Bad shutter or bad processing?

An attempt at correction

An attempt at correction

Rob, this caught my interest -- and I was trying to avoid getting down to work anyway ;) -- and as you and I probably represent most of the GIMP users here, I thought I'd try a little correction with the program (version 2.2.17 running on Open SUSE 10.2 64-bit, for the record).

The stripes look very regular top-to-bottom. I opened your image and used the select-rectangle tool as a ruler to confirm that they were pretty close to vertical. Then I selected a 40-pixel-tall stripe from the very top, where there's little detail except the stripes, and pasted it into a new file. Just in case the stripes changed a little from top to bottom, I took another 40-pixel sample from the bottom, pasted it into my new file, gave it 50% opacity and merged the layers, effectively creating an "average" image of the stripes against a featureless background. I then stretched that image to the height of the original image. (To be exact, since I somehow reduced your original to 800x788 when I started to work on it, I stretched the 800x40 image to 800x788.) The result is the first image below.

Using the "levels" dialog, I then adjusted the range of the "stripe" image to 0-255 (that is, I maximized the contrast). I then inverted the image (made a "negative" of it). As some "texture" was apparent in the "stripes" from top to bottom, I used the Gaussian Blur filter to smooth out the top-to-bottom (but not the left-to-right) differences. Gaussian Blur, unlike the generic blur, allows one to set separate radii for horizontal and vertical blurring. I set horizontal blurring to 0 pixels (no blurring) and vertical to 788 (complete blurring). The result was the second image below.

I then selected that entire image and pasted it into the original image. Using the layers dialog, I set the layer to "overlay" mode. (Not really knowing what I was doing, I tried "multiply," "divide," and "screen" modes. "Screen" was almost as good, but "overlay" seemed to work best.) I then experimented with the opacity and found that about 25% opacity seemed to provide a reasonable correction for the stripes in the original. I "anchored" the new layer to the original ("merged down" in Photoshop terms, I think) and the result was the third image below. I'm sure it could stand improvement, and if I understood alpha layers and correction layers better I might have been able to get a more linear correction across the brightness levels in the image, but I think the result is a start. (I'm sure there are many readers who know more about these matters.)

P.S. It's a lovely photograph! I see why you're eager to get to the root of the problem.
 

Attachments

  • crap 120 mask.JPG
    crap 120 mask.JPG
    63.9 KB · Views: 0
  • crap 120 mask stretched neg smoothed.JPG
    crap 120 mask stretched neg smoothed.JPG
    97 KB · Views: 0
  • crap 120 corrected.JPG
    crap 120 corrected.JPG
    265.5 KB · Views: 0
Last edited:
Because I'm splitting my time between here and home and have no room for a dark room (or another box of "stuff") and really don't have the $ for a decent MF scanner and don't have the time to process and scan...umm, no it would be very much harder and results possibly no better than what I've got now. I'm happy to pay for expertise and skill for this kind of thing, I just thought I had done so.
titrisol said:
not to be a party pooper....
wouldn;t it be easy if you processed the 120s yourself?
OK, now that is what I was musing about above. So I have printed this post out and will be using it as a guide while I look at the rest of the negs.
Thank you so much!:cool:
And, you didn't change the size; the square format of this camera isn't quite square; it's about 6 x 5.92 rather than 6 x 6.

mjflory said:
Rob, this caught my interest -- and I was trying to avoid getting down to work anyway ;) -- and as you and I probably represent most of the GIMP users here, I thought I'd try a little correction with the program (version 2.2.17 running on Open SUSE 10.2 64-bit, for the record).

The stripes look very regular top-to-bottom. I opened your image and used the select-rectangle tool as a ruler to confirm that they were pretty close to vertical. Then I selected a 40-pixel-tall stripe from the very top, where there's little detail except the stripes, and pasted it into a new file. Just in case the stripes changed a little from top to bottom, I took another 40-pixel sample from the bottom, pasted it into my new file, gave it 50% opacity and merged the layers, effectively creating an "average" image of the stripes against a featureless background. I then stretched that image to the height of the original image. (To be exact, since I somehow reduced your original to 800x788 when I started to work on it, I stretched the 800x40 image to 800x788.) The result is the first image below.

Using the "levels" dialog, I then adjusted the range of the "stripe" image to 0-255 (that is, I maximized the contrast). I then inverted the image (made a "negative" of it). As some "texture" was apparent in the "stripes" from top to bottom, I used the Gaussian Blur filter to smooth out the top-to-bottom (but not the left-to-right) differences. Gaussian Blur, unlike the generic blur, allows one to set separate radii for horizontal and vertical blurring. I set horizontal blurring to 0 pixels (no blurring) and vertical to 788 (complete blurring). The result was the second image below.

I then selected that entire image and pasted it into the original image. Using the layers dialog, I set the layer to "overlay" mode. (Not really knowing what I was doing, I tried "multiply," "divide," and "screen" modes. "Screen" was almost as good, but "overlay" seemed to work best.) I then experimented with the opacity and found that about 25% opacity seemed to provide a reasonable correction for the stripes in the original. I "anchored" the new layer to the original ("merged down" in Photoshop terms, I think) and the result was the third image below. I'm sure it could stand improvement, and if I understood alpha layers and correction layers better I might have been able to get a more linear correction across the brightness levels in the image, but I think the result is a start. (I'm sure there are many readers who know more about these matters.)

P.S. It's a lovely photograph! I see why you're eager to get to the root of the problem.
 
Back
Top Bottom