Severe mould damage on slides

[Caitlin]

This is one of 100s of slides I discovered belonging to my parents - that had been stored in a dank cupboard along with the potatoes and onions for the last 20+ years... (No - I'm really not kidding!) Not too surprisingly they are in a dreadful state.

I have only just started to look at them - I have been cleaning them with PEC-12 photographic cleaner, generally with good results (though I found once particularly batch where the pigment had become soluable!! Ekk!)

Anyway, I've been suprised considering their dreadful state how well many are coming out. The attached is one of the worst cases though (excluding those that have just completely faded away) I've been using the trick of replacing the blue channel with the green channel for some of the ones with particularly bad mould (mold) damage, but this one seems to be equally damaged on all channels. I'm pretty sure this one is beyond rescuing, and all I've done is scan it so far (on auto settings) but I thought I'd post it up here in case anyone had some clever tricks I'm not aware of.

I've also attached a photo of one of the boxes of slides, to give you an idea of the state they were in!

It's quite interesting (though a bit tragic), to see how differently different slides have survived that were stored in the same conditions. Down to slides processed by the same company in the same box, but from a different batch (possibly different stock?) being in almost pristine condition, or coated in mould and completely faded away. Similarly while most that have survived are seriously faded, a small percentage are still stunningly vibrant.

BTW - can anyone explain the science behind the technique of replacing one colour channel with another? (I notice you can't use the red channel though...) I know it works, but don't understand why!

It's not a disaster if this photo can't be saved, so if it's a lost cause just let me know ok?

[CJ Swartz]

Caitlin,

This sure isn't much of an improvement, but see if you think it is a bit better -- it's a Calculation of the Red channel in a RGB copy of the image with the Cyan channel of a CMYK copy set to Soft Light.

If you think it's an improvement, here's what I did (you can try other settings to see if there's a better fit):

dupe the image and change the duped image to CMYK format,
select your RGB original and select the Red Channel,
then use Image --> Calculations to bring up the diaglogue box,
the settings I used were Source 1: RGB image, Layer: merged, channel:Red
Source 2: CMYK copy image, Layer:background, channel:Cyan
Blending: Soft Light
Opacity: 100%

There might be a variation that would be a bit better, but the few I tried were worse. Overlay setting was okay except for losing the highlights.

You can copy the resulting alpha channel and add it as a layer to your image, and add color from you original.

[Gary Richardson]

Caitlin,

There's no real reason why you can't replace any channel with any other channel, including red.

When you look at the channels you see a greyscale representation of your image. What you're actually seeing is the luminance image for that channel, ie the intensity of the red in the red channel, the blue in the blue channel, and the green in the green channel.

When you replace channels, what you are doing is taking the luminance of one channel and replacing it with the luminance of the other. That is why you get colour shifts on your original image when you've replaced a channel.

Although I said you can replace any channel with any other, in practice the luminance levels in the red channel are often so different to those in the others, that severe colur shifts occur. But there may be circumstances where this is not the case, so always look at all channels.

As a last tip, when replacing channels, I usually copy the channel I'm using to replace the damaged one to an Alpha Channel, then use Levels and curves to get it as close to the channel I'm going to replace, before I paste it into the channel. That way there are fewer colour issues to resolve.

Hope this rather incomplete description clarifies things a little.

PS. Good luck with your slides, had a look at the one you posted, and decided I had little enough hair as it is, so didn't try "improving" it.

[Flora]

Hi Caitlin,

CJ gave you already a great tip on how tweak with the Channels ... she used Calculations which gives you much more control and options on the procedure, but you could also use the 'simpler' Apply Image.

Can't add anything to Gary's excellent explanation about Channels.

That said, I've completely replaced Channels very, very rarely... usually when the damaged channel was practically non existent ...either a solid black or a solid white block ... and in those occasions, I followed the procedure described by Gary to minimize the colour shift.

The reasons behind the procedure of replacing/tweaking Channels are several: from correcting a colour cast to minimize chroma noise/blotches or marks or writings on the picture, to retrieve details to work on ... etc.

What I usually do first is checking if there's still something I can use in the damage channel (details missing in the other channels) in which case I try Apply Image/Calculations, or the great Tool which is Channel Mixer ... All these procedures let you control the interaction among Channels in different ways until you get, in the worst cases, to 'something acceptable' you can work on.

For your slide, I used Channel Mixer and Hue&Saturation to get to an acceptable basis ... after which it's up to you to decide how much or how little you wish to proceed with your restoration.

The result I got is a far cry from good .... but, at least, it's a 'recognizable' image ....

Let me know if you are interested in the values ....

[Caitlin]

Flora, that restoration is a 100 times better than anything I thought was possible! I have a number of other slides that I probably care more about, but that aren't quite this badly damaged, so if you can acheive this with this image it gives me hope the others can be saved.

Thanks to everyone for their instructions. We are definately heading into unknown territory for me, so I think I will need to put some time aside over the weekend and ponder what you have all said and try to really get my head around it!

[Gary Richardson]

Flora, terrific result with a very difficult picture. Despite what I said earlier, I did have a play with Caitlins image, and I now have less hair than before, (will I ever learn) and all for no usable result. I'll now have to have another go using the your technique (there goes the rest of my locks).

[happyheart]

Originally Posted by Flora

Hi Caitlin,


The result I got is a far cry from good .... but, at least, it's a 'recognizable' image ....

Let me know if you are interested in the values ....

WOW! If no one else speaks up...I will! What values did you use with this? I tried it in both LAB mode and RGB mode and didn't come anywhere close to what you got. Please share!

[mistermonday]

Caitlin, looks like everyone has given you some excellent advise.
Flora, awesome work!!
Channel blending will help get the image into a "recovery" range. I made 2 copies of the red channel. I applied an inverted copy of the Blue channel to one copy the red channel. I turned the 2nd Red channel copy into a new channel by blending it in Hard Mix mode with an inverted blue channel and a copy of the original red channel. Finally applied that channel to the modified 1st red channel in normal mode at 35% opacity. This gave a reasonably good lightness channel which I applied to a duplicate image in LAB mode, where this new lightness channel replaced the L and the A & B blurred. Color tweaked in lab to reduce the yellow cast.
So there is hope for the rest of your slides. If you can get your hands on a scanner with Digital ICE s/w (like Nikon Coolscan) it may save you many hours of time on those slides which have pits in the emulsion. The technology really works.
Good luck with your project.
Regards, Murray

[bart_hickman]

I got a result similar to Flora's and Murray's, but I think my method resembles Flora's more.

What I usually do is look at the channels and see if there's a clean one. In this case there's no noise-free channel, however the blue channel is almost exclusively noise which is nearly as good. This means you'll be subtracting some amount of the blue channel from the image luminance.

Pop up a channel mixer, load the standard luminance preset, then slide blue towards the negative end until the noise disappears. Then slide green and red upwards and then blue downwards some more to get good contrast while still cancelling the noise. I ended up with +70, +102, -126, but there are many correct answers depending on the tonal range you end up with. Be sure to click the monochrome check box.

Then I extract the color channel, use Noise Ninja on it, and blend it with the luminance layer using color blend mode. Then did a saturation boost followed by a cooling photo filter to get rid of the yellowish color cast. Pretty quick operation actually--just a few minutes.

Still needs work on some color stains, but things are under control. Doing 100's of these sounds daunting unless the damage is consistent enough to batch process.

Bart

[richardsullivan]

http://palimpsest.stanford.edu/byorg.../an26-512.html

Here is a link to Stanford University with their successful FREEZING method. When moisture forms on the slide, mold is easily removed with a Q-Tip. I have not tried this yet but I will post with my results when I do.