Scanning film with digital camera

[xkaes]

I would very much like to find something suitable, and if anyone has a link to such a bellows contraption, it would be greatly appreciated!

Here's around 100 of them starting at around $10 -- not all will work with the Nikon mount, of course.

https://www.ebay.com/sch/i.html?_from=R40&_trksid=p2334524.m570.l1313&_nkw=%28vivtar%2Csoligor%2Cspiratone%29+bellows+&_sacat=31388&LH_TitleDesc=0&_odkw=sony+%28a850%2Ca900%29&_osacat=31388&_sop=15&LH_PrefLoc=2

 

[George Mann]

He is using an APS-C camera body, which limits his lens choice to Nikkors 40mm, which is specifically made for this size of sensor.

I found out the hard way that no other lens will work with this sensor type.

 

[Bronson Dugnutt]

I would very much like to find something suitable, and if anyone has a link to such a bellows contraption, it would be greatly appreciated!

The manual for the Nikon PB-4 bellows contains some useful charts for magnification and working distance with various lenses mounted forward or reversed. See Table 1 on pages 6-7 and Table 2 on 10-11. Still a very useful reference even with a 'fixed' bellows aka extension tube.

The D3200 being an APS-C camera means you also gain a 1.5 crop factor worth of 'magnification'. I sometimes use a 1" sensor to digitize 110 and smaller negatives. Using large sensors to digitize small negatives introduces fickle working distances and is probably better described as a special case of digitization, with more considerations akin to enlarging/printing.

 

[runswithsizzers]

So, I already have a camera, a Nikon D3200 with a AFS Nikkor 18-55 3.5-5.6 G lens, the regular one that comes with this camera. I would like to know if this lens will work and if yes, what sort of extensions I would need (if any), or perhaps a different lens would be needed? I also have a Nikkor Micro 55/2.8, ...

Of those two lenses, the Nikkor Micro 55/2.8 is, by far, better suited to your purpose. If the Micro lens works, I would forget about trying to use the zoom for copying film.

If you wanted to copy only 135 and/or 120 film, then extension tubes would be practical -- but for the wide range of formats you want to copy, the extra adjustabilty that you will get from bellows probably makes that the better option. Using an APS-C sensor camera, I've copied a fair amount of 135 and 120 film using both 75mm and 100 mm copy/enlarging lenses on bellows with good results. But I don't know what to expect if trying to use a 55mm lens with the smaller formats. Seems like it ought to work, but haven't tried it. Good luck with your project!

 

[Helge]

Do not be so dismissive. Color balancing is a problem in 99% of camera scans I see online, even in minuscule web sizes. To me this means that before resolution is even mentioned in a thread like this one, beginners need to get advice on how to get good color. Even with slides. I have tried several camera manufacturers, and the default happy path doesn't work for film scanning: Adobe Color or default camera profiles / picture modes produce wonky color. Tools like NLP help but they introduce inconsistency. I would even say that for a beginner with average image editing skills there is no simple and reliable solution to get good color out of a digital camera. In other words, when I look at the difficulty of scanning-related tasks, I assign the 80% weight to color balancing, 10% to resolution, and 10% to workflow convenience. And it's not as simple as choosing a good light source. It's that, plus dealing with camera color profiles, color filter arrays, and having above average image editing skills.

B&W is an exception, sure.

What are your examples of good light sources?
There is a reason dichroic heads is/was so popular with colour negative film.
For slide you basically need something that is as close to daylight as possible, diffused sunshine daylight being a good option.

 

[koraks]

There is a reason dichroic heads is/was so popular with colour negative film.

Yeah. RA4 paper.

 

[MattKing]

I have two Olympus OM bellows units with slide copying units.
The challenge for me is finding a macro or enlarger lens that will fit the Zuiko OM lens mount built into it. My OM 50mm macro lens is too long when I put either a M 4/3 or crop sensor Canon body on the camera side.
I already have adapters that work with the cameras.

 

[Helge]

Yeah. RA4 paper.

Thought you’d be the last person to disagree?

Why RGB LEDs suck for a color RA4 enlarger – Koraks tinkers

tinker.koraks.nl

BTW factoid is a wrong fact.

 

[koraks]

Thought you’d be the last person to disagree?

If only things were so simple. But I'm currently not interested in creating another diatribe. Looks like we have enough for those going on already. So say I'm agreeing or disagreeing to whatever you say, whichever way you fancy. Floor is yours.

 

[Helge]

If only things were so simple. But I'm currently not interested in creating another diatribe. Looks like we have enough for those going on already. So say I'm agreeing or disagreeing to whatever you say, whichever way you fancy. Floor is yours.

Don’t be that way. I’m very interested in what you have to say on the matter. Obviously you have researched this more. And it is very pertinent to the topic of the OP.

 

[Steven Lee]

Steven, this scanner was a very good tool, and to some extent still is. But I cannot use it with Silverfast software, which is what I used to in the past and still want to use for the foreseeable future. I am using currently Vuescan Pro, which is all right, but does not offer the multitude of options, as Silverfast does.

To make things more interesting, the only version I've got for this Plustek 7200 is Silverfast software is 6.0, which cannot be used with Windows 10. I would need to buy a more recent version of the software, which is almost an arm and a leg. I know, there are numerous other software solutions, some even free, but the Silverfast offered me what I needed and I stick with what works for me.

But switching to camera scanning also means buying some kind of color inverting software. There are 4 popular options that come to mind: Negative Lab Pro, Negmaster, and ColorPerfect. There are also two free options: grain2pixel and Negadoctor module in Darktable. None of them are good, unfortunately. By "good" I mean comparable to Silverfast with a supported scanner.

You can try manual inversion in Photoshop without any additional software, but it's quite time consuming and you need to find a good color profile for your camera.

BTW I am not bashing camera scanning, in fact I have switched to it myself. I am simply pointing out that you should be prepared to invest a lot of time into color inverting routine.

@Helge regarding your question above, the best light source I found was the "Pro" model of Negative Supply LED light source. But the difference vs cheaper LED panels or a flash is marginal. In terms of getting good color I've experienced bigger gains with trying different cameras (different color filter arrays?), camera profiles, and picking the right digitization exposure.

 

[koraks]

Don’t be that way. I’m very interested in what you have to say on the matter. Obviously you have researched this more. And it is very pertinent to the topic of the OP.

Alright, you're right and I'm sorry. I guess I'm kind of weary of this topic because it's actually pretty darn complex (more so than I can oversee, frankly) and because I have a feeling (justified, or not) that there's at least some conservatism that plays a significant role in what people say is 'best' in terms of a DSLR scanning setup.

A proper response would take a lot of space and I'd rather not throw this thread off track too much (for a change...) So instead of replying here, I've written a blog on it instead: https://tinker.koraks.nl/photograph...ht-sources-for-dslr-scanning-color-negatives/

I've limited the scope of that post to scanning color negatives (C41 and ECN2). Some or perhaps all of it might be extrapolated to color positive film, I'm sure, and it will apply (perhaps even more so) to maskless color negative film as well. But I only explicitly go into the case of C41, and since that's quite similar to ECN2, I know the principles should apply there as well.

I also don't explore in very much depth different light sources per se; for instance, I don't go into using white LEDs vs. some form of incandescent bulb. Instead, I discuss 'old-fashioned' filters such as the dichroic filters combined with a broad-spectrum light source vs. narrow peak light sources, in particular LEDs.

A further limitation is that the piece is theoretical. It's really a paper napkin kind of exploration, without any empirical work. I'm kind of tempted to undertake the latter, but frankly it's not really a priority for me, now.

Hope you enjoy; at least the piece reflects why my earlier response to the recommendation of using a dichroic head was somewhat evasive/doubtful. I don't think it's necessarily 'the best' choice - although there's no doubt in my mind that it works very well and that in practice, there's very little to be gained to throw it out of the window.

 

[Minox]

I realize now I should've mentioned that the film to be scanned with this digital contraption is 99% black and white, not colour. The C41 and E6 is usually handled by my local lab, with satisfactory results, although I do not shoot much of it anyways.

Thank you !

 

[koraks]

I realize now I should've mentioned that the film to be scanned with this digital contraption is 99% black and white, not colour.

Makes me extra happy I decided to not post at length here, but instead do a blog on it

 

[Minox]

Makes me extra happy I decided to not post at length here, but instead do a blog on it

Any knowledge which I do not possess is more than welcomed. Even if it does not directly relates to things I envisage doing in the future. There is never such a moment when one stops learning.

 

[L Gebhardt]

I realize now I should've mentioned that the film to be scanned with this digital contraption is 99% black and white, not colour. The C41 and E6 is usually handled by my local lab, with satisfactory results, although I do not shoot much of it anyways.

Thank you !

If that's the case everything gets much simpler. There could still be an advantage to using a daylight balanced light source since that's what the sensor in the camera is balanced for, but you're even less likely to see any difference if you give enough exposure.

One more thing to consider is the need for the film plane to be parallel to the sensor plane goes up the more you magnify. I use laser alignment tool with a mirror pressed to the lens to verify this when not using a bellows. Don't let the laser hit the sensor. I don't know if that's sure to cause damage, but I've heard cameras have been damaged at events like concerts that use lasers (how the human eyes survive is another question).

 

[Helge]

Alright, you're right and I'm sorry. I guess I'm kind of weary of this topic because it's actually pretty darn complex (more so than I can oversee, frankly) and because I have a feeling (justified, or not) that there's at least some conservatism that plays a significant role in what people say is 'best' in terms of a DSLR scanning setup.

A proper response would take a lot of space and I'd rather not throw this thread off track too much (for a change...) So instead of replying here, I've written a blog on it instead: https://tinker.koraks.nl/photograph...ht-sources-for-dslr-scanning-color-negatives/

I've limited the scope of that post to scanning color negatives (C41 and ECN2). Some or perhaps all of it might be extrapolated to color positive film, I'm sure, and it will apply (perhaps even more so) to maskless color negative film as well. But I only explicitly go into the case of C41, and since that's quite similar to ECN2, I know the principles should apply there as well.

I also don't explore in very much depth different light sources per se; for instance, I don't go into using white LEDs vs. some form of incandescent bulb. Instead, I discuss 'old-fashioned' filters such as the dichroic filters combined with a broad-spectrum light source vs. narrow peak light sources, in particular LEDs.

A further limitation is that the piece is theoretical. It's really a paper napkin kind of exploration, without any empirical work. I'm kind of tempted to undertake the latter, but frankly it's not really a priority for me, now.

Hope you enjoy; at least the piece reflects why my earlier response to the recommendation of using a dichroic head was somewhat evasive/doubtful. I don't think it's necessarily 'the best' choice - although there's no doubt in my mind that it works very well and that in practice, there's very little to be gained to throw it out of the window.

Wow, thank you!
That will surely become an oft referenced article on here, and in general.
Good work in putting that together in under 24 hours.

 

[koraks]

Thanks, I appreciate it I just fixed some problems with illustrations that were lost in a copy-paste exercise, but all should work now.

Good work in putting that together in under 24 hours.

Oh, it's been brewing up there for months, or probably years! Thanks for the kick in the pants to make me sit down and actually write the damn thing. I need that sometimes.

 

[brbo]

A proper response would take a lot of space and I'd rather not throw this thread off track too much (for a change...) So instead of replying here, I've written a blog on it instead: https://tinker.koraks.nl/photograph...ht-sources-for-dslr-scanning-color-negatives/

Nice blog. I take scanning quite seriously and have done multiple tests to verify various "paper napkin ideas" in the past. Also some that include using cameras for scanning.

Maybe this will be of interest to you or somebody here...

Unfiltered tungsten light (from subtractive dichroic enlarger) vs. filtered light (dichroic filters set to cancel out the mask):



Camera is Olympus M5 II with pixelshift. Raw files are developed with dcraw to get the linear tiff file (dcraw -v -w -H 0 -o 0 -q 3 -4 -T rawfile.raw) without any Adobe nonsense getting in the way. Inverted, applied Adobe RGB and bumped saturation and vibrance in PS (+25).

You can see that unfiltered light loses a bit of saturation and separation with some hues, but it's not a huge deal.

Now, I also have an additive dichroic enlarger which produces three peaks of R, G and B light so in theory camera sensor colour filter array should have less colour crosstalk (better colour accuracy - higher saturation):



This file only needed +10 in saturation and vibrancy bump to make a suitable starting base.

Camera sensors with colour filter arrays have a lot of crosstalk (much more than typical triple line scanner sensor). You lose colour fidelity even with the best illumination. Any further compromises in light quality, bad exposure, "funky" raw developers... and the results become bad or at least unpredictable.

The best thing would be to have a monochrome sensor which you would expose three times with narrow R, G, B peaks and then assemble that data into colour scan. Unfortunately, at them moment I can't find the files where I simulated that with my digital camera. I took three separate exposures with R, G and B light of my additive enlarger light source and then binned all channels into one. For example, the exposure with green light would have most of the data recorded in G channel, but some would bleed into B and R channel. I took that B and R data and added it to G channel. You repeat that for the other two shots and you have a monochrome sensor. Unfortunately, I can't afford a camera with monochrome sensor or have this one converted to monochrome...

What I do have is some BW film, so I used that in place of a monochrome sensor.

Three exposures on Rollei RPX 25 with R, G and B light, developed, scanned as BW and then assembled into one file:



(link to raw files)

 

[koraks]

@brbo, this is excellent, thanks. I struggle to see the difference between dichroic filtered and unfiltered light, and insofar as I see them, they are partly due to very subtle contrast differences (the area under the chin of the woman looking up, for instance). Maybe in the pavement I can see with some good will a difference in actual hues.

The difference with the additive light source is much more pronounced. In part there is the influence of a different red balance, but this isn't linear, so there's a real difference here. Interestingly, skin tones are warmer (redder) while the white of the car is cooler (more cyan) in the additive filtered image. There's of course always the chance of some crossover emerging in the digital domain; I know that the way I do these adjustments, it's easy to inadvertently include a significant crossover in a channel.

The film-based color separation thing is really creative and I'm surprised at how well it has turned out given the limited gamma you have to work with in the original. The combined grain + dye clouds does very funny things indeed when viewed at 100%!

Good job on matching those scans so closely btw; I don't think I could actually pull that off.

My personal conclusion from this comparison is that the difference is slight at best, so I stick with the things I said on this at the start of my blog.

The best thing would be to have a monochrome sensor which you would expose three times with narrow R, G, B peaks and then assemble that data into colour scan.

Yes, that would be an excellent approach. More work than a single image, of course, but a very good idea.

 

[JerseyDoug]

But switching to camera scanning also means buying some kind of color inverting software. There are 4 popular options that come to mind: Negative Lab Pro, Negmaster, and ColorPerfect. There are also two free options: grain2pixel and Negadoctor module in Darktable. None of them are good, unfortunately. By "good" I mean comparable to Silverfast with a supported scanner.

There is at least one other option. If someone is switching to digital camera scanning from a flatbed or dedicated film scanner they might already own a copy of Viewscan. I have found that Viewscan works very well for inverting the color of my digital camera scans of color negatives. The relevant Viewscan settings are Task: Scan to file, Source: File, Files: <digital camera scans>, Mode: Transparency, Media: Color Negative. The other Viewscan settings are just as if I was scanning the negatives themselves with my Plustek or Epson scanner.

[EDIT: I have the White Balance on the Fuji X-T20 set to Auto and the Viewscan Color/Color Balance set to Auto Levels.]

Viewscan supports batch scanning so the entire operation for a roll of film takes fewer than half a dozen mouse clicks, and it runs very fast on an M1 MacBook Air.

It has been years since I used Silverfast. I do not remember if it had this ability.

 

[MattKing]

For clarity when considering JerseyDoug’s post - it is “Vuescan”, not Viewscan.
And you will need to pay for the Professional version.

 

[JerseyDoug]

For clarity when considering JerseyDoug’s post - it is “Vuescan”, not Viewscan.
And you will need to pay for the Professional version.

Oops! Thanks for cleaning that up Matt. I don't know that I would particularly recommend buying Vuescan for this purpose because there are other good options. But if someone, like me, is switching from a Plustek, Epson, etc. and already owns a copy of Vuescan I think it might be worth giving it a try.

 

[Steven Lee]

@brbo kudos for the effort you've put into these experiments! Do you mind if I ask a couple of questions, since you clearly thought about this problem deeply:

I do have an old digital camera to convert into a B&W without a Bayer grid. How would you go about producing a clean R/G/B light for 3 exposures onto a mono sensor without an enlarger? Looking up sensitivity peaks in C41 film datasheets I see wavelength in nm, but I haven't any seen color filters with nm in their specs. Where should I be looking at?

 

[koraks]

Looking up sensitivity peaks in C41 film datasheets I see wavelength in nm, but I haven't any seen color filters with nm in their specs. Where should I be looking at?

Have you looked at the blog post I linked to above? Not to shamelessly promote my own writings of today, but it does adress more or less exactly what you're asking, or at least the most crucial part of it. https://tinker.koraks.nl/photograph...ht-sources-for-dslr-scanning-color-negatives/

If you're looking for an easy solution, I'd consider any R G B controllable LED system. The wavelengths will not be precisely perfect, but they will be quite close. There are all manner of color-controllable bulbs, led strips etc. that you could fit a diffusor over and then dial in the desired color for each exposure.

A more optimized solution would comprise selecting the optimal wavelengths and purchasing some LEDs, drivers and some sort of a controller. There are countless ways of doing this, of varying complexity, but they all require some knowledge of electronics.