Restoring the Cecil Stoner photos

ruttrutt Registered Users Posts: 6,511 Major grins
edited October 23, 2005 in Finishing School
Hey everyone, now that we've done what we could with these in the time we had, I'd like to take a little time so that we can all learn from what we did. I know this was a different kind of PS work for me than what I usually do, but I have the feeling that at least some of the people who worked on this really know what they are doing (Bear, what's your day job?)

So I'd very much like to learn what went into each retouch. It would also be great to share what you learned along the way. Often knowing what doesn't work is just as important as knowing what does. I know that I did better later on.

I'll post my before and afters and the steps that went into them as soon as I can get to it. Everyone else, please take the time to do this. I'm sure it will be interesting.
If not now, when?

Comments

  • behr655behr655 Registered Users Posts: 552 Major grins
    edited October 20, 2005
    rutt wrote:
    (Bear, what's your day job?)
    Design/sell/install custom kitchens. Our top line is Rutt by the way.

    Anyways, on to photo retouching.
    Long before I had any kind of camera worth talking about I had a scanner. Several years ago my wife started working on her genealogy. She gathered all the old family photos that she could from her family and relatives. Just like Lindsay's photos they were faded and many were torn and tattered. Many of the photos were from the late 1800's and early 1900's. With my primitive scanner and a trial version of Paint Shop Pro 3, I started hacking away at them. To my surprise I was able to do a fairly decent job of touching them up. Fixing tears, re-constructing body parts, removing dust and blemishes, etc.
    I really wish I could give specific detailed steps that I use to restore photos but it really is just trial and error. Many many years ago I read the comic strip B.C. (still do). In this particular strip a caveman asked one of the other cavemen how he was able to sculpt a likeness of Peter. His reply was "Simple, I just chip away anything that does not look like Peter". That's how I approach photo re-touching and a lot of other things for that matter. Funny how a comic strip can give you a life lesson.
    The military photo I worked on for Lindsay was approached in this same way.
    Here is the original photo:

    scan0009.jpg

    The first thing that is obvious is that the photo is faded. PSP X (and starting with 7 I think) has an easy button called "Fade Reduction". Sorry I don't know how this works other than very well. The first thing I do is use Fade Reduction where I have a slider from 1% to 100%, In most cases anywhere from 1% to 20% is enough. If you were to look at the full size file you would see that it was loaded with dust and blemishes. Although this was not a digital photo I then hit it with "Digital Camera Noise Removal". I have found that this helps remove the patterns that are often associated with scans. It also helps in removing some dust, etc. Then onto the dust and blemished that are left behind. I never use "Dust Removal" as I find it degrades the photo too much. I use the "Clone Brush" which is a tedious way of doing it but the results are best. I usually zoom in to 100% to 200% while doing this and start dust busting. I zoom out every so often to see the results and make sure I'm not goofing thing up. Next I noticed that his ascot?.... scarf?.... thing around his neck, had some sort of pattern to it that was possibly the result of scanning or some of my other alterations. To correct this I selected "it" and used "Edge Preserving Smooth". This function basically blends the pixels to give it a smother appearance.
    Then to the really hard part IMHO. Getting the white balance, color balance, Highlight/mid-tone/shadow, sharpness, etc. While dust is dust and it has to go, these other things are subjective. What is the right color? How sharp should it be? I can make it look good to my eye but what about others? I really do have the most trouble with this part of re-touching. Here is the result:

    40710379-M.jpg

    So it really is trial and error and patience, patience, patience. On the three photos I worked on I spent a total of 5 1/2 hours and I'll let you know that in this case it was a labor of love. Not much more than a year ago I was doing this with photos of my father who passed away. I know how much it means to preserve these memories.

    So to sum it up. Read comics.

    Bear
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 21, 2005
    I gathered up the before & after shots here: http://dgrin.smugmug.com/gallery/899573

    Take a look; it's pretty interesting to see what we did.
    If not now, when?
  • spocklingspockling Registered Users Posts: 369 Major grins
    edited October 21, 2005
    Here’s my attempt at explaining the step I took with these photos. This is just off memory as I didn’t write anything down and I’m not really that good at this kinda stuff.


    40853468-S.jpg

    With this picture, I first straightened the horizon, then cloned a few areas to get rid of the worst of the damage. I used a LAB curve on this one. The toughest part of this picture was it’s toning. It’s not a B/W nor is it a color picture. I used the black from the sunglasses as a starting point. Then just got it to look what I thought a sepia toned picture should look like. I used a very small amount of the dust and scratch filter (too much and everything goes soft), then I added some USM to L channel. Finally I used Dan Margolis method of sharpening using the halos method from this thread.

    40894843-S.jpg

    This same method was used for this picture, and this time it was a B/W. So it was a little easier to get the black point for the curve layer. But I had to guess on the highlight. Again, done to look right to me.

    40853453-S.jpg

    40894830-S.jpg

    This color picture was almost done the same way except that I added a CMYK curves layer after the initial LAB layer. This allowed me to fine tune the highlight and shadow points. But, as Dan explains in his book, you have to look for those points which you “know” are true colors. I used the Coca-Cola truck. I knew the tires would be black and the side of the truck would be white. Those were good starting points. The trick obviously is the colors have to look believable.

    40853454-S.jpg

    My version was 40656969-S.jpg


    I know Rutt has redone this one. He’s much better than I am. :^)
  • DavidTODavidTO Registered Users, Retired Mod Posts: 19,160 Major grins
    edited October 21, 2005
    spockling wrote:
    The toughest part of this picture was it’s toning. It’s not a B/W nor is it a color picture. I used the black from the sunglasses as a starting point. Then just got it to look what I thought a sepia toned picture should look like.

    Interesting. I would have probably just made it a grayscale image, made the repairs and then added sepia at the end to make it look like the original sepia.
    Moderator Emeritus
    Dgrin FAQ | Me | Workshops
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 21, 2005
    The last one I did (except for my tweek of Bear's) was this one:

    Before:

    40853454-M.jpg

    After:

    40894838-M.jpg

    By this time, I had learned a few tricks. So I suppose it was my best effort.

    Here's what I did:
    1. Nearly doubled the resolution from 1059px x 689px to 2499px x 1625px using the trick of increasing by 10% over and over. If nothing else, it was a lot easier to work on once I did this.
    2. L channel Dust & Scratch filter, L channel, on a layer. Then I used a layer mask to reduce the opacity of this (sometimes to 0) where sharpness matters, in particular on Cecil's face. Flatten.
    3. Neat Image to reduce noise. Again on a layer. And again I used a layer mask afterward to limit the damage to the places that matter most. In this particular image, the scarches and noise were most offensive in the black background which obviously doesn't have to be sharp. Flatten.
    4. A little work with the clone tool reduce the flash reflection in his glasses above his right eye.
    5. Copied the green channel into a luminosity blended layer to recover some detail and depth in his face. Again, this wasn't so great for a lot of the image. Made a duplicate layer of the background on top, added a layer mask to it and revealed just the face. Flatten.
    6. LAB curves. I optimized color and contrast for the face. Some of the toys ended up with pretty nice color after this, but the shadows ended up with a pretty bad purple cast. And the whites were pretty yellow.
    7. Selective color - white to neutralize the highlights.
    8. CMYK curves to steepen the K curve and try to get better shadows.
    9. Selective color / black to cut back on the CMY inks and reveal the shadow details plugged by the previous step.
    10. L channel USM, HIgh RAdius, LOw AMount. This image wasn't sharp enough for conventional sharpening to work; it only increased noise. But 24/45/9 really did seem help the illusion of sharpness.
    11. Voila

    I struggled to find a crop to get it down to 10x8 aspect ratio, but I didn't know where to crop. Any one of those toys might be a precious memory. So I left this for the family to decide.
    If not now, when?
  • ginger_55ginger_55 Registered Users Posts: 8,416 Major grins
    edited October 21, 2005
    Really cool, the whole gallery is. I just looked at it.

    On the panning for gold shot, I think it is a tad over to the "dark" side. In particular I would love to be able to discern facial features again.

    I like that shot in particular. By the improvements, I think it either got too contrasty or too K curved. Either way, on my monitor I can no longer see his facial features.

    You guys are the LAB angels for doing this, smile!

    ginger
    After all is said and done, it is the sweet tea.
  • mereimagemereimage Registered Users Posts: 448 Major grins
    edited October 23, 2005
    Rutt, I will try to explain my workflow
    Unfortunately I am a bit eclectic in my photoshop work and I don't always remember all the steps I take. With this photo:

    40763369-L.jpg


    I started by duplicating the background layer. I switched to Lab mode and did gross tonality adj. by steepening the L curve and then adding an S curve bulging slightly in the midtones. I then steepened the a and b curves a bit, one slightly more than the other and adjusted the color cast by shifting the center pts of the curves. This was a good starting pt but I wasn't happy with it because the shirt was still sl yellow. I tried using lab-locking center and adj only the highlights, adding assym curves but I couldn't seem to get it where I wanted it(my limitations). I changed back to RGB and used the color adj sliders for a little fine tuning. Finally I brought up levels and used the dropper to pick my white pt-I didn't like the overall result but I looked at how it changed the histogram and the made subtler but similar changes and fine tuning till it suited my eye using the individual color sliders in levels-kind of a backassward workflow. Sharpened with Noise Ninja, used clone tool and healing brush to remove dust and scratches. I then burned in the edges by making a marque selection around the gentleman, inverted it, feathered the edges ~100 pixels, copied the selction as a layer and set it to multiply-adj by eye with the slider, and then some use of the burn tool.

    This photo I used Greg Gorman's duo-tone technique(mostly to practice), adding some curves to adjust tonality, sharpening and adding noise:

    40763317-L.jpg

    This last photo the work flow was much like the 1st. I couldn't adjust the color through the pictue without leaving a yellow cast along the shadows of his skin. Here I made a selection using the color selection wand,feathered it and adjusted the skin tone with color balance sliders and fine tuning with the saturation controls:

    40763264-L.jpg

    ...........................Mereimage
Sign In or Register to comment.