Nostalgia & History > Artifacts With or Without Stretching the Contrast of a Scan

Evan_Werkema and I have been having discussions about scanning for some time.  The other day I said I have found that you introduce more artifacts if you stretch the contrast of the scan to the black and white points of each color because that darkens the scan and you have to do more lightening to get the scan suitable for presentation.  I think we all have found out the more you diddle with a scan the more you enhance the artifacts.  Conversely, the stretching of the contrast for each color does much color correction.

Evan asked me if I could do a comparison of not stretching the contrast and correcting the scan manually to stretching the contrast of each color and correcting the scan manually.

So here it is.

I chose an awful slide (first photo below).  That's the scan right off the scanner.  It's a Burlington Northern inbound Dinky at Lisle, IL pushing.  Yes the headlight is on.  The fog is heavy enough that there are no shadows.  The camera (a Cannon A1 SLR) is looking mostly east at the sun.  Thus the photo is very flat (little contract).

First I stretched the colors to the black and white points as shown in the second photo below.  It is darker (look at the ballast and the pilot) and the greens seem a little more vivid.


Continued ...



Edited 1 time(s). Last edit at 04/19/16 21:18 by MartyBernard.

---

---

Second I manually lightened mainly the dark areas of the stretched scan, took out a little blue due to the aging of the Kodachrome, and sharpened it.  I did not do any noise reduction which would, of course, change the artifacts.  The result is the first photo below.

Third, I manually lightened the dark areas of the scan, did a little color correction, sharpened it the same amount, but did no noise reduction.


Continued ...



Edited 1 time(s). Last edit at 04/19/16 21:08 by MartyBernard.

---

---

Here are three enlargements of the E-unit's nose, the scan, the stretched and lightened, and the non-stretched lightened photos.  Yes they are all awful.  And the stretched shows the most pixilation because I had to do more drastic lightening than I did to get the non-stretched version lightened.

Note the enlargement of stretched but not lightened version look like the enlargement of the scan in terms of artifacts.

So Evan, what do you think?

FYI, below the enlargements are the histograms of the stretched and unstretched versions.  They look as expected.



Edited 2 time(s). Last edit at 04/19/16 21:50 by MartyBernard.

---

---

---

.



Edited 1 time(s). Last edit at 04/19/16 21:15 by MartyBernard.

---

---

Thanks Marty.  I can see what you mean about pixelation, but fiddling with the same images in Photoshop, it looks like the sharpening is what's really introducing the noise.  Sharpening works by increasing the contrast between pixels that already differ by a certain threshold value, so it stands to reason that the higher contrast of the stretched image would enhance the effect.  The third image above looks much "sharper" than the fourth even though you sharpened them by the same amount.  This is a case where I would be reluctant to sharpen at all just because of the noise issues.  Was this a multi-sample scan?  If not, you might try a multi-sample scan to see if it reduces the noise.  It takes longer, but it can help to make shadows less noisy when lightened and sharpened. 

I do think this image works better with higher contrast.  In fact, I would boost the contrast even more: darken the shadows and lighten the highlights so that the image approximates a "glint" shot as below.  Yes, the undercarriage is reduced to a few highlights as are the ballast and the tracks in the foreground, but I would argue that those details aren't really important in this particular image.  As a documentary or modeling guide, as you said, it's not a very good slide.  As a mood shot, though, those dark-and-busy areas can be darkened more, minimizing them and drawing the focus more to the train and the side of the locomotive.  On my monitor, the roofline of the coaches against the sky is still distinct, and the edge of the E's nose is discernible against the foliage.  Sharpening will again introduce noise, and I still think I would forego it. 

---

Yes it was a multi-scan so to speak.  My Nikon Coolscan does infrared cleaning during the same pass as the scan.  Noise reduction was on too but I'm sure that has to be done after the pass.

Marty
 



Edited 1 time(s). Last edit at 04/20/16 19:20 by MartyBernard.

Well, I always like a challenge. Sorry about the signal bridge - forgot to replace it. I'd give my effort a "meh +".....

Bob

ATSF100WEST......Out

---

Bob,

I was not trying to show a finished product, just a comparison after sharpening.  Before I'd call to photo done, I would have at least denoised it and resharpened it. 

I do like your version.  With which version did you start and what did you do to it?

Marty Bernard

 

Evan, some of the reason you might be able to get away without sharpening it further is that VueScan did some sharpening.

Marty