Looking for experiences/advice on high-end scanning options!

[Alan Edward Klein]

I'm in a similar position to the OP, developing photography facilities for students. We also have a set of V800/V850s and a flextight X1. The flextight is now unsupported and will eventually break down or become unsustainable, so looking to the future, I am thinking along similar lines to yourself.
In general, it's clear that camera capture is likely to be the only way forward in the long term, but building a system which is workable in a University/student setting is not easy. Getting something which can reliably produce scans of the same quality as the Flextight is not trivial. Having done quite a bit of testing myself, I remain unconvinced that any of the current off-the-shelf solutions within a reasonable budget would provide the quality and consistency we would like when operated by students.
For 35mm, the 5DS R (50mp) which I have tested does produce more or less the same quality as the Flextight (max res 6300ppi) within the focal plane, but keeping the entire frame on the focal plane is difficult to say the least. If you want to achieve the highest resolution, you need to be shooting at f5.6 or even f4 to avoid diffraction. The only way I've managed to get the entire frame sharp at these apertures is by sandwiching the film between two sheets of ANR glass, which unfortunately produced newton's rings. Although the Negative Supply holders look pretty good, I seriously doubt that they will be capable of keeping the film this flat - especially with frames at the end of a strip. For me at least, any solution needs to be able to deal with cut strips because that's how film tends to be handled.
Without an extremely high precision system, I'd say it's more realistic to aim for more like 3200ppi on average across the frame, which is still substantially better than an Epson, but even this would be quite hard to maintain with student operators. It doesn't take much for the resolution to drop to below what you would get with the Epson. It would probably be best to recommend shooting at f11, which would limit the maximum resolution in favour of consistency.
In medium format, assuming a single-shot workflow, you would be hard-pushed to get any better than the flatbed with anything larger than 645. Using a 5DS R to shoot 6x6, 6x7 or 6x9, the theoretical maximum resolution is around 2600ppi - realistically you would be getting less than this, and the Epson gives you around 2400ppi. Multi-shot stitching is a possibility but I would not have confidence that many students would a) want to go to this trouble, or b) be capable of doing it properly. 645 could work well, however - with a theoretical max res of around 3500ppi. Because of the lower magnification and subsequent greater depth of field, it might be realistic to get a consistent 3000ppi, which roughly equals the Flextight (3150ppi).
So my thinking at the moment is that it would be feasible to set something up which could consistently get better results than the Epsons for 35mm, but at present I can't see a solution which could reliably get close to the Flextight. For 645, a high-res camera-based system could potentially equal the Flextight (though this is only moderately better than the Epson anyway). For MF 6x6 and above, I don't really see a solution. For large format, the Epsons are probably already the best option.
To be honest, given the above, I think it might be worth considering the Plustek 8200i for 35mm, since it's a self-contained unit, requires very little setup or maintenance and does the negative conversion for you. The optical resolution is around 4000ppi and it's a lot cheaper than a camera setup with good components. It would be a lot slower than a well set-up camera system however, so it depends on whether you want to prioritise batch-scanning or getting the best out of a few frames.
On the subject of control, personally I would not want to attempt a camera setup without live view, which is invaluable for levelling and checking focus. So I'm not sure that using the H3D is a good idea - also its firewire interface will even make tethering a problem going forward.
Anyway, there are some thoughts from someone in a similar position to yourself - sorry I can't really offer any solutions. I'm sure there is a way to make the camera scanning thing work, but I think people generally underestimate how hard it is to get something that works consistently well with operators that don't know what they're doing (students). If you do go down this route, I'll be really interested to hear how it goes.

It might be easier to go back to all chemical.

 

[Alan Edward Klein]

I honestly think that too many scanning threads get ruined by focusing on irrelevant variables, particularly resolution. User experience, consistency, throughput, tonality and colors are much more important, in my humble experience. If you want your students to enjoy photography, eliminate agony & torture first, THEN think about fringe variables like resolution. Scanners = Agony+Torture. Silverfast = Agony+Torture. Give them the most pleasant method of shooting film and posting gorgeous-looking results on Instagram. That's what they want.

I have already provided some medium format samples above, the kind that you can get by scanning a 12-shot roll in 3 minutes, and be done in less than 15. Here's a 35-mm scan and another one, these your students can get at a rate of 5-7 rolls per hour, using modern software tools, with minimal color corrections, as compared to something like Silverfast. Look at the kids in the corner of the 2nd shot. Do you honestly believe there's anything else you can squeeze out of a 35mm negative that your students can't live without? I can keep producing (and sharing) results of scanning cheap consumer films like this or this by a hundred per hour, knowing that scanner operators get that kind of output only after monumental amount of suffering, because I used to be one.

And don't get me started on ICE. Dust is yet another reason why scanners are painful. I stopped using scanners, I stopped seeing dust, because apparently films don't attract much of it by themselves. Scanners do.

My new V850 seems to attract more dust than my old V600 did. Maybe it's the extra glass in the film holder. In any case, there's just more of it.

 

[PhilBurton]

I scan directly into Lightroom, select a RAW image, press Ctrl+Alt+N to invoke the NLP plugin. Usually I start with a "Soft Lab" preset, set the white balance to "none" or "standard" and that gives me 90% of the final look. I like to adjust some brightness, then correct Portra warmth in the shadows by adding a bit of cyan there, set "soft highlights", and sometimes apply the Fronter LUT as I like how it improves mid-blues. Then I just export to JPEG using one of the presets that I have for different negative sizes. I find that this gentle method preserves the natural emulsion look without making it looking overly color-corrected.

I didn't know that you could scan directly into Lightroom. I have always believed that you need separate scanner software like Vuescan or Silverfast. No? (That would save me some nice money.) Do you use one of these? At what settings in that software?

 

[Les Sarile]

Absolutely massive boost in visible resolution, especially if your lens can render it onto the sensor.

That is actually the reason I did this test was to see how good or bad these used manual focus lenses bought super cheap on craigslist were. And after doing these tests, I still don't know which is the limiting factor - film or lens, as clearly the scanners and DSLRs I used did not come close to resolving the details captured on the film. One thing's for sure, these lenses will not be the reason I am unable to capture all the details in the scene.

 

[Les Sarile]

The number one film that comes through my lab for processing is..... *drum roll*..... Kodak Colorplus 200, followed by Gold 200, and Ultramax 400. Most of the images are shot on crap point and shoot cameras they picked up at a garage sale, or on a camera they had in storage for the last 20 years and pulled out, or was given to them by their parents. A huge number of the images aren't exposed correctly, have a way too slow shutter speed to stop motion, and/or aren't in focus because the camera either doesn't autofocus, or they haven't figured out how to focus the camera yet, and in general are what you would expect to see if a complete amateur was shooting it, which is actually what is happening.
. . .

I'm sure the old guys on here would totally choke if they saw what comes through my lab. The younger generation by and large doesn't give a flying nit about the stuff many of the older folks on here hold near and dear to their heart. They care about turnaround time, and file size. If you can't turn and burn in less than a week, you aren't going to get their business.

I'm pretty sure that in order to make good comparisons you can't go with the lowest common denominator. Keep in mind that an scene - no matter how poorly captured, still has to be rendered accurately. I am not above using "lowly" films or even film that has been setting out in a Georgia driveway for years. You may be surprised that even those come put perfectly fine with a good scanner like a case of Kodak Ektar 100 I picked up . . .

Or some way long expired Kodak Gold 25 donated by a forum member . . .

Simply slip the strips of film into the Coolscan press scan and go. They come out ready to go - clean, and no post work to convert them to accurate representation of what they're supposed to be. Of course if you want, you can always apply whatever post correction you want as with any digital image like stitching multiple frames of Kodak Ektar 100 . . .

Or using HDR, shadows & highlights . . .

I mean you run a lab, you try to manage scanning costs and be able to provide a good product. Can you do that with DSLR scanning? Sure the capture is fast but when it comes to color negative comversion, there is no way you can have human intervention on every frame for the prices you charge. Even if you provide them the lowest scan resolutions! If the majority of scans the OP will be doing are of color negatives, then I am reasonably certain that's not sustainable in a school environment.

 

[Adrian Bacon]

I mean you run a lab, you try to manage scanning costs and be able to provide a good product. Can you do that with DSLR scanning? Sure the capture is fast but when it comes to color negative comversion, there is no way you can have human intervention on every frame for the prices you charge. Even if you provide them the lowest scan resolutions! If the majority of scans the OP will be doing are of color negatives, then I am reasonably certain that's not sustainable in a school environment.

The most common scan I get asked for is the "straight scan" which is no intervention by me. I simply scan it in, run it through my software to create the files, and send them. It's fairly fast throughput. I run control strips and have a pretty tightly calibrated setup. The people that care pay for the basic scan which is intervention by me as needed, and the premium scan is I touch every frame.

 

[Adrian Bacon]

you try to manage scanning costs and be able to provide a good product. Can you do that with DSLR scanning? Sure the capture is fast but when it comes to color negative comversion, there is no way you can have human intervention on every frame for the prices you charge.

Well, here are two screen grabs of a roll of superia 400 xtra and gold 200. Both shot on a little crap point and shoot. This is straight out of my scanning software. My current hardware setup is a Canon 90D, Sigma 70mm macro art lens (the new one, not the old one), a strobe for the light source. Both rolls were shot in all kinds of mixed lighting with all kinds of different exposures, pretty representative of what comes through my lab pretty much every day.

The straight scan isn't a terrible place to start.

First, the Superia

Then the Gold 200

 

[fs999]

Is this 35mm or one of the medium format sizes?

Medium format 6x6

 

[Tom Kershaw]

Medium format 6x6

That is faster than my Nikon scanner, although I do always use the 'fine' mode, otherwise the Coolscan 9000 is prone to banding.

 

[fs999]

That is faster than my Nikon scanner, although I do always use the 'fine' mode, otherwise the Coolscan 9000 is prone to banding.

Never had banding with iSRD, I use always standard settings. The IR scan is really fast compared to Epson digital ICE...

 

[Helge]

Ah! The same old bullshit with exotics like "The Spur Orthopan film, developed in Nano Edge" comes out. You could have used a better known edge case like CMS20. All right, let me just stitch two shots then, by adding 30 seconds to my workflow. But most importantly is that nobody here shoots "spur orthopan", on a tripod, with $5K lens at optimal resolution, developed in some kind of nano edge. Gold 200, HP5+, Delta/T-Max 100, HP5+ and Delta 400, all CN 400 films, developed in Xtol, D76, or Flexicolor - everything the real world people shoot, comfortably falls within capture capabilities of any 24MP sensor manufactured within last 5 years. Which is very different from your stated BS of "Even an 8x10 print will show of the deficiencies and artefacts." I posted full-size scans. Print them at 8x10" or even bigger, like I did. They're more than enough to prove my point. This is something that is enjoyable and available to anyone in their own home, approximately nobody in the world is interested in "spur orthopan developed in nano edge and printed in a darkroom".



I scan directly into Lightroom, select a RAW image, press Ctrl+Alt+N to invoke the NLP plugin. Usually I start with a "Soft Lab" preset, set the white balance to "none" or "standard" and that gives me 90% of the final look. I like to adjust some brightness, then correct Portra warmth in the shadows by adding a bit of cyan there, set "soft highlights", and sometimes apply the Fronter LUT as I like how it improves mid-blues. Then I just export to JPEG using one of the presets that I have for different negative sizes. I find that this gentle method preserves the natural emulsion look without making it looking overly color-corrected.

Here is the Fuji 400H Pro scan stitched from 2 shots (Ctrl+N in Lightrooom gives you a stitched DNG), 6555x6555 pixels, so over 42MP which is way more than 99.95% of people will ever need. Looking at this I find Helge's comments about "insufficient for actual prints" laughable. In software industry we say "show me your code or go home". I showed my scans, and I suspect they print beautifully at all sizes Helge ever tried. The sample above will print and look great at 30x30" (and I have never met anyone in person who ever printed anything at 30x30")

My point is, camera scanning gives people better scans that they know what to do with, at a fraction of a time compared to a scanner, and without agony+suffering associated with scanning that potentially turns folks away from film.

Well now you're just into relativist territory.
Microfilm derivatives are not weird or cheating. They are available and quite usable in daylight or in a studio like setup. With the correct developer they have better than good tonality. I'd say even great, with careful exposure.
There is really nothing like them in the digital world if you want very high resolution black and white.
The same exceptions and low-pass cutdowns you employ, can be used for digital too. Motion compensation and high QE are not magic bullets.

TMAX and Delta or any of the good colour films (negative or slide) are not as far behind the microfilms as you suppose. This example done with calibrated drum scanners and a microscope from Tim Parkin clearly shows that:

https://www.onlandscape.co.uk/microscope/

Compare the 4000dpi scans with 8000dpi. There is a clear advantage (at the risk of being insultingly obvious: No, format does not matter. Dpi does). The microscope shots shows there is even further potential resolution and cleanness to be had by further improvements in scanning. Even at these high resolutions there is still grain aliasing beating going on.

Your example is clearly from a 6x6 medium format, with which the egalitarian, "down with the common man" virtue signalling you possibly have going, is not really congruent. The average "pundit" doesn't even know there is something other than 135 film, let alone how to shoot it. Your students does, but they are lucky to have you to show and tell.
In this case format matters, since you need to scan it at the highest possible resolution for two (somewhat connected) reasons.
As already said, high res scanning gives you obviously high resolution, but also a cleaner less "grainy" looking photo due to the (many a time) aforementioned grain aliasing. Medium format certainly gives you, some of that for "free" or earlier than 135.

 

[Adrian Bacon]

Helge, I provided not one but about 10 examples, both 35mm and 6x6. Scroll up, download, print and see for yourself. Apologies for making them hard-to-notice hyperlinks instead of using the attachment feature, but that's the only way I know how to post full-size scans here.

I am agreeing with you that there are emulsions out there with amazing resolution potential (my personal experience is limited only to CMS20 in that regard), I am just pointing out that:

* The films we actually shoot are nothing like that.
* One cannot buy a new scanner with an effective resolution higher than a modern digital camera.
* Your remark about "drug store print quality" is wildly off and borderline offensive.
* Resolution is overrated anyway, as most people do not print and even when they do, that's rarely bigger than 11x14" or so.

Even Ektar in 35mm doesn't begin to stress the capabilities of a Fuji's 24MP sensor. I did not observe any additional detail from Plustek 120 Pro 35mm scans vs a DSLR, even after adjusting the focus. Therefore I have settled to downsampling all of my 35mm scans to 5000x3400px to keep the file size reasonable, and even that feels excessive for 99% of my pictures as they simply don't have 17MP of useful detail.

I think it’s important to note that film resolution is a very different thing than total system resolution. Your total system resolution is never going to be as high as whatever some film can capture. As long as the total system resolution meets your needs (which only you can determine what your needs are), then who cares how awesome some film is? Different people will have different needs, and that is totally fine.

There are a number of people on here that have the position that film has absolutely massive resolution that no digital system can ever match. They usually also rend to be of the position that grain aliasing is a very real thing that happens all the time. The only thing I’ll say to that is film is great, and has its place, but digital has been providing more than acceptable results for a lot of people for a very long time. With respects to grain aliasing, I wouldn’t be so quick... most of the times I see someone pointing out grain aliasing in some sample, they’re doing it with a positive image that’s been run through a bunch of processing to get said image from a negative. If you’re trying to evaluate grain aliasing, shouldn’t you be looking at the raw samples off the scanner before any other processing is done? Otherwise, how do you know what you’re seeing is actually coming from actual aliasing and not a byproduct of something else introduced in later processing steps? I’m not saying it can’t happen, I’m saying before anybody calls it, make sure you’re looking at the right thing first.

 

[Adrian Bacon]

Wise words. Moreover, people should count their own eyes as a part of the system. I am always suspicious of people demanding 300dpi for a 16x20" print.

lol... yeah... it’s nice to have that as a standard, but the reality is 8-10MP will look totally fine printed on an 8x10, a 16x20, and a 32x40 at any reasonable viewing distance for each respective size. If you want to get up close and inspect a giant print from a few inches away... well, all you’re doing at that point Is the equivalent of pixel peeping. That’s kind of like going up to a giant billboard or movie theater screen and complaining that the picture is soft.

 

[Lachlan Young]

Wise words. Moreover, people should count their own eyes as a part of the system. I am always suspicious of people demanding 300dpi for a 16x20" print.

Resolution has come to be regarded as a questionable metric of film comparison for anything other than specific aerial photography roles since about, oh, 1952. It's what's been called 'Image Content' and the ability to transmit that (ie MTF sharpness in specific frequencies relative to format, overall noise/ granularity performance, latitude, quality & quantity of colour etc) that is far more important to how we perceive the performance of film and scanned film. No one cares as much about outright 'resolution' as long as your scan is adequately artefact free for the intended size/ resolution & not so inherently soft in resolving the inherent granularity/ fine detail resolution of the material that the subsequent over-sharpening produces yet more artefacts in areas that should deliver fine sharp-edged detail. The other problem is people using too large a gamut for inverting colour negs, but that's a different story. Regarding the 300ppi at 16x20, a lot depends on print size and viewing distance - with today's print heads you can see a difference out to maybe 500ppi-ish (which is pretty close to the realistically useful resolution limits of darkroom paper), but you need to be at 8x10 viewing distance to really see it. And most viewers are really not going to notice unless you stick them side-by-side.

 

[138S]

For a photography workshop at an university we're looking to upgrade some of our scanners. I'm struggling with finding the best solution and I hope you'll share your experiences with certrain methods of scanning.

Single problem of the V850 is that it requires attention from a good operator to get a top notch result.

As graph bellow shows, if you miss focus by a 1.5mm (curling) then you may get 1/3 of the line resolution and 1/9 of the area resolution. (See the case you go from 3mm to 1.5mm)

Additionally, a top notch Epson scan also requires a careful image optimization (in Photoshop or the like) not everyone is able to perform.

But those people being able to operate the V850 properly do obtain an amazing yield. If you want the students learn to squeeze all the performance an scanner may yield then retrofitting the V850 with BetterScaning holders would be a choice, specially for MF and up.

If you want the students get perfect scans directly, without asking, then the Epsons are a bad idea.

My personal experience is consistent with this review: https://petapixel.com/2017/05/01/16000-photo-scanner-vs-500-scanner/ , but YMMV

 

[Wallendo]

Does anyone have experience with dSLR scanning in a shared environment? People apparently get great results with these scans when equipment is properly assembled and used, but I would wonder if a long stream of students using such a setup would tend to knock the equipment out of alignment routinely.

 

[Adrian Bacon]

Does anyone have experience with dSLR scanning in a shared environment? People apparently get great results with these scans when equipment is properly assembled and used, but I would wonder if a long stream of students using such a setup would tend to knock the equipment out of alignment routinely.

if you’re looking to set something up in a shared environment for student work, I’d just set up a bunch of midrange flatbeds that came with reasonably good and easy to use software. The goal in that scenario should be ease of use and maintenance, not maximum performance.

 

[Andrew O'Neill]

Personally I work with the V750, as I shoot up to 8x10. I've never been disappointed. Mind you, that's the only scanner I've ever owned. If no one is shooting 8x10 at your university, then there is no point in having the V750. I do have a few V600's at school where I teach, and they perform really well. Hardly high end though, but good enough for students.

 

[138S]

Does anyone have experience with dSLR scanning in a shared environment? People apparently get great results with these scans when equipment is properly assembled and used, but I would wonder if a long stream of students using such a setup would tend to knock the equipment out of alignment routinely.

DSLR can be a good idea or 35mm, for MF and up you may have to stitch crops to get a competitive result, still a good setup is not cheap, if one has a top notch DSLR for other things then it may be also used, but if you have to but a today's Pro DSLR plus a fine lens... then you have to ensure film flatness... remove ambient illumination... and expenssive thing to get the camera aligned and steady... Block lens focus... if students have to touch the dslr/lens then a risk is there.

Another drawback is color inversions, it is difficult to get good results from DSLR scanning, for sure it is possible but it can be challenging, review that Nick Carver video at youtube, IMO he is right...

Also no IR channel dust removing tool is there...

________

One thing it may be useful for that room is an HEPA purifier to mostly eliminate dust from air.

 

[grat]

lol... yeah... it’s nice to have that as a standard, but the reality is 8-10MP will look totally fine printed on an 8x10, a 16x20, and a 32x40 at any reasonable viewing distance for each respective size. If you want to get up close and inspect a giant print from a few inches away... well, all you’re doing at that point Is the equivalent of pixel peeping. That’s kind of like going up to a giant billboard or movie theater screen and complaining that the picture is soft.

Several years ago, I took an 8MP image from my EOS 30D, and had it printed at 16x20, framed and mounted for my mother. I believe the DPI on the final print is either 125 or 150. No one, and I mean no one, has ever had anything but positive comments about it-- even at the shop where I had it framed.

 

[Light Capture]

Single problem of the V850 is that it requires attention from a good operator to get a top notch result.

As graph bellow shows, if you miss focus by a 1.5mm (curling) then you may get 1/3 of the line resolution and 1/9 of the area resolution. (See the case you go from 3mm to 1.5mm)

Additionally, a top notch Epson scan also requires a careful image optimization (in Photoshop or the like) not everyone is able to perform.

But those people being able to operate the V850 properly do obtain an amazing yield. If you want the students learn to squeeze all the performance an scanner may yield then retrofitting the V850 with BetterScaning holders would be a choice, specially for MF and up.

If you want the students get perfect scans directly, without asking, then the Epsons are a bad idea.

My personal experience is consistent with this review: https://petapixel.com/2017/05/01/16000-photo-scanner-vs-500-scanner/ , but YMMV



View attachment 263771

This is absolutely correct. Great care and technique is needed to extract maximum.

Even with quality macro lenses resolution is around 5-5.5 micrometers at 1:1 mag for DSLR scanning methods.
That's again meaningless number without glass holders. Graph would be very similar but with macro lenses resolution drops much faster with film curling.
Fractions of a millimeter make a difference. Also, very few lenses can come to this resolution at required magnification in corners.

If you want fast results, they're not much different than other methods. For top notch results, great care is needed.

 

[138S]

Several years ago, I took an 8MP image from my EOS 30D, and had it printed at 16x20, framed and mounted for my mother. I believe the DPI on the final print is either 125 or 150. No one, and I mean no one, has ever had anything but positive comments about it-- even at the shop where I had it framed.

A 27" Full HD monitor has 2 MPix, and it does not llok bad at all... also resolution is probably overvalued many times...

Still, the trained eye may enjoy a gorgeous sharp print, some images may benefit from that, others won't.

 

[Adrian Bacon]

Several years ago, I took an 8MP image from my EOS 30D, and had it printed at 16x20, framed and mounted for my mother. I believe the DPI on the final print is either 125 or 150. No one, and I mean no one, has ever had anything but positive comments about it-- even at the shop where I had it framed.

Yep. Much more than 2-3 feet away and you can't actually see that much resolution. Most big HDTVs and 4K TVs are 75-150 pixels per inch and nobody is complaining about how low the picture quality is when watching a blu-ray movie. I take scans of my personal images and make jpegs that are 1920 pixels on the long edge, stick them on a thumb drive and plug it into my 65 inch 4K TV, and frankly, they look stunning playing as a slide show. 35mm film has never looked so good. That's not even sizing the jpegs for native 4K display. Anybody who tells you that 35mm film looks terrible printed much larger than an 8x10 or 11x14 hasn't seen a reasonable scan on a nice big 65 inch screen.

 

[Les Sarile]

Several years ago, I took an 8MP image from my EOS 30D, and had it printed at 16x20, framed and mounted for my mother. I believe the DPI on the final print is either 125 or 150. No one, and I mean no one, has ever had anything but positive comments about it-- even at the shop where I had it framed.

Perhaps since you did this for your mother that it has content that glosses over any technical flaws - if there are any. Besides, a negative comment may not be in their best interest if they don't want to be disowned . . .

I just did some scans for a colleague - his wedding shots from back in 90 on medium format color negatives. Obviously his wife has enjoyed the prints all these years but now has new perspective on those pictures. She really loves to check them out on their ultra high def big screen TV. From what he tells me she wants more of those photographs scanned.

 

[Les Sarile]

A 27" Full HD monitor has 2 MPix, and it does not llok bad at all... also resolution is probably overvalued many times...

Still, the trained eye may enjoy a gorgeous sharp print, some images may benefit from that, others won't.

And of course sharp and high resolution are not necessarily the same thing.