Scanning negatives and transparencies with a DSLR

I’ve done a lot of DSLR scanning with both my transparencies and negatives and those of my father, Harry Rowed, a well-known photographer active from the mid 1930s through the 1970s. In March 2018 I’ll be giving a workshop on this at The Camera Store in Calgary. To promote this, the store had these three images of my dad printed large to display in the store.

These images, along with seven other historical images my dad took, are also on display at the Columbia Icefield Centre in Jasper National Park. As the images need to be printed very large, I took four shots of each negative and stitched them together, giving final pixel dimensions of around 8500 x 11500. The entire workflow was done in Linux and open-source apps.

Rawtherapee…Invert and process the raw files
Hugin…Stitch the images in mosaic mode
Gimp…Clean up scratches and add a bit of additional sharpening and local contrast enhancement
Phatch…Reduce the file sizes and add copyright text for email etc


Great pictures, and a job well done on digitising them! Can you explain what your setup was to take the pictures? Especially how to get the negatives aligned with the camera sensor, and in proper focus?

I have digitised my grand parent’s collection of positive slides, and the hardest part was to actually shoot the pictures in focus. The post processing had its own challenges because of the degradation…

1 Like

Wonderful set of photos, especially the middle one! I’d love to hear more about the process.

I’ll post some photos of the setup later, but in the meantime a couple notes about the photos. The young woman climber, Margaret Stock, and the photographer were shot in 1939 on 3x4 inch B&W film (maybe Speed Graphic). The man with the pipe (Fred Brewster, a legend in the Canadian Rockies) was taken sometime in the 1940s on 6x6 cm film (probably Rollei twin reflex).

The woman died in 2010 after a full, interesting life. I was able to connect with her youngest son a couple of years ago and we stay in touch.

I don’t know the identity of the man with the camera.

Due to the size required for the Columbia Icefield display, each photo took hours of work to prepare. I needed to use the healing or clone tools on every tiny piece of dirt, scratch or even imperfection in the film emulsion. There were 10 images for the display and by the end I was getting sore tendons in my hand from all the mouse clicks. The largest image was intended to be printed at 11 feet wide, but in the end they cropped the images into circles.

My scanning setup is with a Nikon D800 for medium and large format film, and a D7100 for 35mm. I currently use a Nikkor 60mm f2.8 G micro lens and a Sigma 70mm f2.8 macro. The camera is mounted on a copy stand with the backlighting for the film either flash (diffused through white plexiglass) or an LED light pad. There are two ways to align the camera and film - a circular bubble level or a mirror. The mirror is faster.


Here is my current setup for using flash with 35mm slides. Note it also needs a continuous light for focusing which is shut off during the exposure. The spacing between the slide and the diffuser is provided by an old picture frame. A small mirror is used to align the camera and film. The centre of the camera viewfinder should be right in the centre of the reflected image of the lens. A circular bubble level can also be used, but it takes a bit longer as you first need to level the copy stand and camera, then the film holding apparatus. The mirror means you only have to adjust the film holder.


Interesting setup. I wonder if you did tests with infrared scratch removal by using an infrared light source for a second exposure? Or flash plus IR filter. It would not work with most b/w film, but probably with some. I am especially interested in difference in focus between both. And if it works in general for such a setup.

I’ve been trying to find out if this is possible but so far no definite answers. That said, a good cleaning does wonders. For 35mm slides I use a bulb blower, then press on 3M transparent tape and pull off (don’t rub across the film surface or it will leave residue), then another blow with the bulb blower. For mediium and large-format negs (not mounted) I use a blower, Kim Wipes or anti-static cloth, then again with the blower. For badly scratched slides you can always wet mount using Kami, Edwal No-Scratch or a light mineral oil. Kami is nasty stuff and I don’t use it anymore. For bad scratches, wet mounting is far more effective than infrared cleaning.

Having a diffused light source through white plexiglass will get rid of all but the worst scratches, much the same as a diffusion enlarger when compared to a condenser enlarger.

I don’t normally clean up the images at 100% size on screen. Just clean up what’s visible at screen size, then do a more thorough clean up on demand, i.e. when an image is needed for printing large. The screen size cleanup is generally quick work, maybe a couple of minutes, after the blower, tape and cloth methods.

1 Like

Thanks for all the info. I am using only a bulb blower, but I wonder what 3M transparent tape you are using. I only know the sticky tape that uses some glue, but I don’t think that’s what you are talking about. Is it something electrostatic?

If you have the opportunity to test with infrared light source, I would really be interested in the results, since used with a scanner this technology is a miracle. It would really be great if it could be adapted to a camera “scanning” workflow as well. But I did not find information on how much the defocusing due to different spectrum is a practical issue or a myth, and if such a technique would lead to some reasonable results. But since I have no test setup, I did not have an opportunity for tests yet.

That is 3M Scotch Matte Finish Magic Tape. Yes, it is sticky but I haven’t seen any evidence of residue if it’s pressed on, then pulled off without moving the tape across the surface of the film. I used to use film leader tape, which was used to hold film to the leader as it went through processing machines, but now it’s old and not sticking well enough. I’d be open to other suggestions for this.

Here’s a bit about dust and scratches.
It’s not infrared, and I’d love to see an infrared solution. That said, if the film is in decent shape, and it’s well cleaned, it’s not as much of an issue as it might seem.

  1. Bring in photo
  2. duplicate layer
  3. One the upper layer apply “remove hot pixels” filter with settings about 6 or 7 on both. This will get rid of the small specs, but not the larger ones.
  4. That also loses a bit of fine detail, such as in hairs or eyelashes. So on those areas use a mask to expose the image underneath.
    This is actually very effective, especially for scanned slides that have a lot of small specs of dirt. It’s still necessary to clone or heal the larger ones but that doesn’t take as much time.

Chris, do you have any thoughts on how infrared cleaning would work with a DSLR? I’m guessing this is a 2-step process:
(1) using infrared to identify the defects and creating another image that shows these defects and nothing else.
(2) an process that would automatically apply something like a healing brush at each defect.

Vuescan, for instance, allows you to save the infrared as a separate channel in a raw file so you can process it again with different cleaning settings without re-scanning the original film. it would be great to have a similar option for DSLR scanning.

I documented some of my findings along the following thread (scroll down a bit to see some example images).

The photos I am scanning are old family photos with some of them already in a very poor condition, therefore infrared is particularly helpful since it separates scratches from other degradations.

This is what I am doing, the process and how it came up is documented in the other thread as well. I am using G’MIC for the tasks, especially the inpainting algorithms that are available in G’MIC lead to much better results than the vuescan algorithm (a comparison is available in the thread as well).

I think it could work similar with a camera by taking a second photo with an infrared light source. But there might be some defocusing due to the different wavelength that passes the glass of the lens. But maybe that’s not an issue, depending on the amount of blur added. Would be interesting to do some tests, but at the moment I am lacking several parts of the setup.

However, not a good option for most b/w images since the silver particles may block the infrared light. So probably not a solution for you, but for colour images it can be magic.

As you say, IR cleaning doesn’t work for silver-based B&W films, but it would still be very useful for colour films. When I have a bit more time I’ll read your other thread in more detail and try to follow the steps with a colour scan. Thanks.

Chris, do you know if the IR channel would show defects on both sides of the film, or would it require two IR shots, one on each side?

I cannot observe any difference on if the film is lying upside down or not (I had several scanned upside down by accident since for old film it is often not easy to detect the emulsion side correctly), so I would conclude that it shows defects on both sides. As far as I understand it is a transmitted light technique. But this is only out of experience, I did not do a direct comparison.

Wouldn’t the IR filter in the DSLR interfere with that?

I am not sure. Interfere, of course, but it may be nevertheless feasible, tt may depend a little on the actual frequency and on the power of the infrared light source.

There are “screw-on” filters sold that block visible light to allow for pure infrared exposure (marketed as “infrared filter”) which give the impression that the cameras do not entirely block parts of the infrared spectrum. For the application of film scanning, doing the infrared exposure at much higher iso values may be feasible to account for losses due to attenuated infrared spectrum. And/Or to use more power at the emitter. Furthermore, one would not have to use the entire dynamic range of the sensor, since the goal is a b/w mask anyway. To get a reasonable feeling, tests are required, and since I do not own everything which is required to do it myself, I try to collect indicators that allow for better understanding of the whole topic.

Thanks for showing your setup, it was quite similar to mine :slight_smile: I only used an Artograph LightPad as my light source, and my camera was hooked up to my pc for tethered shooting. I used DigiCamControl, which has a really nice ‘focus’ functionality that helped a lot in setting the focus just right. This was especially useful when the slides had been slightly damaged or wrinkled.

Still a question: how did you take multiple shots of the same slide? Your setup looks quite immobile.

I’ve only done a bit with tethering, mostly with Entangle, but should try DigiKam for that as well. Have you compared the two?

The film holder I showed in the photos is only for 35mm and there’s no reason to shoot multiple shots as even 24 mp exceeds the useful information in the slide. For medium and large format film I can move the holders around to get multiple shots. But even that 35mm film holder can move around easily.

I’ve been experimenting with light pads as well. While they are very convenient I’m finding two issues. First, my copy stand is not as sturdy as some, such as the Manfrotto. This means there is slight vibration which makes the images not quite as sharp as they are with the flash. Second, with one of the LED panels, the CRI (colour rendering index) is not that great, resulting in slight shifts of some colours. I have a better light pad on loan from the store and it seems more accurate for colour, but there’s still the sharpness issue. With ISO 100 and f8 on the lens, the shutter speed is 1/15 second, which is the worst shutter speed for vibration. I have an LED video light which is much brighter and gives me 1/250 second, thus sharper images. Still working on this as the light panels are more convenient.

1 Like

I believe all the tethered shooting options use gphoto2 as the backend, so they should all be fairly similar. I was happy with Engtangle and thought it was a nice solution.