Camera "scanning" with a fluorescent light box.

[Lachlan Young]

As said above by Les Sarile, if you're happy with the results, fine. But I'd be cautious about using a random fluorescent source for color scanning, as the output contains a number of spectral lines. Worst case is one line halfway between the peaks of two dyes --> maximum cross-talk. Presumably when Epson (for example) uses a fluorescent source in a scanner, they are careful about phosphor selection to avoid that. Better is incandescent (true continuous spectrum); even better RVB LEDs (Nikon) centered on the dyes peak wavelengths. Possibly anecdotal, but I was once struck by the higher saturation of same C-41 frame scanned with LS-2000 versus V-700; not self-delusion because only after did I realize the difference in light sources.

More likely that the DSP is set up to correct the spectral output of a fluorescent lamp/ LED array against what it understands to be an equal energy source. Plenty of high end CCD scanners use fluorescent sources & get far better results than the Nikons.

 

[Anon Ymous]

I would need to make some more comparisons to assess the colour. The Nikon is giving a much 'cleaner' result. The negative here is 10 years old so I can't reliably recall the exact lighting conditions of the day.

IMHO, the Nikon/Vuescan scan still has a green cast, especially in the mid to high tones, even after your adjustment. The sky looks a bit too green, but is easily correctable. I've noticed Vuescan giving quite greenish results in many cases with my Minolta scanner and sometimes you think you've done a good job correcting casts only to realize that correcting a bit more gives even better results.

 

[Les Sarile]

I would need to make some more comparisons to assess the colour. The Nikon is giving a much 'cleaner' result. The negative here is 10 years old so I can't reliably recall the exact lighting conditions of the day.

Since you used Vuescan with your Coolscan, did you use the built-in profile for Kodak 160VC and a setting of White Balance?
The reason I ask is that the sky looks similar to the results I got with that combination. Below is an example of my result using Kodak 160VC with Coolscan+Vuescan and the built-in film profile as well as the various modifiers compared to a completely neutral scan with Nikonscan.

Full res -> http://www.fototime.com/CFB9B61FC2EBFE1/orig.jpg
The sky rendition in #10 looks very similar to your results. It is a more greenish tint but I did no post on any of these except assembling them as one image with labels.

By "cleaner" I assume you mean no dust and scratches due to Coolscan's ICE? Yeah, I am familiar with Coolscan 9000+Nikonscan's magical ICE capabilities . . .

Full res version -> http://www.fototime.com/496ADC2265357CA/orig.jpg
Clearly another reason I have little motivation to enhance my camera scanning skills . . .

 

[Tom Kershaw]

Since you used Vuescan with your Coolscan, did you use the built-in profile for Kodak 160VC and a setting of White Balance?

No, a raw tiff file into ColorPerfect.

 

[bernard_L]

More likely that the DSP is set up to correct the spectral output of a fluorescent lamp/ LED array against what it understands to be an equal energy source.

No way. Once the spectral energy distribution of the source has been folded (convolved, in technical language) into that of the dyes, all electrons are alike.

Plenty of high end CCD scanners use fluorescent sources & get far better results than the Nikons.

No contest. I give the benefit of doubt to Epson (V700) and high end scanner designers to choose the best possible fluo lamp. My remark was about picking any consumer fluo source for the purpose of scanning. Example (quoted from this article)

See that peak near 490nm, sitting halfway between Blue and Green? And the big hump around 600nm, just straddling the transition between Green and Red? So-called "white" LEDs are hardly better.

 

[Lachlan Young]

No way. Once the spectral energy distribution of the source has been folded (convolved, in technical language) into that of the dyes, all electrons are alike.

No contest. I give the benefit of doubt to Epson (V700) and high end scanner designers to choose the best possible fluo lamp. My remark was about picking any consumer fluo source for the purpose of scanning. Example (quoted from this article)
View attachment 244224

See that peak near 490nm, sitting halfway between Blue and Green? And the big hump around 600nm, just straddling the transition between Green and Red? So-called "white" LEDs are hardly better.

Point is that Tom appears to be using a decent quality lightbox - which generally had to be pretty accurate for judging the colour balance of transparencies on.

 

[Tom Kershaw]

Point is that Tom appears to be using a decent quality lightbox - which generally had to be pretty accurate for judging the colour balance of transparencies on.

Yes. There is just very little information contextualising the various different potential light sources and how they fit into a camera-as-scanner system. For practical evaluation I suspect you'd want to have a series of light sources to cross reference. The more sophisticated approaches of multi-shot exposure of a piece of film without moving the camera may also change the results, in that case I would suspect the fluorescent flicker to pose more of a problem.

 

[Lachlan Young]

Yes. There is just very little information contextualising the various different potential light sources and how they fit into a camera-as-scanner system. For practical evaluation I suspect you'd want to have a series of light sources to cross reference. The more sophisticated approaches of multi-shot exposure of a piece of film without moving the camera may also change the results, in that case I would suspect the fluorescent flicker to pose more of a problem.

Diffused strobe is potentially the most consistent - depending on how much power is needed, it should be fairly straightforward to design.

 

[Adrian Bacon]

One thing to watch out for is that some flourescent lights have quite a bit of flicker at 120 Hz (twice the line frequency) in the US or 100 Hz in most of the rest of the world. This can cause artifacts in a scan-by-camera system. Using an extra slow shutter speed can reduce this effect by averaging over several cycles. I think there are some flourescent lights that are low-flicker. Obviously, I don't know if this applies to your light box.

Incandescent lights also have some flicker, in this case it would be at line frequency instead of twice line frequency. However, I believe the flicker from incandescent lamps is less than the flicker from most fluorescent lamps.

Most newer cameras have flicker detection and will trip the shutter at the best time to avoid that when you have a faster shutter speed.

 

[jtk]

To my knowledge there is only one source of near-perfect "daylight" for photo purposes, other than strobe.

https://color.xrite.com/en/macbeth?utm_source=bing&utm_medium=cpc&utm_campaign=01-BI-NA-EN-Lighting&utm_content=Macbeth_Lighting&utm_term=macbeth light&matchtype=p&device=c

I obtained Macbeth fluorescent tubes for my formerly-Xray-viewing lightbox. For duplication it's a lot better than LED purported "daylight" because the spectrum isn't as seriously chopped.

https://www.xrite.com/-/media/xrite...duct_brochure/l10-316_macbeth_lighting_en.pdf

 

[PGillin]

Diffused strobe is potentially the most consistent - depending on how much power is needed, it should be fairly straightforward to design.

Two diffused strobes at 45 degrees to subject should work. A consistent distance would be helpful in establishing a consistent, quick, set up and workflow. Easy enough to mark off a setup like this on a studio or lab floor. I've been expirimenting with "scanning" over a kaiser as well, but I'm planning to move to a pair of speedo heads running on an 805 pack.
The Kaiser is OK, but I've been having issues with long-ish exposures and getting my tripod set up above the light table. I've mostly been doing B&W, so can't comment much on color accuracy.

 

[Tom Kershaw]

To revisit this subject I have spent some time today experimenting with varying methods of processing scans off the Coolscan 9000 with older Kodak 'VC' negatives. I think I now have a better understanding of how to achieve a good result with ColorPerfect in certain circumstances where "off" colours are the initial impression. Judicious use of the Nikon supplied icc profile for negatives (in PhotoLine) seems to help prevent the 'CP' software from 'over-doing' colour and contrast, although I still find I want to make some curve adjustments in PhotoLine, even though the documentation at www.colorperfect.com warns against use of curves to make corrections.

Some of the information about this subject online can seem rather conflicting and I still haven't fully investigated the why or how Nikon machines deliver a particular image quality or characteristic - perhaps related to the scanner light source? My scans of Fuji Pro 400H / 160NS and the modern Portra films have mostly proved uneventful.

 

[Les Sarile]

BTW, have you scanned the same film on the Coolscan using Nikonscan for comparison?

 

[Tom Kershaw]

BTW, have you scanned the same film on the Coolscan using Nikonscan for comparison?

Not the colour negative film, I should probably do so. I have made comparison scans for Kodachrome.

 

[Les Sarile]

Not the colour negative film, I should probably do so. I have made comparison scans for Kodachrome.

I consider DSLR scans of slides and b&w to be relatively easy - except for excessive dust and scratch removal. Color negatives are definitely more challenging, time consuming.and requires a bit of post skill.

 

[Tom Kershaw]

I consider DSLR scans of slides and b&w to be relatively easy - except for excessive dust and scratch removal. Color negatives are definitely more challenging, time consuming.and requires a bit of post skill.

An interesting aspect of colour negatives is that one can produce several excellent files in a row with good conversions and then find one or two negatives that are very challenging. I suspect the slightly more modern emulsions have been built with scanning in mind from that start which may aid production. I should probably do a couple of RA-4 prints from the challenging negatives as a form of control and point of comparison. I can see why some prefer to work with E6 films....

 

[Tom Kershaw]

I'm not using the workflow described here, but it is where I got the idea to extract the .icc file from NikonScan to use with system wide applications.
https://smashandgrabphoto.wordpress...lm-with-consistent-colour-using-icc-profiles/ - the file needs to be opened in a Hex editor in order to make the changes decribed in the article.

 

[Tom Kershaw]

Here is a scan almost straight out of the NikonScan software, apart from a slight curve adjustment in PL.

 

[Tom Kershaw]

Here is another example from the same roll of film. NikonScan with slight curves adjustment in PL as before.

 

[PhilBurton]

I'm not using the workflow described here, but it is where I got the idea to extract the .icc file from NikonScan to use with system wide applications.
https://smashandgrabphoto.wordpress...lm-with-consistent-colour-using-icc-profiles/ - the file needs to be opened in a Hex editor in order to make the changes decribed in the article.

Out of curiosity, can this same approach be used to extract .icc files for slide films? Also are these .icc files designed to correct speifically for the light sources for Nikon Coolscan units, and may not apply exactly to say Epson scanners? Could any of these profiles be transferred to scanner software like Silverfast or Vuescan?

I will freely admit that I am a complete noob about the details of color management?

 

[Tom Kershaw]

Out of curiosity, can this same approach be used to extract .icc files for slide films? Also are these .icc files designed to correct speifically for the light sources for Nikon Coolscan units, and may not apply exactly to say Epson scanners? Could any of these profiles be transferred to scanner software like Silverfast or Vuescan?

I will freely admit that I am a complete noob about the details of color management?

I've only tried the negative profile but as far as I'm aware it should also be possible to extract the profile .icc for slide films. I think these profiles are designed specifically for the Nikon scanners and the one I've used does seem to clear up a few issues, although the dialogue box I have shown abve is from PhotoLine (or Photoshop) - in this case the scan is arriving in the software (PL/CP) as a raw tiff - with the film type set to 'colour negative' in VueScan's film type setting. There are many potential combinations, to the extent that I would be weary of making broad recommendations to anyone else.

The NikonScan software seems to be using a distinct approach and doesn't allow for working in ProPhoto RGB as far as I can tell, so the files from scans I did this morning shown above started off as 'Nikon Adobe RGB 1998' and I converted to sRGB just before saving as jpegs.

 

[Les Sarile]

Here is a scan almost straight out of the NikonScan software, apart from a slight curve adjustment in PL.

Definitely a more balanced looking color/contrast result which is what I expect from the Coolscan + Nikonscan.

 

[PhilBurton]

I've only tried the negative profile but as far as I'm aware it should also be possible to extract the profile .icc for slide films. I think these profiles are designed specifically for the Nikon scanners and the one I've used does seem to clear up a few issues, although the dialogue box I have shown abve is from PhotoLine (or Photoshop) - in this case the scan is arriving in the software (PL/CP) as a raw tiff - with the film type set to 'colour negative' in VueScan's film type setting. There are many potential combinations, to the extent that I would be weary of making broad recommendations to anyone else.

The NikonScan software seems to be using a distinct approach and doesn't allow for working in ProPhoto RGB as far as I can tell, so the files from scans I did this morning shown above started off as 'Nikon Adobe RGB 1998' and I converted to sRGB just before saving as jpegs.

Tom,

I'm a bit lost here, so let me recap what I you have written and please correct me. You extract color negative .icc files from Nikonscan. You scan various color negatives with Nikonscan, and in NikonScan you select the correct color film type. You then import these .icc files into VueScan or PhotoLine or Photoshop, with the outcome that imported scans have colors are better than they would have been had you imported those same scans but without using those .icc profiles. What is "CP" software?

Thanks,

Phil Burton

 

[Tom Kershaw]

Hi Phil,

I'll try and explain myself better; there are really two situations here that I've explored. 'CP' is ColorPerfect - www.colorperfect.com

The preamble is that I've had trouble with scanning the older Kodak film stocks of VC/NC within my usual (VueScan, PhotoLine, ColorPerfect) workflow whereas the Fuji products, e.g Pro 400H and the current line of Kodak Portra films don't give such issues.

1. Scan with NikonScan, which outputs a "finished" tiff file with inversions and gamma applied. - this is what I posted most recently.

2. Scan with VueScan, create a linear (gamma = 1.0) scan, saved as a tiff. - this file goes into ColorPerfect via PhotoLine. In PhotoLine it is possible to assign a profile to the image extracted from NikonScan, i.e which I'm doing prior to the ColorPerfect inversion.

colors are better than they would have been had you imported those same scans but without using those .icc profiles.

- to an extent this is my next question - what impact does the assignment of a profile have on a known "good" tiff file / raw scan file from a negative film type that I know works well in my usual workflow.

If you are scanning E6 transparencies I suspect you may face far fewer issues as there is no orange mask or inversion to be concerned with.

Tom

 

[Tom Kershaw]

Here is an example of the 160VC-2 negative going through VueScan/PhotoLine/ColorPerfect with the Nikon profile applied.
https://tomkershaw.com/apug/160VC-2_curve_comparison_b.pdf