DSLR Film Scanning

[Stevo711]

Can anyone out there help me? I keep reading that I need a macro lens with a 1:1 magnification ratio if I want to scan negatives with a DSLR. I have a Canon 24-70 2.8L Macro lens with a magnification ratio of 1:3.45. Since the 4x5 negative is so much larger than a 35mm negative, why would I need a 1:1 dedicated macro lens? Since I can fill the frame with a focused 4x5 negative, what else could I hope for? Isn't my 24-70 macro more than enough to get great detail out of a whopping 4x5 negative?

[lenicolas]

Your lens would be fine, it’s just that you’d get a bigger file if you could get closer and stitch several shots.

Unless your dslr has very low resolution, 1:1 is overkill for 4x5.
When “scanning” 4x5 I only get so close that the short side of the negative fills the long side of of my dslr, and then I stitch 4 shots to get the length of the negative. 106mm/36mm that’s roughly a 1:3 ratio, and with a 45mp dslr the resulting image is a little over 8.000x10.000 ; plenty of resolution for any print size.

1:1 would produce huge files that would be slower to process, and introduce problems with depth of field and film flatness.

[Stevo711]

Your lens would be fine, it’s just that you’d get a bigger file if you could get closer and stitch several shots.

Unless your dslr has very low resolution, 1:1 is overkill for 4x5.
When “scanning” 4x5 I only get so close that the short side of the negative fills the long side of of my dslr, and then I stitch 4 shots to get the length of the negative. 106mm/36mm that’s roughly a 1:3 ratio, and with a 45mp dslr the resulting image is a little over 8.000x10.000 ; plenty of resolution for any print size.

1:1 would produce huge files that would be slower to process, and introduce problems with depth of field and film flatness.

Oh my God! You have no idea how much time I have spent on the internet researching trying to get an answer! Thank you soooo much! I just joined this forum last night. Not sure exactly how everything here works so I hope you get this reply. Wish I thought of it sooner. Happy New Year my new friend!

[Stevo711]

Oh my God! You have no idea how much time I have spent on the internet researching trying to get an answer! Thank you soooo much! I just joined this forum last night. Not sure exactly how everything here works so I hope you get this reply. Wish I thought of it sooner. Happy New Year my new friend!

[Willie]

Scanning? Use a scanner.

You are just looking at normal Copy work, same as has been done for decades.

[CreationBear]

Isn't my 24-70 macro more than enough to get great detail out of a whopping 4x5 negative?

As always, a lot will depend on your personal needs and infrastructure.:) If your short zoom gives you files you're happy with, excellent--however, many people make a distinction between lenses that have a short minimum focusing distance and "true" macro lenses that are designed to be sharp across the frame. (I.e., the reproduction ratio--whether 1:1 or occasionally 1:2--won't be as important as whether the lens provides a sharp image of a flat object from corner to corner.) It will probably come down to your preferred kind of output whether you need an optimized lens: if you're mostly sharing images on the Web, probably not; if you're printing large, you might be introducing slop that negates the potentialities of a well-exposed and focussed 4x5 negative.

On the upside, in the Pentax ecosystem at least, it's often possible to find legacy "true macro" lenses that work well for the kind of copy work we're talking about for not a lot of money--Canikon might be another story!

[neil poulsen]

Versus a zoom 24-70mm lens, all be it "macro", I would think that a 35mm, fixed focal length, macro lens would be preferable for DSLR scanning. A zoom lens is prone to chromatic aberration and barrel/pincushion distortion that could negatively impact the scan.

For example, I have a Sigma 100mm macro Canon mount lens that could be used. This lens does not have internal focus. So focusing this lens at 1:1 extends the barrel of the lens a substantial distance.

As an comment, I would never purchase a lens with internal focus, since depending on the magnification, focusing can alter the actual focal length of the lens.

[sperdynamite]

The advantage of getting the 1:1 macro lens is that it'll be meant for doing this kind of work and thus be well suited to the task. A 24-70 zoom even at a middle aperture will IMHO be not high performance enough.

A very popular lens for scanning is the 70mm ART macro from Sigma. It's wildly sharp, not terribly expensive, and will all you to fill the frame with a piece of film from 35mm to 8x10 on a copy stand. Stitching is always an option but IMHO that's sort of like ordering a 60" proof. You can always re-scan the film at higher resolution later. It does help that I use a pixel shift camera which creates a very large file in one bite.

[CreationBear]

Stitching is always an option but IMHO that's sort of like ordering a 60" proof.

Excellent, I was meaning to ask you: IIRC, you were using an Panasonic S1 (~24 MP with pixel shift) for a while--if you did a one-shot "scan" of a 5x7 negative with it, how big of a print do you think the file would support?

At any rate, my working theory is that 5x7 is the new sweet spot for a hybrid workflow--hopefully you're finding the same.:)

[grat]

The 24-70 f/2.8 L is a nice lens, but it does suffer from a small amount of distortion. Compare with the 100mm f/2.8 macro:

https://www.the-digital-picture.com/Reviews/Lens-Distortion.aspx?Lens=787&Camera=453&FLIComp=0&LensComp=107&CameraComp=453&FLI=4

Since you're using a Canon lens, I'm assuming you'll be using a Canon DSLR, which means you have access to DPP4-- which can do a very nice job of correcting for deficiencies in the lens (distortion, aberration, etc.), which you'll want even if you're just doing a single shot. If you're doing multiple shots and stitching them together, you'll NEED some form of correction.

[Peter De Smidt]

Always try what you have. That way you have a baseline of quality, and you can judge if it's good enough for you. I did a lot of scanning at 1x....but that was 7 years ago with a low res camera. The only reason to scan at higher magnification than whatever you have is if you need more resolution. There definitely is a law of diminishing returns. For your use, an old macro lens in the 50-75mm range would likely be the best bang for your buck.

[sanking]

Excellent, I was meaning to ask you: IIRC, you were using an Panasonic S1 (~24 MP with pixel shift) for a while--if you did a one-shot "scan" of a 5x7 negative with it, how big of a print do you think the file would support?

At any rate, my working theory is that 5x7 is the new sweet spot for a hybrid workflow--hopefully you're finding the same.:)

I have been digitizing my archive of 5X7 negatives since last December with an a Sony a7r iv with 16 shot pixel shifting. This gives an image file size of about 19000 X 12000 pixel at 350 ppi with one shot. The resolution with this approach is about 50 lpi, which is about what you can get with an Epson V750 or V850, and this allows a vey large print of about 50X35" at 350 ppi from the 5X7 negative. The results don't quite match up to a drum scan, but for this format it is plenty good enough.


With three-stitch using 16 shot pixel shifting I get file size of about 28000 X 19000 pixel, which allows print size of about 80 x 55" at 350 ppi. Unfortunately the work flow with three or more shots with 16 shot pixel shifting is rather challenging as the processing takes a long time in computer due to the large file sizes, but in image quality it beats my drum scanner, and takes a lot less time.

Sandy

[CreationBear]

This gives an image file size of about 19000 X 12000 pixel at 350 ppi with one shot.


Excellent, I definitely won't come anywhere near those limits--though on the processing end, the new Apple Silicon chips (M1 and upcoming M1x) start to make a lot of sense.:)

[sperdynamite]

Excellent, I was meaning to ask you: IIRC, you were using an Panasonic S1 (~24 MP with pixel shift) for a while--if you did a one-shot "scan" of a 5x7 negative with it, how big of a print do you think the file would support?

At any rate, my working theory is that 5x7 is the new sweet spot for a hybrid workflow--hopefully you're finding the same.:)

I would defer to Sandy re his scanning advice generally but I use a S1R, 47mp or 189 with pixel shift. I usually downsample my pixel shifted scans to 100mp as I think it cleans up the file a bit and doesn’t ask so much of the film or the lens. Given that I’ve never in all my years made a print larger than 30” in the long dimension, I’m doing just fine. As Sandy said, stitching is time consuming and can be difficult.

I like pixel shift as much for the resolution as for the fact that it vastly decreases noise. I’ve recovered shadows from chromes to a remarkable degree. For my 100mp usage you could tell me that a drum scanner is better but I’d be hard pressed to imagine how it could be.

[CreationBear]

I usually downsample my pixel shifted scans to 100mp as I think it cleans up the file a bit and doesn’t ask so much of the film or the lens.


Excellent, that's a useful piece of advice I think.

[sanking]

I would differ to Sandy re his scanning advice generally but I use a S1R, 47mp or 189 with pixel shift. I usually downsample my pixel shifted scans to 100mp as I think it cleans up the file a bit and doesn’t ask so much of the film or the lens. Given that I’ve never in all my years made a print larger than 30” in the long dimension, I’m doing just fine. As Sandy said, stitching is time consuming and can be difficult.

I like pixel shift as much for the resolution as for the fact that it vastly decreases noise. I’ve recovered shadows from chromes to a remarkable degree. For my 100mp usage you could tell me that a drum scanner is better but I’d be hard pressed to imagine how it could be.

My comments comparing the high resolution pixel shift files with drum scanner is not about resolution, but what I would call micro-contrast, especially noticeable in the low tonal values. Drum scanners are superb at this because there is very little flare in the system, virtually none in fact. Early CCD scanners, and ones like the prosumer Epson, by contrast, allow a lot of flare.

Flare is also a big challenge with DSLR scanning. On the other hand, the dynamic range of modern digital camera sensors is incredible, and with careful baffling to control flare they can give great results in digitizing film.


Sandy

[grat]

Excellent, I definitely won't come anywhere near those limits--though on the processing end, the new Apple Silicon chips (M1 and upcoming M1x) start to make a lot of sense.:)

Memory and high speed disk are going to be more important than CPU for a task like stitching very large images together.

[CreationBear]

Memory

No doubt I'll never be considered a "power user" but I'll be interested if there's a consensus that emerges regarding the "sweet spot" in terms of RAM and SSD capacity in the new Apple machines for the kind of processing we're discussing. (Of course, I'm sure Apple will charge an extortionate amount for each...:))

[grat]

Not sure-- I've got an AMD 8/16 core system with 16gb of RAM and my working disk is a 500gb NVMe SSD. It does pretty well on high-resolution stitches of medium format, but I haven't thrown a 7000 PPI 4x5 stitch at it yet-- which, I admit, is ludicrous.

Most of the major operations (all under Affinity Photo-- I haven't used Lightroom or Photoshop since the days of dual-core systems) utilized all 16 cores, and I suspect that greatly sped up operations. Memory consumption for 12 images at 30MP each (I said it was ludicrous) was hitting 12-15GB while stitching, framing, inverting, etc.

FYI, stitching took 15-20 seconds, and rendering the final image took about 35 seconds, so just under a minute to auto-stitch all 12 images.

The image history for the medium format image got pretty ugly, and I started swapping if I was doing full frame copies instead of snapshots.

I have had an out-of-memory error with 6400 PPI scans under Epson Scan 2, though (That was Epson scan itself).

I would suspect 32G is probably a bit overkill for most situations, but 16G might be a little small. Since it's very difficult to get dual or quad channel with 24G, I'll probably upgrade to 32G the next chance I get.

[Peter De Smidt]

My laptop has 32 gig of ram and decent ssd drives. Those make the biggest difference in my experience. It's an Eluktronics RP-17. It's not fancy, but it's fast and affordable.

[GRAYnomad]

I'm setting up a "scanning" rig at present.

I have an old 50D (20mpx) that I will dedicate to this task and so far I've been scanning 6x12cm negs. With the long side of the frame across the short side of the neg I take 4 images and stitch them together. That's nowhere near 1:1 but plenty for most purposes. The end result is a roughly 9000x4500 px image or approx 40Mpx.

I plan to start on my 5x4 as well, and with them I will do two passes at the same magnification, so a total of 8 images.

So far it's working pretty well, yes it's a lot of faffing around but I'm slowly getting the workflow sorted and let's face it, with LF we don't shoot 1000s of photos. Anyway it's better than spending a $1000 or more on a flat bed scanner and the results are at least as good, possibly better. So far I've spent about $10.

There is a gotcha with 120 film though that caused me a heap of anguish for a while with the stitching software :D

Oh, and yes I use a proper macro lens.

[Peter De Smidt]

Sounds like you have a very good system, Rob.

[Michael E]

I have been using a DSLR system (Pentax 645Z with 120mm Macro) for black and white for a while. The other day, I tried it for the first time on color negatives and could not get decent colors in Adobe Camera Raw. I tried shooting with and without color correction filters, white balance by eyedropper and manually. The colors were all one brown-ish soup. Do you have any tips for me?

Best,

Michael

[Peter De Smidt]

Make sure that the profile setting in Adobe Camera Raw is "neutral" or "camera neutral."

But you probably want to check out Negative Lab Pro.

[urnem57]

I am curious about other people’s experience with negative holders. I bought two different one thinking I could get away without paying for one from negative supply. The one held together with magnets came apart and scratched a negative. Same for the plastic one. I am waiting on my order from NS. Buy cheap, buy twice. True for me in this case.

[Peter De Smidt]

My dslr scanning was mainly with sheet film. The negative sits on textured acrylic or glass, and I route a a piece of 1/4" thick black ABS with a proper sized hole. I'm make sure that all edges are sanded smooth. I painter's tape the neg in one or two spots to the frame, and then set it down on the glass. No scratches. For roll film, I scan with an acrylic cover sheet on a Screen Cezanne. I would much rather have a slow system than one that leads to scratches.

[GRAYnomad]

Sounds like you have a very good system, Rob.

I think it will be. Just a hodge podge of lights and tripods for now but I will make a proper stand, neg holders, and light source soon.

[GRAYnomad]

I have been using a DSLR system (Pentax 645Z with 120mm Macro) for black and white for a while. The other day, I tried it for the first time on color negatives and could not get decent colors in Adobe Camera Raw. I tried shooting with and without color correction filters, white balance by eyedropper and manually. The colors were all one brown-ish soup. Do you have any tips for me?

Best,

Michael

I am crap with colour I admit, probably why most of my film work was B&W :D

But here is what I do and it seems to work pretty well.


Scan with some of the blank film base included.
Bring into PS (well I use Affinity but should be the same).
Add a layer.
Use the eye dropper to sample the orange base.
Fill the new layer with that colour.
Put new layer below the scan layer and combine them using the "subtract" option (some other options seemed to do the same as well).
Adjust brightness etc. Colour should be OK but also need some work.


I've only done a few and with landscapes you can be a bit out colour wise and not notice, bit it seems pretty good.

Vuescan also works well, but you have to buy the pro version to get the "import file" feature. I could still buy it but I'm a real tight-arse so will see how the above goes first.

EDIT: Modified the steps and tried with an old version of PS, seemed to work on that as well.

[GRAYnomad]

I am curious about other people’s experience with negative holders. I bought two different one thinking I could get away without paying for one from negative supply. The one held together with magnets came apart and scratched a negative. Same for the plastic one. I am waiting on my order from NS. Buy cheap, buy twice. True for me in this case.

I'm making my own.


Camera
-
Clear acrylic, thin, 1.5mm
Negative, emulsion up
Translucent acrylic with a slightly stippled surface facing up
-
LED light source



There is also a mask of black tape on the bottom acrylic to cut down flare and locate the film. The two acrylic sheets are hinged together with tape.

The "LED light" is actually a 300x1200mm flouro replacement unit out of an office refurb, it's rated at 6000k which I think is close enough for Government work. Way over size of course but the extra area is very useful as a light table for sorting the negs.

I've only used this a couple of times so far, but it works well, nice and flat with no Newton's rings.

As I take multiple shots of each neg I plan add some indexing so I don't have to eyeball the alignment of each shot.

[roscoetuff-Skip Mersereau]

Rob: Curious what "stitching software" you're using. I've never tried this, but sure sounds like a way to go - if there's a way to automate the process. I'm using a (now old) Nikon D750 with 24MP, and surely jealous of those with higher MP's and pixel shifts in their cameras. But neither a new DSLR or scanner are quite in this year's budget (already blown on a new 4X5). Appreciate your insights... and kudos to your photos. THanks!

[GRAYnomad]

Rob: Curious what "stitching software" you're using...

I've been using PTgui for about 17 years. Back then stitched panos were pretty new and the few other options available were useless. There are many more options now, in fact pano stitching is built in to most programs like ON1, PS etc. But there is one big difference between them and PTgui. That is that with PTgui you can manually apply control points, it's not normally required but for example if you have a photo with a lot of fuzzy water, or other areas with no real detail, I find that most programs struggle to lock onto what parts of each image should match. PTgui can struggle with this as well, but with it you just go in and create a few control points manually to give it a hand.

PTgui also allows a lot of control over the image placement, rotation of the pano, fixing of panos that point up or down and also format of the output. For example you can output to a layered PSD file where each image is on it's own layer with a mask. Thus if there are some funnies you can get in there and tweak things.

https://www.ptgui.com/

Oh, and the gotcha I mentioned above? If scanning roll film DON'T include the lettering on the edges. The pano software (all 3 I tried) sees "PORTRA" in one place and two frames away sees another "PORTRA" and thinks that they are matching parts of separate frames, the results are interesting if not that useful :D I was tearing my hair out about this and with the other two programs I could do nothing about it as there was no information supplied. Then I looked at the control points automatically produced in PTgui and saw dozens of them clustered around these letters and numbers and realised the problem.

This pic shows a heap of good control points on the rocks but some up on the lettering as well.

http://www.robgray.com/temp/PTgui-control-points.jpg

This actual example would be OK because they are adjoining frames, I couldn't find an example of a bad match, but you get the idea.

All that said it does seem to be a problem with using RAW files as much as the lettering, so I just convert them to TIFF and either don't shoot the lettering or crop it out before feeding the images into PTgui.

YMMV, but I would give it a try...I just had a look at their site, it costs US$154 these days, yikes. Panos (normal ones) are a big thing with me so I don't mind spending a few $, but that's a lot. Still, it's cheaper than buying a scanner :D


EDIT: The other programs I use are Affinity, ON1 and PS and to be honest they work just fine most of the time and you probably already own one of them. If not buy Affinity.

[roscoetuff-Skip Mersereau]

Rob: Thanks for all of that. Let me ask a follow up if I may. Are you using stitching to do panos (I think), or stitching to put together small sections of a DSLR scanned negative into one "scan" of the negative? The latter is more what I think I'm headed towards, and I can see how both processes are very similar. And frankly I'd hoped for an automated means of doing so. But now you make it clear that manual is probably better. Oh well.

[Peter De Smidt]

Well, you can have an automated way of doing it. If you have a precision movement system, then you can use a template (I hope this is the right word) to stitch perfectly in PTgui.

Otherwise, Lightroom can created a stitched pano raw file, which is terrific, if it can handle the stitching.

[grat]

Affinity is pretty good at automated stitching.

[GRAYnomad]

R...Are you using stitching to do panos (I think), or stitching to put together small sections of a DSLR scanned negative into one "scan" of the negative?...

Until last week I only used if for proper panos, but now I'm using it for these multiple DSLR scanned pics of negatives as well.

As for automating, If you are a wiz with embedded processors and stepper motors yes you could automate the process, I could, but I doubt it would be worth the trouble. One thing I plan to do is add indexing to the light table and neg carrier such that I can just slide the carrier along to align index marks and not have to view through the camera. It already only takes 10-20 seconds, but this will be easier.

NOTE: I just redid the stitching of a few negs with Affinity, it worked very well but if it does eventually stumble on a neg you have no control, so I would probably just use Affinity (or PS or LR or ON1) until there is a problem (if ever). If you don't already have Affinity it's worth a look as your main editor, I still have an old PS copy for a few things but hardly ever use it. That said I'm liking the ColourPerfect plugin I have on trial so if I buy that I'll need to keep my PS around just for the neg/pos conversion (I wish people would make programs stand-alone and not reliant on another program).

[grat]

There is *some* control over pano stitching in Affinity-- Once you load the images, and hit "Stitch", it opens in a new window, and I (usually) hit "Apply". But, if you use the icons to the left, you can do some alterations to the pano before the final render.

See https://affinity.help/photo/en-US.lproj/index.html?page=pages/Panorama/panorama_editing.html?title=Editing%20panoramas for more information.

[martiansea]

If you don't already have Affinity it's worth a look as your main editor, I still have an old PS copy for a few things but hardly ever use it. That said I'm liking the ColourPerfect plugin I have on trial so if I buy that I'll need to keep my PS around just for the neg/pos conversion (I wish people would make programs stand-alone and not reliant on another program).
I agree. I recently dumped all my Adobe apps in favor of the Affinity suite and DXO PhotoLab. Affinity handles all the "painterly" editing, compositing, adjusting, collaging, chopping, mutilating, ruining, coloring, etc... that I used to do with PS, and DXO PhotoLab is simply the best raw file processor/developer I've ever used. I tried all the ones I could find and nothing beat PhotoLab for sharpness and detail and noise reduction. Affinity's raw processing is decent, but wasn't even as good as CameraRAW (at least on Windows, I hear it's better in the Mac version).

I've gotten hooked on DSLR scanning, myself. Did this comparison between my Nikon D800 with Minolta 100mm AutoBellows Rokkor (using a plain lens mount adapter with no correction glass) and the Imacon FlexTite I got to use in grad school. The results were more than enough to convince me this is a good way to go. Never got results this good with any flatbed I tried.
213765
This is a crop from a 35mm negative.
Upcoming DIY project is to build a permanent light source with mask that allows for easily making consistent pano shots of the larger formats. I had tried getting the whole negative in one shot, but the results were disappointing. 1:1 of the 35mm frame really looks the best.

[PatrickMarq]

I am crap with colour I admit, probably why most of my film work was B&W :D

Put new layer below the scan layer and combine them using the "subtract" option (some other options seemed to do the same as well).

Adjust brightness etc. Colour should be OK but also need some work..
Be sure that the new layers has 80% transparent with the substraat.
as I’m also using Affinity , you can use the alt key with levels if you change first the white point to 1% . A strange bug ...

[GRAYnomad]

Be sure that the new layers has 80% transparent with the substraat.
as I’m also using Affinity , you can use the alt key with levels if you change first the white point to 1% . A strange bug ...

Thanks Patrick, I'll try that with the next one.

[GRAYnomad]

I agree. I recently dumped all my Adobe apps in favor of the Affinity suite and DXO PhotoLab. Affinity handles all the "painterly" editing, compositing, adjusting, collaging, chopping, mutilating, ruining, coloring, etc... that I used to do with PS, and DXO PhotoLab is simply the best raw file processor/developer I've ever used. I tried all the ones I could find and nothing beat PhotoLab for sharpness and detail and noise reduction. Affinity's raw processing is decent, but wasn't even as good as CameraRAW (at least on Windows, I hear it's better in the Mac version).

I've gotten hooked on DSLR scanning, myself. Did this comparison between my Nikon D800 with Minolta 100mm AutoBellows Rokkor (using a plain lens mount adapter with no correction glass) and the Imacon FlexTite I got to use in grad school. The results were more than enough to convince me this is a good way to go. Never got results this good with any flatbed I tried.
213765
This is a crop from a 35mm negative.
Upcoming DIY project is to build a permanent light source with mask that allows for easily making consistent pano shots of the larger formats. I had tried getting the whole negative in one shot, but the results were disappointing. 1:1 of the 35mm frame really looks the best.

Looks good. I wouldn't like to be a scanner manufacturer these days :D

Have you got any comparison images of exactly the same area?

[joelio]

Looks good. I wouldn't like to be a scanner manufacturer these days :D

Have you got any comparison images of exactly the same area?

Not too many scanner manufacturers left! And the companies facilitating camera scanning solutions? Staffing up!

[grat]

New version of Affinity has a "divide" function which does a nice job of removing color casts. Robin Whalley on Youtube has a tutorial up on it.

[GRAYnomad]

New version of Affinity has a "divide" function which does a nice job of removing color casts. Robin Whalley on Youtube has a tutorial up on it.

Probably one of the other options I tried that seemed to work.