I have several hundred World War One era photographs I am working with that my grandfather took during his service with the Army Medical Corps. I have no negatives so I am working with nearly 100 year old B&W prints. I have scanned all of these but my scans have problems, one major is discussed here with the hope of getting some suggestions.

Many of these prints have undergone “silver creep” where the silver particles have migrated through the emulsion. Of course, these places have lost at least some of the detail that was there. When looking at the prints, these areas appear dark but if held at the right angle light will reflect off of them and they appear bright. When scanned these areas mostly appear white and are annoying to look and detract the eye away from what detail remains.

My goal for this post is to ask for ideas on how to darken these light areas so they are less distractive. I know there is no hope of recovering detail but if these light areas could be darkened I think it would greatly improve the view.

The example I hope to attach shows the problem quite well. Around the top and right side in particular you should see the problem. When the photo is zoomed in on these areas you can actually see what almost look like bubbles or clumps of silver.

Here is my current thinking as to one approach to darken these areas. A year or so ago I achieved partial success but my notes were one of the things that I could not recover after a hard disk crash. Also, my memory is shot so I am starting from scratch.

I am asking for help on what steps to take to achieve the following again.

1 – At one point I was able to get most of the bright spots isolated and on a separate layer. I had just the spots isolated but can’t recall how. I think I might have been messing with luminosity and I think had used an add-in but no clue now.

2 – My plan was to then turn the separated spots dark instead of light. I was then going to work with layers, put the spot layer underneath and adjust transparency of the upper layer to try to blend them together. Is this a reasonable idea?

I have a lot of learning to do so as much detail as possible on suggested processes will be appreciated. I will also have to learn how to handle transparency layers for step 2 but I think I can figure that out without too much trouble.

Again, any thoughts about this process overall will be greatly appreciated. Will my thoughts work? Is there a much better way? Any suggestions on add-ins to try, etc.

You can get a little better feel for the overall project and see some additional photographs at any of the below sites.

Thank you very much for any help!!!

Paul Hurm paulhurm@gmail.com

http://typicalfrenchkiddies.tumblr.com/

https://www.facebook.com/TypicalFrenchKiddies

http://www.typicalfrenchkiddies.com/ No photos on this site but the Introduction may be of interest.

...big snip....

Apart from anything else is there a reason for the sample image to be color indexed mode?

There is no particular reason for color vs gray or any particular mode. It is probably a remnant of something I've done along the way while playing with this project.

When I originally started I had read several places where the preferred scan process was to do true color at 600 dpi minimum even if the subject item was B/W. When I did the scans I just tried to follow those suggestions. Plus, the image I posted has probably been through a couple of iterations / conversions so I can't guarantee what state they are in as far as any number of gray scale points, etc.

Did that even answer your question? I'm very green with all of this.

To start from scratch on any of my original scans they would be true color and generally 600 dpi. If conversion to gray scale would be the first step, I would also appreciate suggestions on the best way to do that. Looking at GIMP tutorials I have run across several ways to go from color to B/W. Perhaps there is a best way for this project???

Thanks.

Paul

On Wednesday 04 January 2017 14:57:07 paulhurm wrote: I don't know as I would call it silver creep. What it is chemically, it
that the developer blacked silver, because of inadequate washing, has reverted to the silver.

I haven't personally seen enough of this to consider the obvious solution, which would be to develop it again, with a modern developer, Dektol if you can find it, perhaps even D-23 which ypu have to make yourself, then a weak acetic acid stop bath for perhaps a minute to neutralize the calcium carbonate accelerator in the developer, then refix to get rid of any residual silver, 3 or 5 minutes, then wash in running tap water for half an hour or more to remove the fixer, all at temps of 68 degrees or so. 75 works faster if you can muster up the warm
water fast enough. Then let air dry for at least an hour. I would not use a heated dryer for that because it will take on the gloss from its hard chromed surface. That will interfere with the scanning.

Yeah, I'm an old fart of 82 now, but I helped in the darkroom at my local
weekly fishwrap since about 1947, and had my owm chemical darkroom, B&W
and Color printing, for nearly 35 years starting in the late '60's. So I
know a wee bit about this. And while I have never done this, I'd sure take one you could waste and see if this proceedure would help mitigate
the effects of the aging. Locating the chemistry supply in 2017 will take a determined search.
Cheers, Gene Heskett

This discussion would probably be helpful in different circumstances but won't help with the ones I am faced with.

I really want help with the scans so I did not give other details that should show that any further processing of these prints in not possible.

I am actually working with photographs that are mounted into an album, not separate prints. This album is the "old style" with black paper pages onto which the photos were glued. Doing anything to re-wet would most likely destroy this album which should obviously be out of the question.

Another aspect that I did not mention in my original post is that each of these pages has hand written captions next to each photo. Again, any wetting would probably destroy these which is also not acceptable.

For others who may read this, again, I need help on how to work with the current scans I have. I am also not willing to do any re-scanning since some of the captions seem to be deteriorating and I don't want to handle the pages any more that is absolutely necessary.

Thanks.

Paul

On Wednesday 04 January 2017 18:58:20 paulhurm wrote: The black paper, and the glue, totally preclude any thoughts of its being
an archival quality mounting medium. There is some acid in all that stuff that old.

So you must do what can be done with the scans you have. You it appears, are truely between a rock, and a hard place. The only other possible procedure would involved crafting a very narrow band color filter, centered on the silvery tint, which might be further filtered to recover its luminance detail, and re-add that detail to the
K channel of a cmyk encoded copy of the original scan. I doubt that it could be scripted because of the variations in the aging of the individual print. Under the conditions you describe, you would probaby have to separate each print on the page into its own file for individual
processing. You would need to develop a filenameing convention that included the page it was on, the upper left corner position marker, and
the images own size, including any captions involved, so that once they
were recovered as best as can be done, a relatively simple script could
put it back together with everything in its original location for then making a single page image that could then be printed at high enough resolution that other researchers could learn from it.

I don't envy that position...

How about you make an "unsharp mask", crank up the contrast, then print
with the unsharp mask subtracted to reduce the contrast back to maybe 20% more than you started with? That might find usable detail you cannot see very well now.

Cheers, Gene Heskett

I do have a bit of this already done. I do have each photograph as individual scans, both with and without the captions and I am not too worried right now about the final "reassembly". My current desire is just to get the issue resolved with the quality of the prints. I already have ideas about "reassembly" but not worrying about this step until I get better quality images.

As mentioned, I had a few steps accomplished once before but can't remember what I had done. I still seem to recall having used an addin somehow and playing around with luminance so that comment by you helps validate some of my current thinking.

I think my next step might have to be to try to go through the various lists of addins to see if I can find one that might help or perhaps stumble upon what I used before.

I'm way too green to develop my own filter as you suggested above although the concept you described seems good. I kind of had that accomplished in the previous attempt where I did get the desired areas separated onto a separate layer. It's just that I now have no idea what I did to get to that point. Thus my post hoping for suggestions!!

Thanks.

Paul

I do have a bit of this already done. I do have each photograph as individual scans, both with and without the captions and I am not too worried right now about the final "reassembly". My current desire is just to get the issue resolved with the quality of the prints. I already have ideas about "reassembly" but not worrying about this step until I get better quality images.

As mentioned, I had a few steps accomplished once before but can't remember what I had done. I still seem to recall having used an addin somehow and playing around with luminance so that comment by you helps validate some of my current thinking.

I think my next step might have to be to try to go through the various lists of addins to see if I can find one that might help or perhaps stumble upon what I used before.

I'm way too green to develop my own filter as you suggested above although the concept you described seems good. I kind of had that accomplished in the previous attempt where I did get the desired areas separated onto a separate layer. It's just that I now have no idea what I did to get to that point. Thus my post hoping for suggestions!!

Thanks.

Paul

Gene,

And...

I just went back out and read a bit about RGB vs CMYK. I have never gotten far enough along with graphics programs to have had to worry about this but obviously now I do. As a former physics teacher I have been exposed to RGB much more than CMYK.

Just one more topic I have to go out to learn!!!

Paul

You are up against an intrinsically difficult problem; restoring the photos to their original condition is not possible, and making them "look like" they were restored to original condition would amount to using them as guides for creating new images from scratch. But making radical improvements is not especially difficult.

I did the following things with the image you provided:

1) Converted it from indexed to RGB mode. Indexed images do not work and play well with filters.

2) Saved it to XCF, created a duplicate as a new layer and named it "white balance."

Then I had a go at correcting for the overall fading, with the Levels tool (Colors > Levels, or add it to your toolbox buttons via Edit > Preferences > Toolbox). The histogram is cut off at both ends, which might be an artifact of destructive compression by the scanner or some program used to process the image.

I used the draggable markers at the bottom of the Input Levels display to set the black point to 50 and the white point to 235. This has the effect of "stretching" the histogram from true black to true white. Then I moved the mid-point marker to the right, and settled on 0.45, as
a "natural looking" adjustment.

3) Then I created a copy of the corrected layer, and tried a few items
from the G'MIC filter pack. Repair > Smooth [median], with a radius of
five pixels, knocked the sparkly silver clumps way down with minimal degradation of "wanted" image content.

4) Then I added a mask to the filtered layer, and painted on it in white to bring back some lost resolution in areas with important details
(faces, hands, etc.) and, switching to a smaller brush and black, painted on the mask to wipe out some bright specks that were restored along with the "wanted" details that were softened by the smoothing filter.

5) Finally, I created a new layer as a copy of the visible image, set its mode to Multiply, and dialed its opacity down to 33%. This brought
the contrast way up.

Result: http://pilobilus.net/xfer/p48_TEST_GRAY-EDIT.jpg

There is still room for a LOT of corrections, and someone may come up with fundamentally better ideas than the process outlined above. But it's a starting point of a sort, at least.

:o)

Steve

Definitely a good starting point but also reiterates to myself how much of a learning curve I'm on. At least, though, this does give some ideas to play with. I had already downloaded GMIC but got a bit overwhelmed when I saw all of the tools but had no idea what any of them do without testing. This gives me one suggested starting point though!

I may take some time and dig through the stuff I recovered after my HD crash with the hopes of finding some notes. I now do recall having at least looked at GMIC before but can't recall if that was a tool I used before or not.

I still want to go down the path of trying to separate the white spots onto a transparent background, turn them dark then try some kind of overlay. Don't know if this idea will work in the end but that thought still haunts me.

Now I "just" have to get a better feel for working with layers and transparency. I knew from the start that I had / have a steep learning curve but the idea of getting suggestions to narrow things down is why I posted here.

Definitely still open to other methods or thoughts that anyone might have.

Thanks.

Paul

PS If you read a previous reply of mine, you might recall that my very original scans are TrueColor and that the one I posted here has definitely been played with in other apps and is might not be my best starting point.

Would you suggest going back to my original scan and starting from there?

When I open my original scan in GIMP I get a message to the effect that the original contains an embedded color profile of "KODAK Srgb" and it asks if I want to convert the image to the RGB working space (sRGB built-it).

In general what do you think is the best path to take?

Should I then work with that "color" image before converting to BW at the end (or somewhere) or should I convert to BW (or grey) first???

Thanks.

Recurrent question, and it appears the answer is to not scan them, but make photos of them with a
decent camera/lens using a side lighting so that the dark areas do not reflect light into the lens.

Yes, I have had the take photographs idea mentioned and it may be eventually necessary but I am avoiding handling the pages any more than I have to. The handwritten captions appear to have been done with something like a crayon, certainly not a pencil or ink pen. They are starting to flake off in a few areas and I am afraid to handle the pages too much more.

I do not have the camera equipment to accomplish that task by myself. Interestingly the "Instructables" web site recently had an article about building a light box to use with the camera on a phone. I do not have a phone to use and my digital camera has recently died also. No bucks to build even a cheap light box either so I am trying to get going with the scans I already have.

Now, after other responses I received I have gone back and started to dig through GMIC. That utility is kind of overwhelming with all it contains and my unfamiliarity kept me from digging too deeply. Yesterday though I did go through a few of the sections that I thought might be most promising and I hit on something that, to my eye at least, helps the appearance. I plan to post be before and afters a bit later on for everyone to comment on.

Thanks.

Paul

On Wednesday 04 January 2017 21:02:15 paulhurm wrote: Chuckle... Make the connection with RGB in the video realm, where its all additive. So much red, so much green, and so much blue is white. Hence its RGB. In the video realm, its basically a matter of ajusting each colors gain once black is established, to get a pleasing white on the video screen. THis is RGB.

OTOH, The print realm is all subtractive, the Cyan reflecting green and
blue but absorbing red,, the Magenta reflecting red and blue but absorbing green, the yellow reflecting red and green while absorbing blue, and K is the combined luminance expressed as the density, hence the CMYK designation.

One, RGB references just barely visible as black, while CMYK references
the base color of the whiteish paper as its white, and inks are laid down to subtract the RGB from the white. And often, to save the expensive colored inks, the blackest areas will get a reduced amount of
color in favor of getting a more consistent black by useing black to build the density needed with far less ink.

Neither method can at the end of the day, represent the complete color spectrum the human eye (with good color vision) can see, but generally can be tweaked to do a decent job in the center 2/3rds of that color chart we've all seen in the past few decades.

Thats not all of it, but thats about as simplicated as I can make it. I
hope this helps.

Cheers, Gene Heskett

So much to learn and become familiar with. I did happen to go out and read up a bit on the RGB vs CMYK issue but the above helps too.

If you read my other recent reply, watch for a reply where I plan to post my recent before and after attempts.

Thanks.

Paul

OK, I'm replying to myself here and will attach my recent before and after work for comments.

I will post details of steps if anyone is interested but basically I relied on GMIC tools to achieve what I have here. Luckily, if I decide that this process might be worth pursuing with the rest of the photos, it is a fairly easy set of steps.

To my eye at least, the after shot does a pretty good job of toning down the white areas that I complained about. You can still tell that the image is degraded but at least they are not so bright and distracting to the eye. If you look at the middle guy's chest area you will see that my result does not solve the problem of lost detail and I don't expect there to be any way to regain what has been lost. That is just what I have to accept.

I just wish the negatives had survived! My 93 year old aunt did not even know this album existed when I showed it to her for the first time so it apparently is something that my grandfather put away and never displayed. She has no idea what ever might have happened to the negatives so I don't ever expect to find them at this point. OTOH, I did actually find 2 negatives mixed in with other more recent photos my grandfather took which have survived, so I think the disappearance happened later rather than that they never made it home from France.

I'm still definitely willing to listen to any other advise anyone has but please take a look at my two attachments "Original" being the original scan and "Method 1" being my GMIC work. Comments appreciated!!!

Paul

Here are some notes for someone's possible future use about what I did here.

My original scans are True Color and usually 600 dpi. When I open them with GIMP I am first prompted that they have an "embedded color profile KODAK Srgb" and I am asked if I want to convert to "RGB working space". I select "Convert".

For the example scan in this post I did the following:

Image Mode
Grayscale

Filters
G'MIC
Details
Dynamic range increase
Recover highlights set to 1.0
OK

Filters
G'MIC
Repair
Despeckle
Tolerance 15
Max area 10
OK

I have tested a few of my other scans and as would be expected, each of my photo scans takes a slightly differect process from the above and with different settings.

I have found that many of my other scans will not need to be Despeckled. If I do play with Despeckle I have to be very careful of the settings. I have found that in some cases I can lose details that I don't want to such as bright reflections off of posts, faces etc., things that I really want to keep. I have found that setting the Tolerance rather high will help with catching many "true" speckles without losing too many wanted reflections.

Again, the above is meant to be a reference to what I did with this example for someone's possible future use. This was a rather extreme example that I needed suggestions with. As would be expected, each photo will take some differences in process and settings. Bottom line though I have found that the G'MIC - Details utility seems to be a good starting point.

I hope this helps others in the future.

Paul