versions below show the three-step process that I took. The first image on the left has a curve applied to it that removes the cast from the highlights and seriously reduces it in the quartertones. As you can see, the rest of the image is still crap. The middle version has a second curve adjustment layer applied that removes the cast using the black of the phone as a reference. Using blending options, I restricted the effect of this curve to the shadows and faded it away through the mid-tones. The third version has a third adjustment layer added using a drastic color enhancement curve, pulling both anchor points in towards the center 40 units each on both a and b channels. Then, using blending options, I removed the effects of this curve from the highlights and shadows. That enabled me to go back to the b curve and favor the yellow end of the spectrum. I put the opacity of this layer at around 50%.

This is a first shot effort. The variables are the areas affected through blending options, the intensity of the third curve and the opacity of that layer. More experimentation might yield better results. There were no contrast moves in the lightness channel, and once the color comes out, other serious problems appear, such as the extra-greenish band on the right side. I would probably wait to go into CMYK and push the black channel to get some contrast, or perhaps make a copy, convert to CMYK and borrow the black channel as a layer mask in LAB or RGB to heighten shadows. But, like I said, this is just an example of lab curves, not an effort to produce a print-ready masterpiece.
Attached Images
File Type: jpg 3Steps.jpg (61.8 KB, 37 views)

Last edited by edgework; 09-19-2005 at 11:01 AM.
Reply With Quote top
  #13  
Old 09-19-2005, 03:42 PM
Duv's Avatar
Duv Duv is offline
Senior Member
 
Join Date: Oct 2003
Location: Nanaimo, British Columbia
Posts: 1,213
Here's what I've ended up so far in LAB. Fairly nauseating a curve. Not as good but perhaps it can be brought closer after converting back to CMYK and/or RGB for further manipulation. Lots of reading yet ahead. Thanks for your effort.

Cheers

Dave
Attached Images
File Type: png a channel.png (8.9 KB, 13 views)
File Type: png b channel.png (8.9 KB, 10 views)
File Type: png lightness.png (9.2 KB, 7 views)
File Type: jpg D-PinkGirl.jpg (93.0 KB, 21 views)

Last edited by Duv; 09-19-2005 at 03:56 PM.
Reply With Quote top
  #14  
Old 09-19-2005, 04:58 PM
edgework edgework is offline
Senior Member
 
Join Date: Jul 2004
Location: Brooklyn
Posts: 494
The problem with your curve for the a channel is that, while it certainly eliminates the red cast, pulling the entire bottom half of the curve above the center line turns ALL the red in the image to green. So now you have a green cast.
Reply With Quote top
  #15  
Old 09-19-2005, 06:05 PM
Duv's Avatar
Duv Duv is offline
Senior Member
 
Join Date: Oct 2003
Location: Nanaimo, British Columbia
Posts: 1,213
Ya, pretty nasty. Are you able to send your curves?

Dave
Reply With Quote top
  #16  
Old 09-20-2005, 08:16 AM
edgework edgework is offline
Senior Member
 
Join Date: Jul 2004
Location: Brooklyn
Posts: 494
The method for removing the color cast comes from the chapter on LAB color in Dan Margulis' earlier book, Professional Photoshop. He seems to have abandoned this technique for the more intuitive approach he takes in his latest work, but I find it's an incredibly simple and powerful way to kill whatever cast you have.

The white of her dress is reading a:9, b:9. The black phone (front panel) is reading a:54, b:17. Right off we see that shadows and highlights will need a separate treatment. While RGB and CMYK allow curves that address different tonal ranges, the lightness channel in lab allows you to target highlight, shadows or midtones specifically, using blending options. I used one curve for the highlights, then placed an adjustment layer on top targeting the shadows. The approach is identical for both curves. I've included the second set of curves, as well as the blending settings.

Make sure that the light/dark gradient on the curve window has darks set at the left. (This is contrary to the method Dan uses in his LAB book, but it is necessary to match the Info readouts). Since the highlights read 9 in both color channels, the curve will be the same. Click somewhere in the midrange, and in the input/output windows, plug in the values; 9 for inupt, 0 for output. Do this for both a and b curves.

Now the highlights read 0 0 and the phone reads 47 9. So the second curve uses those values to address the shadow cast. Those are included below. Then, double clicking to the right of the layer name brings up the blending options window. Setting the channel to lightness, the settings shown will begin fading out the effects of this curve at the first white arrow and totally exclude it by the second arrow. These settings are intuitive and to some extent a judgement call. I was concerned that the drastic moves did not affect the highlights that the first curve had already fixed. The quartertones had been helped by that curve but still needed further moves. The fade-out was set to address both those needs.

The third curve is a straight color enhancement curve, as described in the early chapters of Dan's book. For this, I reversed the light/dark gradient. Note that the blending options emphasize the midtones, leaving the highlights and shadows, previously adjusted, intact. Note also that the curves do not move through the center point. Ordinarily this would cause the highlights and shadows to take on a serious cast, but the exclusion of the blending options prevents this from happening. I favored the green in the a curve and the yellow in the blue. The curve moves are pretty extreme, and so I set the opacity of this layer to 36%.

These settings are all variable, and a matter of judgement. Any number of modifications could produce a better result, I'm sure. But the approach, the ability to address highlights, shadows and midtones separately, while not unique to LAB, is much more effective when coupled with the power of it's cast killing curves.

What is obvious, now that there is some definition to the colors, is that this is one seriously messed up image. It's going to be impossible to address the rest of the issues without making some local selections. In addition to the green band at the right, the rug along the left side is also disastrous. Contrast sucks and dozens of other glitches show up. These could be addressed in LAB as well, but for this stuff, LAB doesn't offer much of an advantage over RGB or CMYK and for the local color moves, might not be as good. LAB is a battering ram. From here on, something more subtle would be called for.
Attached Images
File Type: gif shadow curves.gif (14.2 KB, 9 views)
File Type: gif Shadow blend.gif (23.4 KB, 9 views)
File Type: gif Color.gif (13.9 KB, 9 views)
File Type: gif ColorBlend.gif (23.8 KB, 9 views)

Last edited by edgework; 09-20-2005 at 08:03 PM.
Reply With Quote top
Reply

  RetouchPRO > Technique > Photo Restoration


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
LAB Book mdavis Non-RetouchPRO Resources 14 08-08-2006 08:24 PM
LAB versus HSL bart_hickman Photo Restoration 5 03-12-2006 03:16 AM
Why the lack of LAB and other general ?s imann08 Website Feedback 10 01-30-2006 10:29 AM
Matte vs Glossy chiquitita Input/Output/Workflow 39 01-23-2006 12:27 PM
PC/Win filter for Lab mode: Minor ab Manip Stroker Software 15 09-25-2005 08:16 PM


All times are GMT -6. The time now is 01:24 PM.


Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
SEO by vBSEO ©2011, Crawlability, Inc.
Copyright © 2016 Doug Nelson. All Rights Reserved