Very dark old slides – how best to adjust them?

JS
Posted By
Jon_Shippam
Sep 3, 2006
Views
1135
Replies
30
Status
Closed
I’ve been scanning some old family photos, and discovered that some of the darker areas have become disproportionately dark. I’ve tried increasing the analogue gain setting and this just over exposes the light areas but doesn’t help much with the dark areas (using Nikon Scan 4.0.2 and Coolscan 5000 ED).

I’ve come to the conclusion that the adjustment is best done in photoshop, and so far I have tried using Image:adjustments:curves… and Image:adjustments:levels. I’ve been able to improve the images with both of these techniques, although it’s hard to get it looking natural using curves. In levels… I’ve found that dragging the middle slider to the left significantly improves things.

My question is – is this the best way of improving my image?

An example image can be seen here: <http://web.mac.com/shippam/iWeb/Site/Photoshop.html>

Is it best to do colour correction before or after this? Example3 is done after adjusting levels.

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

R
Ram
Sep 3, 2006
What version of Photoshop and the OS? (Makes a difference.)

Generally speaking, using levels for adjusting brightness, contrast or color is kind of like using a chain saw for brain surgery.

Curves are better.

Also, you’ll find some images respond well to the technique described here: CLICK HERE <http://www.adobeforums.com/cgi-bin/webx/.3bbf8b64/3>

You might also want to get the best scan possible during the scanning process.

<http://www.scantips.com/>
JS
Jon_Shippam
Sep 3, 2006
Thanks for the reply and the links.

I’m using Mac OS X 10.4.7, and Photoshop CS2.

I did try comparing the results by adjusting levels and curves, and found that curves were very difficult to get right and tended to make the image look horrible.

I hope I’m getting the best scan possible during the scan process. I’m scanning in Nikon Scan, saving as 16 bit TIFF, using Fine setting for ICE, a low GEM setting and but I have turned off some of the other ‘enhancement’ features as I found they produced some really nasty results. I have it set to 4x multisample scanning.

I’m using a technique described in Scott Selby’s Photoshop CS2 for Digital Photographers for colour correction using a customised setting in Curves:Auto (used in Example3 after adjusting levels).

Do you have any tips for using curves? Your comment about chainsaws seems to be the opposite way round to what I’m getting. I can adjust the mid-point setting in levels to get a subtle and natural change, but when I try curves (I mean pulling the straight line into a a sigmoid curve) I just get bizzarre results.
JS
Jon_Shippam
Sep 3, 2006
PS. I forgot to mention that about the only thing I change between slides during the scanning process is Analogue Gain (and rotation and cropping). I make sure that the histogram isn’t cutting anything off at either end. For example, if the curve extends off to the left I increase the analogue gain, and vice versa. Most of the well exposed images fortunately don’t require any adjustment.

However, it’s these old images, where I presume the colours have darkened over time, where the histogram is within the range, but very ‘bottom heavy’, skewed dramatically to the left.
R
Ram
Sep 3, 2006
described in Scott Selby’s Photoshop CS2

That’s probably Scott Kelby. He has his following for sure, but I rather look elsewhere. Not that I’m suggesting you should not pay attention to him; I’m merely explaining why I can’t comment specifically on any of his techniques. Just a personal preference, that’s all.
E
eltee
Sep 3, 2006
there’s a learning curve to using… curves : )

this one may help a bit:

<http://www.retouching.com/LT/2ladiesHalfGuy.acv>

control-click on link, save file to desktop, remove .html, load curve in Photoshop as an adjustment layer

I used sRGB (you don’t have to, but the curve would work different)
CB
charles badland
Sep 3, 2006
Sometimes a slight Shadow/Highlight adjustment can greatly improve dark images. (note the "sometimes")
RR
Reed_Reed
Sep 3, 2006
Jon,

I have 2 suggestions for you. The first, if these images are very valuable to you and your family is to have them drum scanned. There are many drum scanning services on the Internet. You’ll pay between $25 to $50 for a 100 mb scan of a 35 mm transparency. Drum scanners have MUCH greater D Max and dynamic range than anything short of a high end Heidleberg flatbed CCD scanner and will produce detail easily in dark images that a Nikon Coolscan simply cannot give you. I operate drum scanners and the hype the manufacturers publish about the so-called dynamic range of CCD scanners is downright fraudulent.

I am not familiar with the operation of the Nikon scanner, but when you mention changing the "analog gain setting" I am not sure that the machine does anything to change the sensitivity of the CCD sensors. I don’t think that can be done, but I’m not sure. It IS possible on a drum scanner to change the sensitivity of the photomultiplier tubes, and that is one of the great advantages of a drum scanner. Drum scanners can dig down into dark shadows that are just impossible for CCD scanners to see.

I believe that the so-called analog gain settings only change the way the Nikon software handles the scanner’s raw image data. So you are just applying the same kind of curve controls in the scanner that are available in Photoshop after you open the image there. Not much advantage doing it in the scanner.

The second suggestion is a method we used to "rescue" a project that was given to us after it was scanned on a Minolta Dimage film scanner (with very similar specs to your Nikon – for all I know, the CCD sensor is the same in both). The customer was convinced that his "5000 dpi" scanner was the equivalent of a drum scanner (because that’s what the industry buzz said).

All the scans lacked detail due to very poor dynamic range in the shadows, but there was no apportunity to rescan. We had to use the files we were given. We tried all kinds of curves adjustments in Photoshop and we were able to improve the shadows a lot with many different combinations. But when we got the shadows lightened enough, we always blew out the highlights. We could never adequately save both, no matter how many combinations of adjustment layers we used.

The method that worked was to create a copy of each image and re-combine the 2 as though we were working with an under-exposure and an over-exposure. The same way that a digital camera can be used to make several images at widely different exposures that are then digitally blended in Photoshop to capture a very wide dynamic range.

The general method is described on this page:

< http://www.luminous-landscape.com/tutorials/digital-blending .shtml>

(Scroll down to the section labeled "The Layer Mask")

Essentially, you paste a darker version of an image into a layer above a lighter version of the same image and follow the instructions on that page to blend the 2 together.

We applied a huge curve to our images, completely blowing out the highlights and lightening the shadows, saved that altered image (different name than the original), then placed the dark version onto a layer mask above the lightened version for digital blending.

I didn’t think the method would work. I reasoned that we were creating no new data and that re-combining the 2 images could achieve nothing that Photoshop couldn’t do with curves and adjustment layers in the first place. But it DID work. It worked MUCH better than anything we could do in Photoshop with a single image.

(Although we improved the images, they weren’t close to the quality that drum scanning would have done)

If the "analog gain" controls of your Nikon scanner actually do allow you to over-expose your images, then you could actually produce 2 different real exposures which you could digitally blend, and probably get very good results.
R
Ram
Sep 4, 2006
Good post, Reed. Thanks!
CC
Conrad_Chavez
Sep 4, 2006
I have used Analog Gain on a Nikon scanner and can tell you that its effects are hardware-based and can’t be replicated by Curves or anything else in the software. Analog gain alters the CCD exposure time <http://www.nikon-image.com/eng/ei_cs/faq/qa/qno_1531.htm>. It’s perfectly reasonable for you to try it though it does have its limits <http://www.lumika.org/gear_nikon_scan_analog_gain.htm>, such as increased noise and loss of contrast. Here’s another link <http://www.sphoto.com/techinfo/analoggain.htm> from Google.

As far as dark slides, if it isn’t easy to get what you want with Curves, try two other techniques before applying Curves. One is the Shadow/Highlight feature mentioned by Charles, particularly the Shadow slider in your case. The other thing you can try is duplicating the slide layer, applying the Screen blending mode to the upper copy, and adjusting opacity. Both of those techniques can give you a better starting point for Curves tweakage.
R
Ram
Sep 4, 2006
One is the Shadow/Highlight feature mentioned by Charles,

That’s what I suggested in post #1.

The clickable link(CLICK HERE <http://www.adobeforums.com/cgi-bin/webx/.3bbf8b64/3>) takes you to a tutorial posted by Phosphor in The Photoshop Lounge Resource Repository showing Mathias Vejerslev’s advanced technique of using the Shadow/Highlight adjustment tool.
RR
Reed_Reed
Sep 4, 2006
Conrad,

"Anaolg gain alters exposure time."

Ah, then Jon can use the digital blending method to very good advantage by combining a "normally exposed" file and an "over-exposed" one.

Using the Shadow/Highlight feature was a far better tool than any combination of Curves adjustment layers we could construct (I’m sure one of the Photoshop engineers could show us a complex combination of Curves settings with blending sliders settings that would do what Shadows/Highlights accomplishes in its single dialog).

But the 2-image blending method we finally used was a lot better than even the Shadows/Highlight tool.
CB
charles badland
Sep 4, 2006
Ramón: That’s what I suggested in post #1.

Ooops. I didn’t "Click Here"!
PF
Peter_Figen
Sep 5, 2006
There is no substitute for a photomultiplier tube, the light gathering sensor of a real drum scanner. Reed is correct. Even with the controls on the Nikon hardware, you’ll still be minus at least two stops of real shadow detail, and possibly more. On a good drum scanner with good software (and that makes a huge difference), there’s no need to scan twice. Just scan without clipping highlights or shadow and correct to your heart’s content in PS.

I CCD scanner can only "see" to a maximum density of somewhere between 2.7 – 3.0 at the outset. A pmt on a drum has no problem pulling detail all the way down to 3.9 or 4.0. Do the math from there. If pulling max shadow detail, have it drum scanned by someone who really knows how to scan.
JF
john_findley
Sep 5, 2006
That’s all well and good. I’m sure all this "drum scanner advice" is correct, I believe it. But, practically speaking, I (like Jon) am in the process of scanning (and correcting) approx 1800 slides and 700+ old family photos to preserve them for the kids and grandkids. It’s been a two-year project, so far.

What’s all this drum scanning gonna sat us back?
R
Ram
Sep 5, 2006
What Reed said:

"You’ll pay between $25 to $50 for a 100 mb scan of a 35 mm transparency."

You can buy a luxury car instead. 🙂
PF
Peter_Figen
Sep 5, 2006
With that many slides to scan, it would indeed be prohibitive for most people, but it’s good to know why cheap slide scanners may not deliver what you expect or need, leaving you the choice or opportunity to perhaps have a few of the more special shot receive special treatment.
JF
john_findley
Sep 5, 2006
Re-reading my message, I realize it sounded confrontational. Shoulda put a smilie after that last line.

🙂

Thank you for the low-cost solutions.
A thing I’ve found, besides the darkening, is that the dyes seem to be spreading somewhat. My slides are from about 1944 through 1960s. There is also mold damage (they’re from Florida) which can be clone stamped, generally. But the dye problem seems insoluble.

Sorry to divert.
AS
Ann_Shelbourne
Sep 5, 2006
To remove mould damage, try stamping or healing in Color mode.

Soluble dye problems probably are "insoluble" — but I wonder if sharpening in Luminosity might help?
RR
Reed_Reed
Sep 6, 2006
John,

Unfortunately, there is no escaping the reality of the difference between a PMT-scanned image and one done on a typical CCD "film scanner." The reputation of high-end CCD scanners is also somewhat suspect (in my opinion).

For a project as large as you have, it would not be a bad idea to consider getting a used drum scanner. I have seen machines such as a Howtek D4000 sell on Ebay for less than $1000. (They’re darn good. Museums use them to digitize tone of images) Of course, you could get a pig in a poke there. Or a gem. Reputable equipment sellers sell such machines in the range of $2000-2500. And they guarantee their condition. Software such as Silverfast is another $500-700.

Then there is the learning curve! It’s not unusual to spend more than a year getting proficient, and that’s with other people who know how to use a scanner, teaching. Oil mounting is fun! In the end, though, you would have all those memories in much better image shape than otherwise, and you would have a free drum scanner.

Somehow, I don’t think I convinced you!
JF
john_findley
Sep 6, 2006
That’s all well and good. I’m sure all this "drum scanner advice" is correct, I believe it.

–me 🙂
AW
Allen_Wicks
Sep 6, 2006
This technique mentioned by Conrad has been little discussed:

The other thing you can try is duplicating the slide layer, applying the Screen blending mode to the upper copy…

It is very fast and for certain images gets excellent results. After creating a duplicate layer in screen mode you can adjust the opacity of the dupe layer if too light, or even merge layers and repeat the process if still too dark. If an image has light and dark you can erase the dupe layer at the overly light sections.
AR
alan_ruta
Sep 7, 2006
Conrads technique can work provided there is detail there to begin with. This is where $50,000 (and up) drum scanners excel. It may be dark but the detail is there so it can be brought out. Same issue with highlights–if there is something there they can be multiplied.

Of course ecconmics (vs. time) will be a big factor in any decision.

If only I had room–last year someone I knew thru out a Linotype 3300 scanner. Unreal.

alan
AS
Ann_Shelbourne
Sep 7, 2006
You can get great scans from a NikonScan if you work at learning how to drive the NikonScan software to the max.

For those really contrasty trannies with dark shadows and over-exposed highlights, you are probably going to have to make two scans using different Analogue Gain (and Analogue Loss too!) settings and then combine the two in Photoshop with suitable Layer masking.

You should be able to get perfectly acceptable results for most purposes without having to go to the expense of buying drum scans.
PF
Peter_Figen
Sep 7, 2006
Ah, the venerable 3300. A great scanner, but quite complicated to use. No live preview on screen. You looked at the image through a scope on the scanner, picking areas through a crosshair for things like white point and black points. 8 bit CMYK only. No RGB output, and usually oversharpened at the same time.

No matter what techniques you use with a Nikon, it will never come close to both the shadow and highlight detail of a pmt in good condition. Yes, pmt sensors see much better highlight detail as well as shadows. Just one more thing to consider.
AS
Ann_Shelbourne
Sep 7, 2006
Sure Peter.

But for some purposes, economics DOES come into play — and I suspect that for the OP and John Findlay that applies in both of their cases?

I am fairly certain that, provided there is actually some smidgeon of detail in those shadows and HLs., it will be possible to get a sufficiently good image using NikonScan and Photoshop.
JC
Jonathan_Clymer
Sep 7, 2006
My personal experience is that you will see shadow detail from a drum scan (a good scan) that you won’t see from a CCD scanner. There are lots of drum scanners sitting around gathering dust, but perfectly usable. If you don’t want to buy one maybe you can rent some time on one.

If you can’t go the drum scan route, you might try Dan Margulis’s false profile method: Create several low gamma profiles, and preview each of them on the image. Apply the one that looks best, and then convert to a normal profile.

Jonathan Clymer
AR
alan_ruta
Sep 7, 2006
Good drum scanners pull out detail that you can’t even see on a bright lightbox. They really are amazing machines.

But I agree Ann economics can matter. I think one has to see if the CCD result is good enough. Maybe it will be for 95% of the scans and some can be farmed.

alan
SP
steve_peters
Sep 7, 2006
What about a Imacon Flextight scanner? I know that those can be rented, and are usually slot cheaper than a drum scan. Also, what is the final image going to be printed on? You may spend alot of time trying to pull out shadow detail that may not even show up when you print it. Of course that will depend on how it is printed.
KN
Ken_Nielsen
Sep 7, 2006
"Very dark old slides – how best to adjust them?"

It’s nice to see that the question has been answered here, not only for how to deal best with an existing flatbed scanner, but beyond to the best way available, albeit at more cost.

On the practical side, for someone who wants the best quality, I think I would recommend taking the transparencies in to a lab with a good scanner, operator and software, and discuss the project with them and get a quote. You might find out that there would be cost savings by ganging up scans or through some other technique the operator can recommend. Someone in the business that looks at the project first-hand might be able to give a total cost that would be feasible while also giving the customer first-rate results.

It costs nothing to ask.
PF
Peter_Figen
Sep 7, 2006
Steve,

The Imacon’s CCD is no better than the Nikon in extracting shadow detail. In many ways, the Nikon is a better scanner. The Imacon, especially on 35mm simply does not deliver its claimed resolution. No matter what is claimed, they max out at 4000 ppi or less. Imacon had a great marketing campaign that many people bought into. I demoed one for a few weeks and bought a Howtek instead. One of the best investments I’ve ever made.

One thing is for certain. If your scanner can’t pull out the detail, you will never be able to print it. If your scanner can, then you at least have the option of trying to print it.

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

Related Discussion Topics

Nice and short text about related topics in discussion sections