DSLR scanning 120 film

[GLS]

I just received a high quality apochromatic 20x loupe (Peak 20x measuring loupe), and looking at my film with this thing has been a revelation, especially the slides. I never realised until now just how much detail a well exposed 6x6 or 6x7 frame of Provia 100F or Velvia 50 contained. Amazing! However, my mind was truly blown when it came to look at the Agfa Copex Rapid negs. All hyperbole aside, the detail was absolutely astonishing; in one shot I could literally count every vein and ridge on every small leaf of a tree which had to be at least 100 feet away, or in another see every single needle of a conifer clearly and sharply defined at about the same distance.

This has really made me realise that my DSLR scanning method - good though it may be - is still woefully inadequate to really do justice to what is on these films. I think I will have to start seriously investigating 1:1 stitching methods of digitisation, at least for the more important shots. That or pony up the cash to get them drum scanned.

 

[RalphLambrecht]

Hello all,

The power supply on my scanner has failed and while I’m awaiting a new one (2-3 months lead time) I’d like to use a DSLR setup to digitize my negatives. I’ve got a quadrapod, a Canon 5D3, an LED light table, Lomography Digitaliza (35 and 120) and a converted FD 50mm macro. So far I’ve shot my 35mm film in RAW and gotten 30 MB images. When I shoot 4 images for 6x6 or 6 images for 6x7 and stitch them in Lightroom I end up with 200 MB files.

Comparatively, my medium format scans from Indie Film Lab are around 30 MB (scanned at 4800 x 5900). That’s gotten me thinking that maybe I should be shooting in a lower resolution since the individual images are getting stitched together. Is that accurate? And if so, how low should I go? Canon offers me small, medium, and large JPEGs in addition to RAW. Can I Shoot medium JPEGs and have enough resolution to print 16”x20”? What about 11”x14”?

Then there’s the issue of JPEG vs RAW. I’d prefer RAW files but 200 MB files are too impractical for me. I shoot mostly medium format and have no desire to have such extreme file sizes even if it means compromising quality. I’m of the belief that my images will live or die based on the content, not some technical constraint. In my book, bigger is not automatically better.

Once my Imacon is back up and running I may end up sticking with this method as it seems much faster. Not to mention it takes up far less space than a scanner, G4 Mac, and monitor.

Thank you in advance for considering the question and offering any feedback.


Omid

You can always loer the reolution in PS to your needs. I'd shoot in as an as-high-as-you-can quality,do all your post and then loer res prior to saving but save another high-res for archive.

 

[Eric Rose]

I just posted some images shot on Ilford HP5 and "scanned" using my Panasonic GH5 equipped with a Nikkor 60mm macro lens. The resolution of the digi files are outstanding, not that you could see the benefit on Photrio's gallery. Here is a link, click on the image to see it with better resolution and tonality. https://www.photrio.com/forum/media...tage-park-with-the-blad-click-on-image.60671/

 

[jim10219]

After trying everything to get rid of Newton rings and keep the image as sharp as possible, I discovered a standard glassless negative carrier works best. If the negative is warping, you can tape it down along the register holes with tackless blue masking tape.

As for copy stand, I just use a tripod and don't have any issues with the camera moving. The trick is to use a heavy tripod, like one big enough to support an 8x10 camera. I also use a manual focus (and aperture) lens and by zooming in to set the focus, I never have a problem. I typically use my Nikkor 55/3.5 macro (first version) as it's just as sharp (practically speaking) and easier to use than my enlarging lenses.

The one thing I haven't seen mentioned (and maybe I missed it) is to do it in a darkroom or put a dark skirt over your setup. Stray room light can cause reflections and cause you to lose contrast. Also, it helps if you can control the temperature of the light when scanning C41 film. It can be hard to get the red and blue of the negative within the camera's dynamic range (check the histogram) due to the yellow mask if the color temperature of the light skews too far one way or the other. I use a combination of LED's in my lightbox for this.

 

[jim10219]

I found that bracketing the exposure brings some advantages: you can chose the best exposed shot, or even combine multiple shots using HDR techniques.

My 2 cents,
Etienne

How do you do HDR with film scans? Every time I've tried, the HDR software in Photoshop really brings out the grain, to the point that it becomes very distracting.

 

[etn]

How do you do HDR with film scans? Every time I've tried, the HDR software in Photoshop really brings out the grain, to the point that it becomes very distracting.

I use Photoshop indeed. I haven't noticed what you describe, but I have mostly used HDR with color slides (where there is little grain) and my digital camera is relatively low resolution by today's standards. I might also not have looked closely enough
I plan to scan a bunch of film, both Velvia 50 and Tri-X, in the coming weeks (about 20 rolls from my last vacation, which I want to share with friends). I will run more detailed tests and keep you posted.

 

[etn]

The one thing I haven't seen mentioned (and maybe I missed it) is to do it in a darkroom or put a dark skirt over your setup. Stray room light can cause reflections and cause you to lose contrast. Also, it helps if you can control the temperature of the light when scanning C41 film.

+1 on this!!! My experience too.

 

[etn]

What I really, really would enjoy with DSLR scanning is a kind of rig to advance the film automatically and trigger the camera, so as to scan an entire roll at once.

 

[GLS]

What I really, really would enjoy with DSLR scanning is a kind of rig to advance the film automatically and trigger the camera, so as to scan an entire roll at once.

Not sure how practical this would be if you wanted the optimal DSLR exposure for each frame (i.e. as far to the right as possible whilst avoiding clipping). With changes in the frame density throughout a roll, this can mean a change in the DSLR exposure of a stop or more.

 

[etn]

Not sure how practical this would be if you wanted the optimal DSLR exposure for each frame (i.e. as far to the right as possible whilst avoiding clipping). With changes in the frame density throughout a roll, this can mean a change in the DSLR exposure of a stop or more.

The following has provided good results:
- Aperture-priority auto exposure (aperture set to the optimal aperture of the lens)
- Averaged exposure metering (or however it is called - i.e. no spot or fancy matrix metering)
- bracket at -1.7, -1.0, -0.3, +0.3 and +1.0 EV from the measured value
- lowest ISO on the camera for maximal dynamic range (100 in my case)

With my light table, exposures range from about 1/10th of a second to a few seconds depending on the density of the frame and the exposure compensation of the bracketing. It therefore takes 5-10 seconds to scan a picture, plus the time for film handling.

 

[GLS]

The following has provided good results:
- Aperture-priority auto exposure (aperture set to the optimal aperture of the lens)
- Averaged exposure metering (or however it is called - i.e. no spot or fancy matrix metering)
- bracket at -1.7, -1.0, -0.3, +0.3 and +1.0 EV from the measured value
- lowest ISO on the camera for maximal dynamic range (100 in my case)

With my light table, exposures range from about 1/10th of a second to a few seconds depending on the density of the frame and the exposure compensation of the bracketing. It therefore takes 5-10 seconds to scan a picture, plus the time for film handling.

Ah, fair enough, although that will create many redundant digital files if you aren't doing HDR blending. I prefer to just set the optimal exposure manually for each frame, based off the matrix metering value. However I appreciate this could get quite tedious with 35mm film (I only shoot 120, so a max of 12 frames per roll to deal with).

 

[etn]

Ah, fair enough, although that will create many redundant digital files if you aren't doing HDR blending. I prefer to just set the optimal exposure manually for each frame, based off the matrix metering value. However I appreciate this could get quite tedious with 35mm film (I only shoot 120, so a max of 12 frames per roll to deal with).

Yes it indeed generates a lot of data I then pick up the best exposure and dismiss the others, unless I decide to do HDR (which is rare). Contrarily to what I thought at first, I don't believe that HDR brings much for B&W neg. I find it good with some color slide pictures (I don't shoot color neg so can't comment on this). Obviously, this is purely personal, others will certainly have a different opinion - which is fine, as there is no "one size fits all" recipe with scanning!

 

[etn]

As a side note, I also use autofocus. I find the autofocus much more capable of nailing focus than I can myself. (I probably just haven't tried hard enough )

 

[GLS]

Contrarily to what I thought at first, I don't believe that HDR brings much for B&W neg. I find it good with some color slide pictures (I don't shoot color neg so can't comment on this).

I have never needed HDR for any B&W film either. Certain frames of slide film require almost the entire dynamic range of my D810 to capture in one exposure, but again I have never needed to use exposure blending eith E6, although I suspect there may be some slight advantage to getting better shadow detail in doing so. I have never tested this though, as I find boosting the shadow recovery of a single RAW to give good results, at least with my DSLR.

As a side note, I also use autofocus. I find the autofocus much more capable of nailing focus than I can myself. (I probably just haven't tried hard enough )

I also found autofocus via a cable release to be essential to really nail the focus on the grain.

 

[etn]

I suspect there may be some slight advantage to getting better shadow detail in doing so. I have never tested this though, as I find boosting the shadow recovery of a single RAW to give good results, at least with my DSLR.

Exactly the reason why I am using it. I find it brings better results than shadow recovery, but of course the latter is much easier/faster! (and sufficient in most cases)
I also enjoy very much the low lights clipped to black which Velvia provides.

 

[PhilBurton]

I just received a high quality apochromatic 20x loupe (Peak 20x measuring loupe), and looking at my film with this thing has been a revelation, especially the slides. I never realised until now just how much detail a well exposed 6x6 or 6x7 frame of Provia 100F or Velvia 50 contained. Amazing! However, my mind was truly blown when it came to look at the Agfa Copex Rapid negs. All hyperbole aside, the detail was absolutely astonishing; in one shot I could literally count every vein and ridge on every small leaf of a tree which had to be at least 100 feet away, or in another see every single needle of a conifer clearly and sharply defined at about the same distance.

This has really made me realise that my DSLR scanning method - good though it may be - is still woefully inadequate to really do justice to what is on these films. I think I will have to start seriously investigating 1:1 stitching methods of digitisation, at least for the more important shots. That or pony up the cash to get them drum scanned.

High quality flatbed scanner, so you do only a one-time purchase?

 

[GLS]

High quality flatbed scanner, so you do only a one-time purchase?

From all the comparisons I have seen, high resolution DSLR macro scanning is already superior to the output of any consumer flatbed.

 

[Patrick Robert James]

.....
This has really made me realise that my DSLR scanning method - good though it may be - is still woefully inadequate to really do justice to what is on these films. I think I will have to start seriously investigating 1:1 stitching methods of digitisation, at least for the more important shots. That or pony up the cash to get them drum scanned.

This is why I stay out of these discussions. With my tests you need a pixel density with a Bayer sensor of more than 13000 dpi to really image the film, and that is for color neg. The new Sony looks promising with its pixel shift 240 megapickle resolution. I'd have to do the math in my brain, but that might still not be enough for a Copex Rapid neg, plus you would have to have a lens that is good enough to resolve that otherwise it is just mush anyway. I think that is the biggest fallacy with digital conversion of film. Most of the time it is just mush, and people then say film isn't good....

 

[GLS]

Now I know why I have avoided 1:1 multi-shot stitches up to this point.

I've just spent the last couple of hours trying to do the above on a frame of CoRa, and it has actually proven highly challenging. The main problem is getting deviations from perfect focus across each digital segment, so that when you then stitch them you can get obvious transitions when viewed at 100%. After several attempts I got an OK result from a 9-frame stitch, but it's still far from perfect (although admittedly the issues are only visible at 100%, and even then you'd have to hunt for them, but I like to be meticulous about these things). Sandwiching the film between two pieces of ANR glass to ensure perfect flatness didn't help much either, and doing so introduces potential issues of its own.

At these focus distances with a 100mm macro the DoF even at f13 is absolutely miniscule, and I didn't want to go beyond that because diffraction would start to nullify the benefits. Obviously this technique relies on A) perfect parallelism between the film emulsion and the DSLR sensor, B) perfect flatness of the film, C) perfectly flat field performance and optical centering of the lens, and D) the stitching software not introducing any weird artifacts. I hope the problem is simply the practicality of maintaining A whilst moving the film around, rather than C! To avoid tearing one's hair out I think this whole approach would probably require engineering some kind of rig to guarantee condition A, and also to enable measured & guided incremental movements of the film in relation to the DSLR.

Anyway, here is my first tentative go at it. The original gear used was a Hasselblad 501CM with the Sonnar 4/180 shot at f8 & 1/60th using MLU, mounted on a Gitzo GT3532LS tripod with an Arca-Swiss Z1 ballhead. There wasn't much wind at all that day, so I think all motion blur of the foliage was avoided. The shot in question was only intended to be a test of resolution at moderate distances, rather than to have any great artistic merit. The Copex Rapid was rated EI 25 and developed with SPUR dokuspeed SL-N exactly as per the data sheet. Digitisation was via 9 stitched D810 frames shot with a Tokina 100mm macro on a Kaiser RS 2 XA copy stand (I have written a detailed rundown of my DSLR scanning method elsewhere). The resulting stitched RAW file is 120.8 megapixels, which when printed at 300 ppi would produce a print almost 40 x 40". Shown below is a resized version of the whole frame, followed by a 100% crop of the indicated section:

This still doesn't quite do justice to what's on the film, but it's close.

Thoughts?

 

[etn]

Thoughts?

Beautiful. The grain is clearly visible and well focused on. I love it.

 

[etn]

Just thinking about it, if high resolution is needed, maybe using a camera with sensor shift might be better than stitching? (like the Pentax K1, or the new Sony A7R IV, or the insanely expensive Hasselblad H6D-400c I wish Nikon had this on their cameras...)

 

[GLS]

Just thinking about it, if high resolution is needed, maybe using a camera with sensor shift might be better than stitching? (like the Pentax K1, or the new Sony A7R IV, or the insanely expensive Hasselblad H6D-400c I wish Nikon had this on their cameras...)

Possibly so, assuming you were using a lens good enough to capture all the required details in one shot.

 

[John51]

For the precision needed, I'm thinking that a permanent setup would be best. Once the rig is built and aligned, the dslr stays there and the only thing to ever move is the neg holder.

 

[etn]

Possibly so, assuming you were using a lens good enough to capture all the required details in one shot.

Absolutely!
I read the following somewhere (can’t remember whether it was a forum or an article) : the question was raised, why paying big $$ for a medium format digital when modern full frame sensors also offer 50 Mpixels? Someone stated that all comparative reviews search for the difference at the sensor level, but the difference is actually due to the lens. Being larger it simply has more optical resolution (all other parameters being equal, of course.)

 

[GLS]

I think one has to be cautious claiming larger format lenses automatically have better absolute performance. In fact when talking about digital sensors, for a fixed pixel count less resolving power is needed for a larger format lens in order to resolve details at the pixel level, as each sensel is physically larger. Regardless, there are excellent and not-so-excellent lenses for every format, and each one has to be evaluated on its own merits.

For me the bigger advantage of MF digital is not the larger megapixel count but the sensor size, which affects the DoF and degree of perspective distortion you get for a particular angle of view and framing. But I digress...