DSLR scan with green light source

[tvdpid]

Hi,
I am relatively new to B&W film-photography.
I develop the negatives myself and used the Epson V500 to scan them.
I read lots of articles on ‘scanning secrets’ and post-processing and tried them all, but was never very happy with the results, till I tried DSLR scanning.
The setup is as follows:
- Nikon D90 body
- 13 mm extension tube
- Nikon F-mount to M43 screw thread adapter
- M42 tot M39 screw thread adapter
- Elmar 5 cm f3.5
- Two pieces of glass (from photoframes) in between which I put the negatives
- White piece of plastic foil
- Light source: Smartphone which shows a White rectangle.
The body is connected to my PC and mounted on a cross slide vise which allows me to fine tune the focus distance. I shoot in RAW-format.
The ‘scanning’ process is very fast compared tot he Epson scan.
The image quality is much better. I don’t need special sharpening recipes anymore.
The only thing I have to do to get a decent picture is a Levels-adjustment and (eventual) a single basic sharpening of the edges. (From there - if you need to – you can start being creative and do other things )

Last week I tried different light sources: Red, Green and Blue.
The pictures shot with the green light are sharper than those shot with red or blue light.
I have the impression that the image quality of the pictures shot in green light is better than the quality of pictures shot in white light and also better than the green channel of those pictures shot in white light.
Does anyone have experience with that?
Maybe I am so excited that I see things which are not there .

Regards,
Tom

 

[sevo]

At the data level, that positively cannot happen. The Beyer grid has twice as many green pixels than red and blue - so white light should give you full resolution, green half and red or blue one quarter. If you perceive an advantage with coloured light, you should experiment with different raw converters - you probably have stumbled upon conversion artefacts.

 

[tvdpid]

Hi sevo,
Thanks for your reply.
I forgot to say that I only use the green channel of the shots in green/white light. The red and blue channels are crap when shot in green light.
There is a distinctive difference in the histograms between those green channels. The histogram from the 'green shot' looks more refined (less peaks). So my camera sees different things ...

 

[calebarchie]

Did you know white light consists of Red, Green and Blue light?
You are doing a number of things wrong.

 

[Dwig]

Several things are likely to be at play here.

One is the Bayer matrix issues mentioned by servo

Two involves the lens. You may be encountering some significant optical chromatic aberration (I'm NOT referring to the CA produced by the sensor's micro lenses). Lenses are never perfect when it comes to focusing all colors exactly the same. When you use a narrow spectrum light source (e.g. green only) you often get sharper images. With "white" light, when one color is in focus others sometimes aren't. There are also issues with the images sizes changing with the color change which produces lateral color "smearing".

Combine the two and it isn't surprising that green only is better than either red only and blue only simply because of the issues with the Bayer matrix population ratios. Green only should, in theory, be less sharp that white except lenses aren't perfect and you're using a lens that was not optimized for macro and is likely producing enough chromatic aberration to degrade the image more than the increased number of photosites helps.

 

[mfogiel]

You should probably look for a repro lens like one of the APO Rodagons.

 

[mdarnton]

Interesting. The next time I camera scan I will try this. All explanations aside as to why it can't work (and I see a lack of consistency to the explanations, so I'm more skeptical of them than of you) results are what talks, not opinions.

 

[Larry H-L]

I don't know about the Bayer issue, but when I scan with an Epson, I choose the green channel, it gives the best results.

 

[sevo]

I don't know about the Bayer issue, but when I scan with an Epson, I choose the green channel, it gives the best results.

Scanners do not have a Bayer mosaic, but a scan line, where each colour has the same pixel count. With Epsons I'd suspect that green, being the middle channel, is the channel they use for focusing, so the others will suffer from whatever chromatic aberration the scanner lens has (and current consumer flatbed scanners have as primitive lenses as they can get away with, so there must be a fair deal of it - the pro flatbeds of the past, which had proper Apo lenses, had the weight and price of a small car).

 

[Dante_Stella]

I don't know about the Bayer issue, but when I scan with an Epson, I choose the green channel, it gives the best results.

Same is true of Nikon and Polaroid negative scanners (line CCDs).

Dante

 

[tvdpid]

Of course I know that white light consists of red, green and blue RGB(255,255,255) 🙂, but because of the Bayer pattern which consists of 50% green pixels, it could be possible that a lack of red and blue may result in something different. That's why I did the test. If you don't test, you will never know.
I don't think that there are problems with abberations. When shot with the white source, the quality of my image is top in relation to the V500. The only thing I want to share is that I have the impression that the quality is even better when I shoot with a green light source. RGB(0,255,0) 🙂 But as I also mentioned, I may be wrong ... There might be a parallel with the story about the green channel when using a scanner.
I will prepare some pictures, where you can see the difference between the histograms of the green channels.
To be continued ...

 

[tvdpid]

I was inspecting the histograms of the 'scans' I made with the white and the green light source and what I see is that the pixels from the scans I made with the green source cover a significantly wider tonal range than those from the scans with the white light source. The difference is ca. 20 on the scale 0 to 255. That's why the scans look less flat. Something has to do with the intensity of the light, which i didn't measure, but as far as I can estimate correctly, it seems that I need 1/3 stop less exposure with the green source than with the white source (which is some characteristic of my smartphone of course). So I compared e.g. f5.6 1s white light with f5.6 0.7s green light.

 

[tvdpid]

Two histograms which illustrate previous reply. Find the 7 differences ...
(I know, things may be subtle, but the difference in tonal range is not ...)
The first histogram is the green channel of the 'white shot', the second is the green channel of the 'green shot'.
The green shot is 1/3 stop less exposed.

 

[tvdpid]

Pics belonging to the two histograms in previous reply.
These are the green channels straight out of the camera. No noise reduction, no sharpening. Just inverted the negative image.
The flattest is the one shot with the white light source.
click to enlarge.
If you want, I can upload bigger pictures ...