How can I reconstruct a color image from a three-frame glass negative?

Asked 7/12/2013

4 views

2 answers

0

I have a digital color composite made from a glass negative containing three black-and-white frames used for three-color separation. The source appears to be a single glass plate with three exposures taken through different color filters. In the composite, the channels do not line up perfectly, and lighting may have changed between exposures. Is it possible to restore the "original" image, and what is the best way to process it?

Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0

Photography Stack Exchange contributor

13y ago

2 Answers

4

There is not "as clicked", I'm afraid, unless you are willing to settle for a black and white picture.

The main problem with the picture is that it comes from three separate exposures, made on black and white glass plates using colour separation filters (probably Wratten numbers 29, 47 and 61). Or, more specifically, it was made using three exposures separated in time as well as by colour on a day when there was sunlight peeking through moving clouds. That means that no matter how carefully you align the image's edges (and they are well-aligned here, if not quite perfect), you will never be able to make a true-colour composite image from the three plates. (There were ways available to make three simultaneous exposures using three cameras and a beam-splitter arrangement, but that would have meant a whole lot more weight and expense for the photographer.) Tricolour photography didn't always work out nicely.

You can do a fresh colour separation digitally if you don't have access to the original plates. That leaves you with several options to improve the photo.

The hard way would be to do some heavy-duty dodging and burning on each of the colour channels to try to even out the tonalities and make the pools of light appear in the same places on each of the channels. Then you could recombine the channels to approximate what the picture might have looked like if taken on a less-windy day. Depending on your personal perfectionism quotient, your patience, and so forth, that may or may not do permanent injury to your mental health.

The second approach is less "real" but much more practical. Choose the channel that gives you the best overall tonality, and use a copy of that as the basis for a "hand-tinted" image. (You may still have to do some dodging and burning to remove tonalities that cam from colour contamination.) Then make a bunch of different mixes of the three channels to find convincing swatches of the colours that were most likely to have been there in "real life". ("Real life", in this case, meaning the colours as they were likely to have appeared in an ideal contemporary tricolour image; that may take some study. If this is more about creating an image than "restoring" an image, then feel free to use something closer to what a modern camera would have captured.) You can sample the target colours and either add them to a separate colour source image or to the swatches collection of your image editor. Once you have found the colours you want, then there only remains the tedious process of colourizing your best separation plate. Yes, it's a lot of work, but it's a whole lot less frustrating than the procedure above.

Originally by user2719. Source · Licensed CC BY-SA 4.0

user2719

13y ago

0

AI Answer

Generated from our catalog & community — verify before relying on it.

You generally can’t recover a single true "as-shot" color image from this perfectly. A three-color separation glass negative was made from three separate black-and-white exposures through different color filters, so if anything in the scene moved—or the light changed between exposures—the color composite will never be fully accurate.

What you can do is improve it:

  1. Split the scan into separate red, green, and blue channel layers.
  2. Align each channel independently to reduce color fringing.
  3. Recombine the channels into a new RGB image.

This can produce a better-looking result, but it won’t fix differences caused by moving clouds, changing sunlight, or subject motion between the three exposures. In those areas, color errors are unavoidable because the three records were not captured at the same instant.

If your goal is historical fidelity, the most reliable "original" may simply be one of the black-and-white separation frames, or a monochrome rendering from the scan. If your goal is presentation, careful channel alignment is the right restoration approach, with the understanding that some color artifacts cannot be eliminated.

UniqueBot

AI

13y ago

Your Answer