DSLR scanning vs Dedicated flatbed scanning.

[robertraymer]

Currently I am using an Epson V850 to scan all of my negatives from 35mm up to 8x10 including wet plates, and while I don't have any specific complaints about it, I always feel that the process takes quite a while and that I am never quite getting what I want out of the negatives. Im sure that some of this has to do with me simply needing more practice, comfortability with the scanner and software, etc, but I have recently been considering other options that may be faster/easier/better. This lead to me considering purchasing a copy stand and using my LED light panel I use as a Lightbox and my D800 to "scan" my images as RAW files and then edit the files in Lightroom, something that has been made even more potentially appealing by a review of NegativeLab Pro I just read.

I was curious as to what other peoples opinions on the matter are? Do you feel there are significant pros/cons to either? If I did make the switch, is there anything specific I should know or consider?

Just some other information that may be useful (or may not, I don't know). Currently I do not batch scan or batch process film scans, and do not "need" that capability, though it would probably be nice if it was easy to do. Most of my scans are usually made so that I have something digital to share on social media/forums/etc, but I do have/would like to keep the ability to make quality prints up to around 16x20 or possibly larger. Also I still shoot a lot of digital work, so RAW editing in LR/PS is actually a process I am more familiar/comfortable with than evaluating/processing scans with the epson/vuescan software.

Thoughts?

[Larry Gebhardt]

Years ago I wrote up my experience doing this for 35mm film compared to a drum scanner. The upshot was the D800E was almost as good resolution wise, equal on tonality, and was much more convenient. I use a Nikon PB4 bellows with the slide/negative holder.

When I have scaled this up to medium and large format some issues were found, almost all related to alignment, film flatness, and moving the film for stitching. Ideally I'd design a jig so the alignment wouldn't drift and there would be a repeatable x-y table for moving the negative for stitching. So far I've just found it easier to use the Epson or the drum scanner than designing that hardware solution or setting up the copystand.

[Pere Casals]

The D800 "scanner" outresolves the V850 for 35mm, with right lens/procedure. For MF, to outresolve the V850 you have to take several images and stitching in Photoshop(or other). For LF...

As the format gets larger the V850 is more suitable. Just make some tests with a tripod before purchasing the copy stand. The V850 is decent for MF, good for 4x5 and superb for 8x10. So IMHO the D800+copy stand would be suitable for 35mm and for jobs requining not much quality.

Just check how it goes before purchasing the stand, use a tripod on a table or place the lightbox on the ground.

With the stand you have to control dust even better than with the scanner, you can use a HEPA air purifier (home models) to get rid of dust.

[robertraymer]

Thanks for the input so far. I will keep it mind. right now it sounds like "right tool for the job" kind of thing. Ill be gone for a few days but am anxious to hear advice/experiences from others as well.

[Peter De Smidt]

There are very extensive threads on this in the DIY section.

[rdeloe]

There are several massive threads on DSLR scanning on the forum that you should check out. Many people have been down this path and are reporting very good results with standard digital cameras like your D800. Some museums seem to be going this route too, albeit with much more specialized equipment. Here's an interesting story about digitizing Autochromes using medium format digital backs to camera scan these very delicate photographic media. They're using stitching when they need higher resolution. (Thanks to Jeffrey G. for putting me on to this story.) https://nglibrary.ngs.org/public_hom...of-Autochromes

I "camera scan" 4x5 negatives (black and white only so far) and process them in Lightroom. I adamantly reject any notion that camera scanning is for times when you don't need much quality. If you have good equipment and a controlled workflow, you can get excellent results. I digitize my 4x5 negs at 2,667 ppi. This is more than enough for making large, high quality prints. If you only need a good quality "scan" for electronic display, you can easily get away with 3 overlapping frames for one 4x5 negative; on my setup that's still 1,500 ppi. If you math that out, you get a file that is 6,000 x 7,500 ppi -- or 45 MP. Over in digital photography land, nobody who has a clue would ever tell you that 45 MP isn't "enough" to make a good quality print!

This is a 13,221 x 10,554 scan (2,667 ppi) after processing. It's stitched from 12 individual frames.



And here's a 3,294 x 2,626 pixel crop.



I'm currently not using anything fancy. I have a home made scanning template that allows me to quickly make 12 frames from a wet-mounted negative. I use an APS-C camera with a 24 MP sensor, and an Olympus OM macro lens. A copy stand would be better, but a sturdy old tripod is working well in the interim. Details here if you're interested: https://www.largeformatphotography.i...ample-approach The only update to this setup I've made is adding additional masking to control extraneous light.

As for processing, I do everything in Lightroom except for spotting, which is excruciating in Lightroom with files this size, but painless in Photoshop. I've tried various add-on tools for inverting the negatives, but (like presets in general) don't see any benefits relative to my own needs. I do the stitching in Lightroom, then invert the negative, export as a TIFF, spot in Photoshop, and re-import the cleaned-up TIFF for further processing. Of course all of this can be done with lots of other tools.

There's only one occasional major downside with camera scanning and stitching, and that's cases where the component images won't stitch. A solution is to use a very precise template when you shoot the frames, and then stitch them in PT GUI with templates. This is discussed at length in the "DSLR scanning" threads; see especially the posts by Peter De Smidt where he talks about the very cool semi-automatic system he and a friend made.

[Pere Casals]

with 3 overlapping frames for one 4x5 negative; ...

The OP's D800 crafts around 18MPix effective in a single frame, not all the 36Mpix are worth, there is a loss in the lens and D800 has a low pass filter (OLPF) (800E and 810 don't have the OLPF, so they are 25Mpix effective able in practice), also we have the bayer mosaic so true resolution may even be well under 18MPix for color resolution, anyway color resolution is less important for accutance tha resolution in the luminance.

APSC cameras are usualy crafting 12MPix effective perhaps.

Then the frames have to overlap for the stitching so a frame yields less than in theory, as a frame overlaps with all neighbour crops then the effetive average recorded pixels can be around 60%, so a D800 perhaps averages around 10MPix per frame. I say "averages" because frames in the corner have a better yield than frames in the center, as there are less overlaps.


A V850 4x5 scan yields around 125MPix effective, so to compete with a V850 a D800 requires from 9 to 16 frames-crops to be stitched in Photoshop, this is a 3x3 or a 4x4 image mosaic. A D800E would require 9 shots.

The question is if it's worth to be stitching a lot of crops or not, as many LF shots have little or no detail beyond what the V850 is able to scan.

A DSLR scanner may even outresolve any drum if the number of crops is large enough and the setup is smart enough...

[Peter De Smidt]

One further note: cameras have improved significantly since when I first started to do this. Back then, I had a Nikon D200. It had limited dynamic range and resolution, but it still wasn't hard to beat an Espon scanner. These days there are a number of great cameras that offer high resolution, high dynamic range, and zero-vibration shutters. All of those things matter. What this all means is that one could get very high quality while doing less stitching. If I remember rightly, I used 25 exposures to capture a 4x5 piece of film with my D200. If you're making 16x20s, I bet that 3 exposures with a D800 or better camera could do it, and it'd be very fast.

Dslr scans and film scanner scans don't necessarily line-up pixel-wise. What I mean is that even when the pixel dimensions of the two scans are the same, it doesn't follow that they are equal in quality. I'll limit myself to Epson scanners, in particular I'm not talking about drum scanners, but my D200 outperformed my Epson scanner significantly, even when the pixel dimensions were the same. The dslr produced cleaner and richer files. Yes, those are subjective terms, but they were also supported by step wedge and resolution target scans. Lastly, Lightroom, Capture One, Photoshop, all are significantly more powerful and easier to use than any scanner software I've used, including Nikon Scan, Epson Scan, Vuescan, Colorgenius....

I believe that Lightroom now allows outputting a raw file from stitched raw files. If so, that's huge.

[rdeloe]

This is a very important point Peter. It's noteworthy that DT Cultural Heritage (the company whose equipment is showcased in that post I linked to above) went with a camera-based system for digitizing items for conservation purposes. Have scanners reached the end of the line in terms of development? It rather looks that way. In contrast, no end is in sight for improvements to sensors for digital cameras. In that respect, it makes perfect sense that a company that wants to sell equipment to museums for digitizing flat artefacts would go with a camera-based system.

When I read the article, my first thought was that this was a strange business to get into given how niche film is, especially large format film. But then the light went and I thought about the vast amount of existing historical material in collections around the world that needs to be digitized -- negatives, positives, prints, etc. -- all eventually (hopefully!) to be made available online. And then there's all the new material HABS/HAER photography. The conservation world knows what it's about, so presumably they have no concerns about the quality of the files generated by "camera scanning".

The business model makes a lot more sense in that context... Sadly, these systems are seriously expensive so they won't be making an appearance in my life!

One further note: cameras have improved significantly since when I first started to do this. Back then, I had a Nikon D200. It had limited dynamic range and resolution, but it still wasn't hard to beat an Espon scanner. These days there are a number of great cameras that offer high resolution, high dynamic range, and zero-vibration shutters. All of those things matter. What this all means is that one could get very high quality while doing less stitching. If I remember rightly, I used 25 exposures to capture a 4x5 piece of film with my D200. If you're making 16x20s, I bet that 3 exposures with a D800 or better camera could do it, and it'd be very fast.

Dslr scans and film scanner scans don't necessarily line-up pixel-wise. What I mean is that even is the pixel dimensions of the two scans are the same, it doesn't follow that they are equal in quality. I'll limit myself to Epson scanners, in particular I'm not talking about drum scanners, but my D200 outperformed my Epson scanner significantly, even when the pixel dimensions were the same. The dslr produced cleaner and richer files. Yes, those are subjective terms, but they were also supported by step wedge and resolution target scans. Lastly, Lightroom, Capture One, Photoshop, all are significantly more powerful and easier to use than any scanner software I've used, including Nikon Scan, Epson Scan, Vuescan, Colorgenius....

I believe that Lightroom now allows outputting a raw file from stitched raw files. If so, that's huge.

[rdeloe]

I believe that Lightroom now allows outputting a raw file from stitched raw files. If so, that's huge.

Indeed it does. In my current setup the input is 12 DNG files (I convert my Fuji RAF files to DNG with a third part developer, outside of Lightroom). Inside Lightroom, I stitch them together, and it creates a DNG. Unfortunately, I end up having to work on the file as a TIFF because Lightroom does not have a usable tool for inverting negatives. It's easy to invert them: you simply flip the tone curve. However, all of the sliders are then "backwards", which makes editing the files a huge pain. Plus, as I mentioned above, spotting is awful in Lightroom compared to Photoshop -- so I end up exporting the DNG as a 16 bit TIFF, spotting it in Photoshop, and then bringing it back into Lightroom.

I was initially concerned that this would be a problem, but a DNG is basically a TIFF so it hasn't made any difference that I can detect.