Scanning 35mm Black and White Negatives with the D800E

I wrote up how I scan black and white negatives with a DSLR in Scanning 35mm Black and White Negatives with the D800E. I show the results compared to a drum scanner and a 4000dpi Canon CanoScan FS4000US.

The methods works equally well on slides, but I haven't been able to get the color negative process exactly where I would like it. Hopefully that will be next. I also think this could work well for medium format film if a proper film holder could be devised. I hope to work those details out too.

Let me know if things aren't clear, typos that need to be fixed, or any other comments. It's a pretty straight forward process, but I'm probably glossed over a few things that may not be clear to others.

Thanks for posting this. I feel that scanning using a DSLR may well be the way that most people do it before too long except for the most demanding applications where high-end scanners will probably always have a place.

At the lower end of the market it is already possible to buy devices that are really the "guts" of a cheap digital camera, a light source and a simple neg/slide holder and these things produce credible 4x6 scans for an outlay of a hundred bucks or so.

After all, at their basic level, the only difference between a scanner and a camera is the technique employed to capture the pixels - line by line or all at once. With increasing pixel density in DSLRs it seems to me to be perfectly logical to use a suitable camera, if available, for no other reason than the time advantage over scanners. The D800 is an outstanding example of such a camera.

I also think that reversal of negative images in external software rather than in a scanner is a more desirable path. There is a large body of dedicated scanners out there who do in fact scan their colour negs as if they were trannys because they believe that scanners, like digital cameras, are designed as direct positive devices and that reversal algorithms built into most scanners are a compromise at best that do not take into account the characteristics of different film types. Keep up your good work. OzJohn

Hi

L Gebhardt said:

I wrote up how I scan black and white negatives with a DSLR in
Let me know if things aren't clear, typos that need to be fixed, or any other comments. It's a pretty straight forward process, but I'm probably glossed over a few things that may not be clear to others.

Click to expand...

a neat and informative site. I'd like to ask how (if its there I missed it) you are able to set base points for the black and white points (where there is only fog and where density maxes out).

Did you mention that the demosaic of the RAW file needs to be done in a linear manner or you'll be applying curves by default (as that's what cameras do). I may have missed that too.

Also, the cost of the lenses and bellows for capture and lack of film holders makes this method perhaps unattractive when compared to a scanner. A V700 for instance can allow you to set and scan a whole roll of 35mm on a platern without needing to stand there. Probably the rapidity of shutter release is faster than the scan of a 35mm neg but what about film transport (acquisition transport)?

Also, as you scan progressively larger formats you are not getting the same DPI without stitching. So a capture of a 4x5 neg will of course still be the same number of pixels as a 35mm frame is.

But nice work and good to see people covering the alternatives.

Chris, thanks for the feedback. I'll address your points here and look at making a few modifications to the article. Let me know if I've missed your points.

pellicle said:

I'd like to ask how (if its there I missed it) you are able to set base points for the black and white points (where there is only fog and where density maxes out).

Click to expand...

I guess I should expand on this. By setting the exposure so that the histogram is moved towards the right you effectively normalize the film base, which should be the brightest portion of the frame. I think the part that I'm missing in the article is that I capture part of the non image area of the film. By including this and being sure not to clip it I lock in the black point (once reversed). The inverted white point will be dependent on the contrast of the negative and will vary from film to film. For this I think the simplest way is to adjust the Tone Curve by starting the upper left hand point farther to the right, which will stretch the negative's limited contrast range over the full image range.

Even if you don't expose to the right you could also set both black and white points with the Tone Curve, but you run the risk of more noise in your highlights.

pellicle said:

Did you mention that the demosaic of the RAW file needs to be done in a linear manner or you'll be applying curves by default (as that's what cameras do). I may have missed that too.

Click to expand...

My goal with this was really to get fast input for a lot of images. So I haven't experimented with linear demosaicing. My simple solution is to again adjust the Tone Curve with a simple S curve (inverted because I also invert using the Tone Curve). This restores a proper looking contrast curve to the image, and is good enough for proofing and LightRoom display. I think as long as the image is captured correctly in RAW format you could optimize an individual image with other techniques.

I haven't experimented with a linear conversion, but it makes sense that you could get a much more accurate representation of what's on the negative. But you would still need to adjust the final output curve by hand for most artwork, so I'm not sure there's likely to be an effective difference.

pellicle said:

Also, the cost of the lenses and bellows for capture and lack of film holders makes this method perhaps unattractive when compared to a scanner. A V700 for instance can allow you to set and scan a whole roll of 35mm on a platern without needing to stand there. Probably the rapidity of shutter release is faster than the scan of a 35mm neg but what about film transport (acquisition transport)?

Click to expand...

I have an Epson 4870 and I wasn't impressed with the software to batch capture. I found I sent a lot of time tweaking individual images if I set the black and white points wide enough to not clip, and if I set them correctly for one image the were invariably wrong for many others resulting in time needed for rescanning. But finally the 4870 doesn't scan as wide of an area as the 700 series Epsons. So it wouldn't have been as much of a labor savings. For me the cost of buying yet another scanner just to test it out wasn't as compelling as this method.

The lens I used was certainly not cheap, but I also have a Nikon 55 f/3.5 pre AI lens that works fine on the bellows (will damage the D800 if mounted). This worked pretty well, but the grain wasn't as sharp out to the corners. I paid $30 for this lens. So it can be done fairly cheaply. There are knock off bellows on ebay for dirt cheap as well. I bet it's possible t get a working solution for under $100, but I haven't tried. But you then need to rig up film holders as you say.

pellicle said:

Also, as you scan progressively larger formats you are not getting the same DPI without stitching. So a capture of a 4x5 neg will of course still be the same number of pixels as a 35mm frame is.

Click to expand...

Definitely an issue. I think the D800E would do well in single shot scanning up to a medium format (6x9 would fit the aspect ratio perfectly). Probably not close to the drum scanner, but probably about as good as the Epson lineup when you factor in their real resolution. Either way it would make good 16x24 prints, which is impressive. The hard part would be fabrication of a holder. I made a quick and dirty attempt at this with a negative carrier made from a 4x5 enlarger Speed Carrier (Carlwen, I think). Again 4x5 would work and should still make a decent 16x20 print, but larger would definitely need to better scanner. For me, I would just use the drum scanner, and probably will for anything that I plan on actually printing.

Thanks very much for all this, Larry.

Have you tried the old M2 (27.5mm) extension tube on the 55/3.5? It clears the Ai tab on anything I've put it on and takes the 55 down to life size.

CGW, thanks, that's good to know. I don't have the M2 tube, but the bellows lets me use it with the D800 as well. I also have an AIS version of the 55 that I can use on the camera.

I just picked up a set of the Fotodiox tubes and I think they will also allow the older lenses to be used. The are very rigid for the price - $10 or so shipped off eBay.

L Gebhardt said:

Chris, thanks for the feedback.

Click to expand...

no worries ... I didn't give it a full heavy duty read, so I just wanted to make sure I didn't miss things.

just quickly ...

I have an Epson 4870 and I wasn't impressed with the software to batch capture. I found I sent a lot of time tweaking

Click to expand...

did you read this?

in my view ...: bulk scanning with Epson flatbed

I found that if I selected an entire strip first I got a very good idea of the black and white points. By then cloning that out over the other strips I got really consistent results (as one would on a contact sheet) for the entire roll.

I scan in strips and crop out later, as it saves time in scanning too (no too and fro for each neg), but you can have presets saved for each selected thumbnail too. As long as you allow a little room for individual strip image placement you can move them in batches too (should you want to stay with thumbnails).

For me, I would just use the drum scanner, and probably will for anything that I plan on actually printing.

Click to expand...

well if you have a drum scanner, sure. Personally I found that for 4x5 there wasn't HUGE differences in drum scans vs 4870 except on the upper limits of enlargement AND if you don't tweak your Epson for focus, and treat your scans properly with sharpening.

back when I wrote this: http://cjeastwd.blogspot.com/2009/05/digital-vs-film-scans-screens-and.html

I was only using a 4990, but I am certain that my 4870 is a better scanner (product variation?) ... I should dig out the neg and see if I can compare that too. Have a look at the images in there, and from what I see the drum scans are more 'pixel / peper grain' than the Nikon LS4000 and the 4990 blurier than the Nikon LS4000. I really should get around to redoing that. But I digress...

The TWAIN data buffer width is the problem on the Epson, so scanning 4x5 (which you aren't doing anyway) will mean that you can't scan 5inches wide at 2400dpi ... but I recall one can snuggle in 4 (by orienting the film the other way on say an Epson 3200 holder).

Anyway ... will discuss more later, just dropped by to see.

OzJohn said:

Thanks for posting this. I feel that scanning using a DSLR may well be the way that most people do it before too long except for the most demanding applications where high-end scanners will probably always have a place.

At the lower end of the market it is already possible to buy devices that are really the "guts" of a cheap digital camera, a light source and a simple neg/slide holder and these things produce credible 4x6 scans for an outlay of a hundred bucks or so.

After all, at their basic level, the only difference between a scanner and a camera is the technique employed to capture the pixels - line by line or all at once. With increasing pixel density in DSLRs it seems to me to be perfectly logical to use a suitable camera, if available, for no other reason than the time advantage over scanners. The D800 is an outstanding example of such a camera.

I also think that reversal of negative images in external software rather than in a scanner is a more desirable path. There is a large body of dedicated scanners out there who do in fact scan their colour negs as if they were trannys because they believe that scanners, like digital cameras, are designed as direct positive devices and that reversal algorithms built into most scanners are a compromise at best that do not take into account the characteristics of different film types. Keep up your good work. OzJohn

Click to expand...

One niggle with the above, OzJohn. At their basic level, there is a big difference between a scanner and a camera. A scanner does not have Bayer filtration. It picks up all three color channels (plus infrared if you choose) on each sample. A camera will lose very fine color detail to interpolation, while a scanner will not. This is the same thing you would see in a digital versus film capture of a scene in which there exists very fine color detail (small red berries in a field, for example, where the size of the berry image at the focus plane becomes smaller than one block of the Bayer pattern [typically 2x2 photo sites on the sensor]).

At the final resolution of the capture this may not be relevant. If the target of the scan is the web, it is unlikely such differences would be noticed. But it could make a difference in large prints, depending of course on the image content.

Another difference is in the color depth of the channels. The better scanners have 16 bits per channel. The best cameras have 12-14 bits per channel. It is the difference between 4096 color steps (12 bit) versus 65,536 color steps (16 bit). Perhaps the eye can't see this, but you might find the additional steps extremely valuable in post processing.

I use my D800 for quick scans sometimes. Since I only have an Epson v750, I think the quality of the D800 is as good as if not better than the Epson for 35mm, and is certainly faster. But the Epson is more convenient when proofing a roll of 35mm. And while the quality is not great, it is certainly good enough for web use of 35mm. When I scan medium format or large format, I think the Epson ends up ahead, especially when I use the betterscanning mounting station and do a careful scan.

Again, for web use, the D800 would do just fine even with these larger formats. It all depends on the final purpose of the scan. If the final target is the web, then for most purposes the D800 is a fine scanner. It does beg the question though - if the D800 is good enough as a scanner (assuming you are not stitching multiple images), why isn't it good enough as the original capture device? (Ok, movements, sure. But other than that?)

chuck94022 said:

One niggle with the above, OzJohn. At their basic level, there is a big difference between a scanner and a camera. A scanner does not have Bayer filtration. It picks up all three color channels (plus infrared if you choose) on each sample. A camera will lose very fine color detail to interpolation, while a scanner will not. This is the same thing you would see in a digital versus film capture of a scene in which there exists very fine color detail (small red berries in a field, for example, where the size of the berry image at the focus plane becomes smaller than one block of the Bayer pattern [typically 2x2 photo sites on the sensor]).

At the final resolution of the capture this may not be relevant. If the target of the scan is the web, it is unlikely such differences would be noticed. But it could make a difference in large prints, depending of course on the image content.

Click to expand...

The Bayer array certainly concerns me as a possible problem area. For black and white negatives it seems to be a non issue, but for color I expected to see some issues. In practice I am not finding a problem with color slide film. I think the details on the film aren't small enough that it's showing the issue. I don't see Bayer issues with color neg either, but I am having color issues I don't fully understand.

chuck94022 said:

Another difference is in the color depth of the channels. The better scanners have 16 bits per channel. The best cameras have 12-14 bits per channel. It is the difference between 4096 color steps (12 bit) versus 65,536 color steps (16 bit). Perhaps the eye can't see this, but you might find the additional steps extremely valuable in post processing.

Click to expand...

Here I think the scanner manufacturers are stretching the truth, at least the ones using CCDs (all of them but the drums). The 16bit claims are just for the analog to digital converters. In reality I don't think the scanner CCD comes close to actually being able to distinguish that many shades. I think 10bits is all most of them are capable of producing.

The D800 in this regard is much better than the scanners I have tested. Take a slide with dense shadows, such as Velvia. Scan it on the Epson and with the D800, exposing to the right using a flash or daylight to illuminate the slide. Lighten the shadows of each. I think you will see noise free detail from the D800E and blocked up noise from the Epson. I'm going to write this up soon, but in preliminary testing the scanners seem to come out poorer. For negative film I don't think it's an issue, as both seem good enough for the job.

chuck94022 said:

I use my D800 for quick scans sometimes. Since I only have an Epson v750, I think the quality of the D800 is as good as if not better than the Epson for 35mm, and is certainly faster. But the Epson is more convenient when proofing a roll of 35mm. And while the quality is not great, it is certainly good enough for web use of 35mm. When I scan medium format or large format, I think the Epson ends up ahead, especially when I use the betterscanning mounting station and do a careful scan.

Again, for web use, the D800 would do just fine even with these larger formats. It all depends on the final purpose of the scan. If the final target is the web, then for most purposes the D800 is a fine scanner. It does beg the question though - if the D800 is good enough as a scanner (assuming you are not stitching multiple images), why isn't it good enough as the original capture device? (Ok, movements, sure. But other than that?)

Click to expand...

If the V750 is really that easy to proof a whole roll of 35mm I should look into it. It might be worth it to replace my 4870.

I think the D800E is excellent as the original capture source. In my testing it easily out resolves 35mm film, seems to be slightly better than the Hasselblad for the same size image (24MP equivalent when cropped square). It's not quite competitive with 4x5 or 5x7 in absolute resolution, but I think it's close enough for the print sizes I make. But for some reason I still like to shoot black and white on film. For color I'm regretting how much color film I have acquired before I bought the D800E.

OzJohn said:

I also think that reversal of negative images in external software rather than in a scanner is a more desirable path. There is a large body of dedicated scanners out there who do in fact scan their colour negs as if they were trannys because they believe that scanners, like digital cameras, are designed as direct positive devices and that reversal algorithms built into most scanners are a compromise at best that do not take into account the characteristics of different film types. Keep up your good work. OzJohn

Click to expand...

I think as long as you fully capture the full image detail it's possible to get equally good results with either method. I never trust the scanner software, or a plugin that claims to support film profiles, to get the color right. Both seem to clip channels by default. Once you start doing it by hand it doesn't seem to matter which app does the inversion.

L Gebhardt said:

If the V750 is really that easy to proof a whole roll of 35mm I should look into it. It might be worth it to replace my 4870.

I think the D800E is excellent as the original capture source. In my testing it easily out resolves 35mm film, seems to be slightly better than the Hasselblad for the same size image (24MP equivalent when cropped square). It's not quite competitive with 4x5 or 5x7 in absolute resolution, but I think it's close enough for the print sizes I make. But for some reason I still like to shoot black and white on film. For color I'm regretting how much color film I have acquired before I bought the D800E.

Click to expand...

Here are a couple of examples of whole roll scans of 35mm. First one is from Fuji Astia and the second is from Kodak 400UC.

Full size version -> Fuji Astia

Full size version -> Kodak 400UC


Have you optically magnified your examples - using the bellows setup, to see if the D800E is actually resolving all the details and by how much? If not then maybe you can scan my Techpan example?

• In the test results below, I used a X wide by 4 high arrangement of ISO12233 res charts shown in the bottom left. I then cropped and provided 100% of the center portions using very low compression. All results are therefore multiplied by 4.
• Second image up is the crop from my 14.6MP K20D + manual focus Pentax M 50mm F4 macro lens + Pentax auto bellows from the full frame shot of 35mm Kodak Techpan shot at ISO 25 and processed in Technidol.
• Third image up is the crop from my 14.6MP K20D + manual focus Pentax M 50mm F4 macro lens shot of the 4 X 4 chart arrangement. It is pretty much the same as DPREVIEW K20D resolution results of about 2400LPH. Testing of all my Pentax lenses maxes out the K20D's sensor.
• Fourth image up is the Coolscan 9000 scan of the center portion of the shot made on 35mm Kodak Techpan shot at ISO 25 and processed in Technidol using the Pentax LX + the same Pentax M 50mm F4 macro lens. As you can see, it clearly resolves much more than the 14.6MP K20D sensor. However there are two different resolutions - Vertical of 4000LPH and Horizontal of 3200LPH. For comparison, a 24.6MP Sony A900 is rated at 3700LPH.
• The image on the right is a 4.5 optical magnification of the center portion of the full frame shot of 35mm Kodak Techpan shot at ISO 25 and processed in Technidol using my K20D and Pentax autobellows. This is 12.5 X 4 = 5000LPH. By comparison, the 36.3MP Nikon D800 tops the reschart at 3800LPH. The unfiltered D800E surpasses the 4000LPH chart used so we can't say where it reaches.

Link to larger file -> K20D-Techpan-ISO12233 Reschart

Les, I haven't optically printed either of the frames I used for examples. Next time I'm in the darkroom I'll try to blow up a few sections to see how much detail I can get off the film that way. I have compared the drum scanner to optical printing in the past. For most film I found that the drum scanner was able to capture everything I could get by printing optically. Most optical prints from medium and large format film have greater detail than my inkjet can lay down, unless I print them really large (or crop severely). However you need a loupe to see the difference, so it hardly matters.

Can the Epson V700 make individual images out of the all those frames, or is it a manual process to cut them out and save them? If I remember correctly the 4870 could do that if you used the stock film holder, but it was a bit hit or miss as to how well it selected the borders. From looking at it I assume you are not using the stock holders.

If you want to send me your tech pan image I can scan it with the D800E and the drum scanner. It would be interesting to see how things compare to the Epson and your Nikon scanner.

L Gebhardt said:

Can the Epson V700 make individual images out of the all those frames, or is it a manual process to cut them out and save them? If I remember correctly the 4870 could do that if you used the stock film holder, but it was a bit hit or miss as to how well it selected the borders. From looking at it I assume you are not using the stock holders.

Click to expand...

With the standard holder and Silverfast, I can proof 24 frames in batch. Silverfast tries to find the borders, but I find I have to adjust it's idea of frame locations by hand. Once you figure out the interface, setting clip points, etc. is not too bad. When I'm doing this I'm usually proofing, so I just let it do the inversion, which it tends to do pretty well. This batch operation can be done with the Epson and Vuescan packages as well, each with its own (somewhat confusing) method of setting up the batch. Silverfast tends to be most consistently good with automatic color conversion.

But I agree with your other comment, when I'm doing a final image, scanning a single frame, I'll scan-as-positive, and do the color conversion myself in Photoshop.

Dear Larry, great research and thanks for the info-
With the diminishing fortunes of film scanners and Hasselblad's 4-6K rebate program I wanted to ensure I could scan all my film properly and seriously considered an upgrade from my 343, but for a lark I tested my Phase One P20+ on a bellows with an enlarging lens, even used the cold light head as a source, both files unsharpened, the P20+ file was inverted in PSD and then slight s curve in adjustments unlike your approach for treating the files as you import them which I tried and the Histogram looks very similar to the 343. One question should you not have the enlarging lens on backwards in the photo of your setup?

Platonumb said:

Dear Larry, great research and thanks for the info-
With the diminishing fortunes of film scanners and Hasselblad's 4-6K rebate program I wanted to ensure I could scan all my film properly and seriously considered an upgrade from my 343, but for a lark I tested my Phase One P20+ on a bellows with an enlarging lens, even used the cold light head as a source, both files unsharpened, the P20+ file was inverted in PSD and then slight s curve in adjustments unlike your approach for treating the files as you import them which I tried and the Histogram looks very similar to the 343. One question should you not have the enlarging lens on backwards in the photo of your setup?

View attachment 893

Click to expand...

Glad it's working for you. How's the resolution vs the 343? (Edit, I missed the attachment when I posted - it looks very good).

As far as reversing the enlarging lens goes, that will depend on the lens. The one (75mm Rodagon D) I am using isn't really an enlarging lens; it's more of macro/copy lens and designed for 1:1 reproduction. So in this case it shouldn't be reversed, but in general I think most enlarging lenses should be reversed for 1:1 copy use.

Apo-Rodagon-D

Larry the resolution- i liked the P20+ better and possibly your D800 method with the 36MP vs 16 is better again,, is it possible "the king is dead" if so "long live the king".
Cheers Platonumb

Has anyone tried a similar setup with a canon 5d mk II? I'm very interested in doing this since I find the workflow with my ls40 quite slow. What accessories are available for the canon?

With a $20 (or less) adapter you could use the same Nikon bellows on your Canon. Since everything is manual in the process using the Nikon it should work exactly the same on a Canon.

There are also probably bellows and Canon mount adapters, but I don't know for sure. I do know there are tons of bellows available for Nikon.

Also, Ming Thein, is talking about creating a scanning rig for using a DSLR. I don't know when he'll get it finished, or what the cost will be. But he uses a regular macro lens on his D800E to scan. I imagine his macro rig will work with a Canon system and the corresponding macro lens. I am hopeful this will prove to be a simple apparatus.

So right after writing this I started a search. I ordered a bellows from fotodiox for $37. I think ill use my leica r 50 summicron on the end. Though if i find a good eos to ltm I could put my schneider 80 mm enlarging lens on. Would that be better?

I didn't see a description in your set up of how you keep your negatives flat? (I may have missed it) My thoughts are to use my enlarger as a light source. It has a tilt feature for printing on the wall. Ideally I can tilt it and mount the negative in the holder focus and be set!

mexipike said:

So right after writing this I started a search. I ordered a bellows from fotodiox for $37. I think ill use my leica r 50 summicron on the end. Though if i find a good eos to ltm I could put my schneider 80 mm enlarging lens on. Would that be better?

I didn't see a description in your set up of how you keep your negatives flat? (I may have missed it) My thoughts are to use my enlarger as a light source. It has a tilt feature for printing on the wall. Ideally I can tilt it and mount the negative in the holder focus and be set!

Click to expand...

I use the film holder that comes with the PB4 bellows. This holds mounted slides and strips of 35mm film. I have rigged up negative holders for 120 film, but things were a bit loose and I had difficulty with alignment.

I can't say which lens will be better. You will need to test them. When you do, try them reversed as well (another adapter to purchase). Your idea to use your enlarger sounds good, as long as you can keep everything aligned. I suspect you will need to build some parts. Good luck.

I just picked up something like this on ebay Canon Slide Duplicator 55 in Box | eBay

I'll have to come up with a way to mount it to my bellows but I'm sure it's possible.

I'll post a link to a new thread here when I get my rig all set up.

Thanks,
John

Well my first attempt did not go to well. To many moving parts to futz around with. The canon negative holder I got was unable to be mounted to the end of the bellows because it needed to be closer than that to achieve focus and fill the frame. However it was to fat to fit between the bellows arms so it kind of just didn't work!

I'm not completely off the idea, but I think to get everything lined up and flat you need something like the PB4 which unfortunately are commanding ridiculous prices on Ebay right now.

And for the record, the fotodiox macro bellows I bought is pretty much junk. It kind of works but everything is flimsy and the screws that move the bellows don't have any kind of locking system.

Sorry to hear it didn't go so well. Getting everything aligned is the hardest part. I think that's why the decent bellows still have a high price.

One option might be to cut the rails down on the bellows. This would let you mount the negative holder where you need it. Mount it all to a piece of aluminum angle iron. Looks like you could mount the canon slide duplicator to a small piece of angle aluminum as well, then mount that to the main bar. With some oversized holes for one set of screws you should have simple adjustments for left and right, and some shim stock would let you get the tilt right.

Or wait for Ming Thein's scanning rig to be launched.

So I'm thinking with the bellows I have and something like this:
Amazon.com: Opteka HD² Slide Copier for Canon EOS 1D, 5D, 7D, 10D, 20D, 30D, 40D, 50D, 60D, 60Da, Rebel XT, XTi, XSi, XS, T1i, T2i T3, T3i and T4i Digital SLR Cameras: Camera & Photo

That would hold the film level to the lens and the bellows would allow it to focus closely enough?

I think I may go with the nikon version instead since it doesn't have any cheap glass in it.

Amazon.com: Nikon ES-1 52mm Slide Copy Adapter: Camera & Photo

Am I thinking this through properly?

No one has any ideas on that set up? I guess I'll just have to try it and report back.