Scanning Analog Film For The Last Time

Film cameras are capable of great resolution, and for a long time were superior in this regard to their digital successors. However, it’s now possible to store digital copies of analog images in superior detail, so [Jan] built a rig to scan their photos for the last time.

The general idea is to take a high enough resolution scan of film negatives or slides, such that there is no need to rescan the images when technology moves forward. To achieve this, [Jan] decided to employ a DSLR to photograph the materials in question. To do this quickly and accurately, with minimal fuss, special lens hoods were 3D printed to hold slides in perfect register in front of the lens. With a flash to provide even light, the results are excellent. Film negatives proved harder, requiring a carefully designed transport mechanism to avoid damaging the fragile materials. With some perseverance, the final tool worked well.

It’s a tidy way of digitally archiving analog photos, and with the resolution of modern cameras, one needn’t worry about lost resolution. We’ve seen mechanised builds for handling other formats too, such as this 8mm scanner. Video after the break.

44 thoughts on “Scanning Analog Film For The Last Time

  1. Now pack them away carefully so they can get rescanned again for the last time in 30 years time when the digital formats you used have bitrotted or need unobtanium hardware or software.

    1. or do what I do, give them to someone who cares, cause I sure as shit don’t need 4000 pictures of my dead family sitting around the living room after thanksgiving

    2. yes, bitrot might have got them – especially if you didn’t keep enough copies (and also checksum them regularly), and copy them onto new hardware when you get it – but unobtanium software isn’t a problem – most image formats aren’t that complex to unpack, someone could write a own display program fairly easily even in 100 years time..

        1. The problem is – as usual – proprietary stuff. I have some files .rif that i can’t open. Iirc the program was called “ArtDabbler” or something and sold with some cheap graphical tablet, but i was not able to find something to open this or specifications of the format on the web. :-(
          So yeah, if you use jpg it shouldn’t be a problem for a long time, but RAW or other (proprietary) stuff…

  2. > such that there is no need to rescan the images when technology moves forward

    Hoo boy….

    It’s not just about the resolution. The dynamic range of film exceeds that of current digital sensors, and the color response to a flash light isn’t the same as to a full-spectrum incandescent light, which too isn’t the full fidelity or fully correct to what the film could provide with a properly adjusted and calibrated light source. Thinking that they’ve got “all that’s there” is just pure fantasy if you want to get pedantic about it.

    1. Lol don’t waste your breath arguing. There’s a whole cottage industry of digital evangelists who come out with highly similar articles every month or so. They actually don’t give a tuck..

      1. That’s his exact point – a flash has a different color spectrum than film.

        Then there is the potential issue of sensor metamerism.

        Of course, these are corner cases and in the end, you might never need anything “better”, but it’s not guaranteed you’ve gotten everything because even with much higher resolution than the film grains and some sort of pixel-shift approach (which is usually not compatible with flash photography) to eliminate Bayer array limitations, you still have a possibility of color metamerism issues.

    2. You have to trade one thing for another with film.
      Dynamic range? The best resolution you’ll get is from slide film, but that costs you in dynamic range (from memory Fuji Velvia any has about 7 stops of DR, modern digital SLRs can manage 13 or more).
      You don’t like digital’s colour response to flash? You can run them at ISO25600 or more and dispense with flash for low light work, or shoot in RAW and push the shadows rather than use flash for infill.
      You don’t like digital’s colour response? It’s actually more faithful than most film (which is a hugely variable thing – Velvia vs Kodachrome fr example) and can be adjusted in RAW post processing to pretty much anything you want.
      It’s pretty much down to personal preference these days. If you like film then all well and good, but don’t pretend that it is technically better in all but limited circumstances.

      1. I think Luke wasn’t arguing tnat film is better than digital. In this case film is the only original we have and digital, however good it may have become, cannot extract “all” information from it. just like duplicating film can’t. They can only hope to interpret it better next time.
        A translator of literature said something similar: Originals are sacred, translations become obsolete every 15 years.

    3. Most pictures, when they were taken, were not done with a calibrated light source, and nobody cared.

      Why should we all of a sudden care when they are transferred to digital ?

      1. But if they were taken outside, you can look up the number of sunspots that day and thus the exact luminous intensity, correcting for visibility and smog reports from local weather stations.

          1. Aunt Ethel be all: “We weren’t wearing dayglo shirts that day, we were pointing to a large flock of starlings almost blotting out the sun..”

  3. How’s the bit-depth of DSLR raw formats these days? Doing a good job converting color negatives to attractive images requires a lot of expansion of the dynamic range and just eats up bit-depth. Even 12 bits per-channel capture might be a bit low for the application. My *entry-level* second-from-last-gen Minolta film scanner (which I’ll admit is now unobtanium) captures about 14-bits per channel after noise, and later or higher models do better. Slides should be fine by DSLR though.

      1. Well actually its the other way round. As negatives ( especially color negatives) have a compressed dynamic range, the chip of a camera, optimized for a much bigger dynamic range only utilizes a few bits of its 12-14 bit depth for this negative. Somewhere in the 5-7bits per pixel. That is not enough color info to actually capture the scene. Dedicated film scanners have a programmable gain stage between chip and AD converter to expand the optical compressed dynamic range of the negative, using all bits available for a pixel.
        Also, to prevent optical distortion, use a non-zoom lens made for macro shots or the lens for a film enlarger. Of course you can correct barrel distortion in software, but there goes the sharpness.

        I have been using a vertical setup made from a old durst enlager with a canon eos 5d mk II and a nikor 50mm enlarger lens on extension tubes with good results for b/w negatives as they have more contrast, but the results of color negatives are poor.

        Slides are no problem. They have the contrast to fill the bit depth of the camera without problem

        Another thing is resolution and larger negatives. If you use a flatbed scanner, the larger your original, the more pixels you get, but with a camera setup, the available pixels stay the same, you just zoom out optically to cover the larger negative.

        Of course flatbed scanners with backlight are more rare and tend to have lower quality optics and an actual resolution much lower than stated on the box. :(

        Good scans are hard.

        1. “Good scans are hard.”

          And this is why *I* still shoot film and have no plans to shift to digital as a primary medium. I’ve seen electron micrographs of real film grain — the “grain” in the tight crop of the ship transparency above isn’t it; it’s a sampling artifact. Analog film, of good quality, well exposed and correctly processed, contains much more information (in the 24x36mm frame of 35mm film, never mind larger formats) than even a 50 MP digital can extract.

          Far beyond that is the long term. Color dyes may fade, but I’m confident that my black and white negatives will be recognizable, to the eye, as containing images in two hundred years, perhaps much longer, if they aren’t destroyed by environmental conditions (flooding, pollution, fire, etc.). Will people two centuries from now even recognize a thumb drive as a data storage device; even if they do, will they be able to read out the data (stored in more or less random blocks scattered throughout the physical flash memory) if the internal controller has died? Seems unlikely. But pick up a strip of negatives or a mounted slide, and you can SEE, with your own built-in scanners, that it has images in it, and the technology to view those images at comfortable scale is easy to recreate. Printing, color correcting — well, archaeologists are pretty good at retrieving data the eye can’t even perceive, they’ll be fine.

          Of course I scan negatives — how can I show my photos on the web if I don’t? I just bought an obsolete, used scanner that pulls 30 MP out of a 35mm frame, and while naysayers poo-poo the quality of the scans, they’re good enough (if I choose) to print that 35mm frame to 16×24 inches. If I need to print bigger, I’ll most likely make an optical print (I could print B&W tomorrow, will be able to print color by end of summer) and scan that. All the detail in the negative will be faithfully enlarged, and dynamic range tamed.

          In the meantime, I’ve been waiting thirty years for digital to beat film.

          I’m still waiting.

          1. “Will people two centuries from now even recognize a thumb drive as a data storage device;”

            On the other hand, they might just stick it into “Baby’s first atom writer” by McDisney-Fisher-Price, hit the autolap button which stops when it’s zapped through to the silicon, then the read/translate button to pull the electron states.

      2. Taking multiple exposures and averaging them together improves the SNR by ~Sqrt[n]. So, if you are archiving film, slides, and prints, I recommend using multiple exposures and storing in a well-documented and lossless format (like TIFF). If you want to kick it up a notch, you can also do a flat-field correction.

  4. The word “analog” (analogue) has spawned yet another meaning.

    By the looks of the lens that the DSLR camera is attached to, I would guess that it has a fair range of analogue zoom.

    The paper that is being used as a reflector/defuser has an analogue color spectrum that I doubt is calibrated.

  5. There is still (potentially) more resolution in a 35mm film frame than can be captured by a modern DSLR. We’re getting close, but not quite there yet.

    High quality ISO 100 slide film is capable of resolving about 150 line pairs per millimeter. That works out to around 78 megapixels of digital data for a 36x24mm film frame.

    Kodak Technical Pan, designed for aerial reconnaissance and one of the highest resolution films ever made, could resolve 400 lp/mm. That works out to about 550 megapixels in the same frame.

    Lower resolution films, like say Tri-X 400 (100 lp/mm), might only have about 35 megapixels of image data. That’s in the range of modern DSLRs. When you accept that you’re also very rarely going to actually have usable information at the limit of the film’s resolution — the image is probably going to have enough blur, defocus, and aberration to cut that down significantly — it is plausible that one of today’s DSLRs could be enough to scan your family’s vacation negatives For The Last Time. But probably not a professional photographer’s slides.

    And 35mm film is the smallest format in common use. Tech Pan originally came in a 5 inch wide roll. A 4×5 view camera negative in that film could theoretically hold about 8.2 gigapixels of image information…

      1. Exactly. If you’re a pro or do lith work or high altitude recon work or deal with microfilm, you know the limitations of digital and this write up isn’t for you. You also know how to deal with this. On the other hand, if you have 1000 negative strips and find yourself with some time on your hands and curiosity in your heart, this is actually an excellent way to digitize your snapshots on deteriorating negatives (again, this is for people with negatives in a shoebox in their attic, not for people who store them at 0C and 40% RH)

        Furthermore, snapshots were likely to be taken on a camera with plastic lens elements and a single shutter speed. If you come across a winner, sure send it off to be wet-mounted and drum scanned. but for the rest of the shots, a 20MP digicam scan is overkill. They’re probably going to end up on social media immediately and forgotten about anyway.

        I did this with ~7000 slides for my grandmother. It was fast and sloppy work, but if I couldn’t have afforded the going rate for professional scanning ($.20 per slide) and she still tears up when she gets to see my grandfather when he was alive and well when they were on their adventures.

    1. Fujifilm Velvia @ ISO 50 can produce 160 lpmm only with a 1000:1 contrast test slide. It is not done in camera. At lower contrast it is less than 100 lpmm. Again it cannot achieve this in camera. It also cannot achieve this with a reflective target.

    2. I remember someone doing an information-theory analysis of DSLRs vs film years ago coming up with a figure that they had eclipsed 35mm and started competing with medium format. The lower noise helps a lot.

    3. Erm, no.
      The limiting factor with 35mm film is the lenses you have. The MTF50 of a good 35mm lens isn’t much more than 75lppm (at the middle of the frame, falling off towards the corners). The highest resolution colour film on the market tops out at around 42lppm. The best 35mm DSLRs (the 60MP Sony A7R IV) push about 52lppm. To get that level of detail you’ll need studio conditions.

      1. One of the basic rules of photography: use the largest format that is practical.
        5×4 is in the large format category. It is capable of fantastic results, but it isn’t practical for many tasks (you tend to need long exposure times, good light and the cameras are large and heavy). These days the equivalent systems would be medium format digital, that can push 150Mpixels.

  6. I’ve got a 4×5 Ektachrome that I took when I was studying. A deep red rose against black velvet. You hold it against a white background, and your eyes are just drawn in. Glorious depth.

  7. I wonder what year 2075’s equivalent of Grandma’s dresser full of photo albums will be. I somehow doubt it will be a pile of usb sticks. A snapshot of Facebook/IG? A dedicated tablet? Just lost? Thoughts?

  8. Nonsense. DSLRs have outclassed 35mm film since at least the 2003 Canon 10D. Possibly even the 2002 D60. Now, there is an argument for medium-format and larger, but only hipsters who think contrarian means cool still use 35mm film.

    1. You’re overly optimistic there. It took far longer than that. Probably the first 35mm digital to exceed 35mm film was the Canon 1Ds mk III of 2008. Resolution before that point was in favour of a good slide film. At around 20Mpixels the resolution was about equal, but at higher ISOs and less noise.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.