Digital Scanning vs. Flatbed Scanning?

[neil poulsen]

I don't know much about digital scanning, but I've been reading threads about this methodology. And I wonder, how does it really compare to flatbed scanning?

For simplicity, let's assume that one is using a full frame camera to scan a 35mm negative or transparency. Let's also assume that one is using a high quality, HM Schneider lens that is optimized for 1:1. (Best possible optic, intended for the purpose.) Given the above, let's see where "logic" leads us and what questions this "logic" might suggest.

RESOLUTION:

The arithmetic implies that, if one is using a 24mp camera, one is capable of "scanning" a 35mm at 4000 DPM resolution. Is this not well beyond the effective resolution of an Epson flatbed scanner? One can "specify" higher resolutions in Epson Scan.

But practically speaking, is this resolution achievable on an Epson 850 scanner?

Is it really achievable using digital scanning?

How does flatbed scanning methodology compare to "scanning" through a high quality lens? Better, or worse?

FLARE:

I've read recently that flare can be high in a flatbed scanner, especially compared to drum scanning. By comparison, I can imagine a setup using digital scanning, where flare can be kept to a minimum. So I wonder, can flare control be an advantage for digital scanning, as compared to flatbed scanning?

Again applying arithmetic, a 4000 dpm would enable one to print a 35mm frame to a 20"x13.3" photograph.

COLOR TEMPERATURE:

With flatbed scanning, it's my understanding that one has a single quality of light having a single, constant color temperature.

Yet with digital scanning, one could use one of those 35mm slide duplicators and control the color of the background light. Would this not be an advantage over flatbed scanning?

GAIN:

I've also heard that drum scanning technology enables one to control the "gain" in a scan, thereby enabling better shadow detail and control. And, this offers an advantage over flatbed scanning.

Back to using a slide duplicator as part of digital scanning, wouldn't being able to adjust the intensify of background light be similar to adjusting the gain used in a drum scan? At the very least, wouldn't the degree of control in digital scanning be an advantage over flatbed scanning negatives or transparencies?

So, lots of questions to ask. But, I'm beginning to suspect that digitals canning offers a huge advantage over flatbed scanning.

But, does it really?

[grat]

The Epson scanners are a strange beast. Because they lack focus (ha), there is an EXTREMELY limited range of "sharpness" in them. A well-calibrated, non-jostled, unabused (ie, every time it's moved, the locks are engaged, it's never been dropped, kicked, beaten, hammered, etc.) scanner should be capable of 4800 PPI resolution with no problem. If you're scanning in the middle section of the scanner, using a film holder, with the film at 3mm height above the glass, then you should be using the "high resolution" sensor, which is claimed to be 6400 PPI.

The problem is getting that 4800/6400 PPI. After much testing, I've determined mine is very close to being at 3mm height. I typically scan at 2400 PPI, and occasionally will scan at 3200-- I've never felt a need to push it to 6400 PPI. I'm also not sure my source material is up to the challenge. :)

Regardless, a target of 2400 PPI, with some mild sharpening, produces what I consider acceptable results.

With my DSLR, I can theoretically hit 7000 PPI by doing multi-image stitching. The quality is impressive, but I'm not sure it's worth doing on a regular basis. I'm working on an automated rig to do the image taking, but my 3D printer has been... problematic... and so the project is stalled probably until next month.

[GRAYnomad]

I have an engineering background but that said I tend not to run the numbers.

I baulked at buying yet another scanner so just tried DSLR scanning and I've been well happy with the results. I get approx 45Mpx files from a 6x12 neg and 5x4 should be nearly double that. I could go larger of course by taking more frames but I can already resolve all the grain and can easily up-res 2 or 4x with Topaz if required.

As for gain/Dmax/etc, I only scan negs so that doesn't matter.

In short, unless you plan to buy your own drum scanner, or pay for someone else to buy one, I think that DSLR scanning it the ducks guts and depending on what gear you currently own it will cost almost nothing to set up.

[Joshua Dunn]

I am considering going down the route of using a DSLR to to make multiple row scans of large format film which I would then stitch together. I am thinking about building a scanner that will house a camera and macro lens and be motor controlled on 3-axis. I could program the area of the scan, I just need to figure out how to effectively convert black and color films from a digitally stitched tiff file. That is one of the nice things about using my V850 and Silverfast. The output for what would be intended as large prints might not be perfect but it's very good and you can use it as soon as you scan it. With the digital camera even if you get the scan done well you still have to work with it in post processing before you can use the scan. How people inverting black and white and color negatives in post to get the most out of a DSLR scan?

-Joshua

[grat]

B&W is trivial. Invert, set the white / black points.

Color is more complicated-- Negative Lab Pro (Lightroom plugin) is a popular option. I use Darktable and it's Negadoctor plugin.

For stitching, I use Affinity Photo. I also found that using the same exposure for each frame made Affinity's job much easier, and produced an image with a greater dynamic range.

[Peter De Smidt]

Is it really achievable using digital scanning?

Yes. With a 24mp camera, I get about 1000 spi more than with an Epson when I scan a resolution target. A friend with a 40mp camera gets a real 4000 spi.


How does flatbed scanning methodology compare to "scanning" through a high quality lens? Better, or worse?
Better than an Epson scanner in both resolution and dynamic range. And that's not even taking hdr into account, with is very practical with 35mm film, but less so as one moves up in film size. (It's usually not needed.)




I've read recently that flare can be high in a flatbed scanner, especially compared to drum scanning. By comparison, I can imagine a setup using digital scanning, where flare can be kept to a minimum. So I wonder, can flare control be an advantage for digital scanning, as compared to flatbed scanning?

Yes.




Yet with digital scanning, one could use one of those 35mm slide duplicators and control the color of the background light. Would this not be an advantage over flatbed scanning?

Yes. You can, if you an RGB light source, adjust the RGB number to get the best information from each channel. PWM can lead to banding.




So, lots of questions to ask. But, I'm beginning to suspect that digitals canning offers a huge advantage over flatbed scanning.

But, does it really?

Depends on the flatbed. I have a Cezanne. It does a true 6000 spi. I can't get that high with my camera system. Do I need to? That's a value judgment.

I use a slide dupe system to scan my fathers old slides. No other system I know is as fast, and the results are very good.

[GRAYnomad]

...How people inverting black and white and color negatives in post to get the most out of a DSLR scan?

-Joshua

For colour:

Vuescan will do a pretty good conversion you just load in a pre-scanned image. BUT you have to buy the pro version to get that feature and I'm too tight to do that. I tried Filmlab as well, that seemed to do a reasonable job.

Other programs will do it also, but I just add a layer filled with the base colour and merge with "subtract" in Affinity. Apparently "divide" works as well.

[neil poulsen]

Thanks for the responses. I've obtained reasonable results scanning 4x5 using an older version, Epson scanner. Not having a Nikon 9000 scanner (for example), my interest is in getting a better, clearer scan of medium format film.

I've seen Nikon 9000 scans compared to Epson flatbed scans, and the Nikon scans are obviously much better.. I wonder how results of a Nikon 9000 would compare to good digital scanning?

[sperdynamite]

Thanks for the responses. I've obtained reasonable results scanning 4x5 using an older version, Epson scanner. Not having a Nikon 9000 scanner (for example), my interest is in getting a better, clearer scan of medium format film.

I've seen Nikon 9000 scans compared to Epson flatbed scans, and the Nikon scans are obviously much better.. I wonder how results of a Nikon 9000 would compare to good digital scanning?

The 9000 is an excellent scanner and Digital ICE alone makes it a very attractive option. The downside will be more noise potential in shadow recovery and very-slow-scans!

The problem is that you can't just say 'camera scanning' as if it's one thing. If you do the work and really build up an amazing scanning set up, you're exceeding the quality a Hasselblad X1 or X5 can output. However, you can easily exceed the quality of a flatbed with a simpler set up.

My recommendations are to not slouch on the light source, and try to get a camera with pixel-shift. Negative Supply will sell you a very good, very bright light with a 99CRI. Used Panasonic S1R cameras are not very expensive. Use the Sigma 70mm ART for the lens. Don't try to be fancy with an enlarger lens, bellows unit, it's not worth the hassle. Then get a good copy stand and you're basically all set. You will be able to bang out 150ish MP files from a stack of 4x5 film in minutes instead of hours or days. For medium format, the transport carriers will allow you to do full rolls with incredible speed.

To process the files you have NegMaster, Grain 2 Pixel, and Negative Lab Pro. All are very good, see which workflow you prefer.

The fact is that the set up I just described probably comes close to the current market cost of a Coolscan 9000. Again, the 9000 is excellent, and ICE is awesome. So, do some thinking on which you really want to dive into. You can process Coolscan files through the above software and still get those amazing color conversions we're seeing now. Plus the 9000 has a smaller desk profile by far. There really are no wrong answers here.

[sanking]

Thanks for the responses. I've obtained reasonable results scanning 4x5 using an older version, Epson scanner. Not having a Nikon 9000 scanner (for example), my interest is in getting a better, clearer scan of medium format film.

I've seen Nikon 9000 scans compared to Epson flatbed scans, and the Nikon scans are obviously much better.. I wonder how results of a Nikon 9000 would compare to good digital scanning?

Some digital cameras will require stitching to get results that are comparable to drum scanners with 4X5 and medium format film. You can get this level of image quality in scanning medium format with a Sony a7r iv, using 16 frame pixel shifting. Pixel shifting, or sensor shifting, is a relatively new technology that allows you to approximately double the effective resolution you get with a single frame with the Sony a7r iv. For example, the a7r iv sensor itself with a good lens is capable of resolution of over 100 lpm, with pixel shifting the effective resolution increases to over 200 lpm. And it woks on both grayscale as well as color film. The increase in file size is from 60 mp with single shot to 240 mp with 16 frame pixel shifting. The Panasonc Lumix SIR is also very good for this work. With 8 frame pixel-shift increase the pixel count increases to 187 mp.

At a much higher end in terms of cost the medium fomat Fuji GFX100 is capable of final file sizes with pixel shifting of 400 mp.

Sandy

[Leszek Vogt]

214099214100

Hey Neil, this was mostly built with leftover pieces of plywood. Some dowels, metric screws, linear screw, etc. and I arrived with a unit that will likely excel the 9000...and processing much faster. Mostly this was done with hand tools, tho my neighbor cut some parts on the table saw for consistent measurements. Not counting the lightbox (Kaiser), since one needs one anyway, it's probably equivalent to couple of dinners out with a partner. The clamp & short lip in the rear was intentional, since I figured the set up would likely tip over. I'll be using D750 (a very capable rig) and either 55/3.5 or Tamron 90/2.8 macro....there is enough room to even use 100mm. Amazingly, the linear screw was made for 3D printer and I have not been able to detect any slop....allowing the handle to dial in the exact height desired. Oh, the angle is adjustable a little....enough to obtain true 90 degrees.

Have some other projects and hadn't had a chance to test it properly.

Les

[GRAYnomad]

Neat.

Have you got a link for that linear screw gadget?

[jp]

I've been using an Epson V700 for years. For large format, it does a nice job; it could be beat by stitching with a DSLR, but I don't have time for that. It's also getting dusty and unclean underneath the glass.
As the formats get smaller, the Epson is less great. I think it does a decent job at 6x6cm but is slow. It doesn't do that great at 35mm and is slow.

I have a six month backlog of scanning B&W film and scanning + editing is the most time consuming part of the process. In order to save time, I'm in the process of acquiring the parts to DSLR scan 6x6cm and save the Epson for the big film. I intend to have the DSLR scanning setup in the darkroom, and when the film is dry, run it through the scanning film holder before cutting it up and putting into printfile pages. I would not have to take the bare film out of the darkroom and that would keep it clean.

I've ordered:
"Essential film holder" https://clifforth.co.uk/ (it will be a couple weeks before it gets here). Negative Supply is an upmarket competitor option.
High CRI LED video panel light as a light source. I will add an additional layer of diffusion under the film.
I have:
Nikon D600 or D810
Copy stand
55/3.5 macro lens.

[Serge S]

Hi JP,

Let us know how you like the film holder one you use it!

My experience similar to yours: I like my Epson, (tried the dSLR route), but for smaller formats DSLR is best.


I've been using an Epson V700 for years. For large format, it does a nice job; it could be beat by stitching with a DSLR, but I don't have time for that. It's also getting dusty and unclean underneath the glass.
As the formats get smaller, the Epson is less great. I think it does a decent job at 6x6cm but is slow. It doesn't do that great at 35mm and is slow.

I have a six month backlog of scanning B&W film and scanning + editing is the most time consuming part of the process. In order to save time, I'm in the process of acquiring the parts to DSLR scan 6x6cm and save the Epson for the big film. I intend to have the DSLR scanning setup in the darkroom, and when the film is dry, run it through the scanning film holder before cutting it up and putting into printfile pages. I would not have to take the bare film out of the darkroom and that would keep it clean.

I've ordered:
"Essential film holder" https://clifforth.co.uk/ (it will be a couple weeks before it gets here). Negative Supply is an upmarket competitor option.
High CRI LED video panel light as a light source. I will add an additional layer of diffusion under the film.
I have:
Nikon D600 or D810
Copy stand
55/3.5 macro lens.

[Leszek Vogt]

Neat.

Have you got a link for that linear screw gadget?

I believe this is the one, tho I paid few bucks less.

https://www.amazon.com/Befenybay-Aluminum-Profile-Actuator-Printer/dp/B07WCXB289/ref=sr_1_35?dchild=1&keywords=linear+screw&qid=1616453063&sr=8-35

[Peter De Smidt]

214099214100

Hey Neil, this was mostly built with leftover pieces of plywood. Some dowels, metric screws, linear screw, etc. and I arrived with a unit that will likely excel the 9000...and processing much faster. Mostly this was done with hand tools, tho my neighbor cut some parts on the table saw for consistent measurements. Not counting the lightbox (Kaiser), since one needs one anyway, it's probably equivalent to couple of dinners out with a partner. The clamp & short lip in the rear was intentional, since I figured the set up would likely tip over. I'll be using D750 (a very capable rig) and either 55/3.5 or Tamron 90/2.8 macro....there is enough room to even use 100mm. Amazingly, the linear screw was made for 3D printer and I have not been able to detect any slop....allowing the handle to dial in the exact height desired. Oh, the angle is adjustable a little....enough to obtain true 90 degrees.

Have some other projects and hadn't had a chance to test it properly.

Les

Looks good, Les. Make sure to mask off stray light.

[Peter De Smidt]

You can also look for a used Velmex unislide.

https://www.dropbox.com/s/m628csda2dn7xhh/side_view_2.jpg?raw=1

[GRAYnomad]

I believe this is the one, tho I paid few bucks less.

https://www.amazon.com/Befenybay-Aluminum-Profile-Actuator-Printer/dp/B07WCXB289/ref=sr_1_35?dchild=1&keywords=linear+screw&qid=1616453063&sr=8-35

Thanks.

[Leszek Vogt]

Looks good, Les. Make sure to mask off stray light.

Thanks, Peter. I've got this black art board, which should do the trick. Can't wait to start on this...it's like years overdue.

Les

[Cesar Barreto]

Far from being the most precise solution, I've been using my old Sinar to make multiple rows scanning and it's nice to have X and Y axis movements right from the start. Resolution is quite ok and obviously quite dependent on each one patience to achieve proper alignment and focus. For those who already have such monorail cameras, it may be a cheap and easy way to follow.214286

[jp]

I got 10 rolls of 120 hanging on clips in the darkroom waiting to be scanned and I don't have time to do that with the epson... It's great for LF, but kinda slow and only adequate for MF. I've got more film to develop, but won't do that till I scan in and printfile the dry hanging rolls.

With the long weekend and rainy weather I skipped the hiking and chainsaw chores and did some indoor work. I had time to let paint dry and assemble my project for this discussion. It took about a month or so for the film holder to show up.

It it is quite obvious that my light source is not big enough as I have some fall off on the sides (evidenced by bright spots on the positive) I've ordered a bigger light which might be here for next weekend. LED video lights with adjustable color temp and high CRI are quite common on the usual online retailers. I recessed the light in a pine board to keep it far enough from the diffusion material and the individual LEDs disappear.

I've tried color film with negative lab pro LR plugin which works well. I've also tried sampling the film between photos with the dropper and making a divide mask to remove the orange cast and that works too. All in all, it's much faster than the epson (think click of the shutter to get a histogram compared to doing a prescan) and quality is good enough that I can see the grain texture of Portra 160 with my D810. When I have the next light source and get the images going, my goal is the set this box on my darkroom enlarger table and zip my MF rolls of film through it without them leaving the darkroom so they stay nice and clean and dust free.

The flocked tube is the hood from an old 300/2.8. I'll make a better flocked tube next weekend too with darker telescope flock material and some 3" pipe.

https://live.staticflickr.com/65535/51216353793_d255c1b32e_h.jpg (https://flic.kr/p/2m2PcCD)Untitled-1 (https://flic.kr/p/2m2PcCD) by Jason Philbrook (https://www.flickr.com/photos/13759696@N02/), on Flickr