Improving Film Scans

I decided to build a better light source for the copy stand film scanning technique nowadays widely employed. Dedicated film scanners are a dying breed and this is a simple method to obtain high quality film scans using readily available tools. Basically an old school duplication process. It also has an advantage in that it’s possible to scan larger film formats. A dedicated medium format film scanner used to cost a fortune (probably still does if one can be found).

P.S. I’m a moron for saying light(n)ing so many times in the video. I know the difference☺

Scanning Technique

Captures are made using a Panasonic G9 camera using its high resolution mode and an Olympus 60 mm macro lens at f5.6. This produce some pretty hefty RAW files and when exported as intermediate 16 bit TIFF files from a RAW converter they are even bigger at around 400MB per file. The intermediate TIFF files are possibly dust spotted in Photoshop or GIMP before being down sampled, possibly sharpened and saved as JPEGs. The last steps are performed by a script. Finally the RAW and TIFF files are disposed of because they are just to large to keep around. To be honest most photos are not that good, film can always be re-scanned if needed.

It’s important to expose to the right (ETTR) as much as possible because negative film conversion require some extensive curve manipulation. A digital camera sensor is linear. With a 12 bit sensor the top stop of captured light will contain 2048 levels, the next one 1024 levels and so on. A B&W negative is usually recorded at around 4 to 5 stops (the scene exposure range is compressed down to this range in the negative). When this information is captured it’s advantageous to record as many light levels as possible by using the ETTR technique. Be aware that the in camera histogram and even RAW converter histograms may lie because they typically do not show the RAW histogram. To be sure about what data is captured a tool such as RawDigger must be used. This is even more important for C41 negative conversion because the orange mask cause the blue channel to be captured at lower levels then the other channels.

Example Scans

Some high resolution scans using the pictured light source.

35mm E6

Converted in Capture One using standard tone curve and minimal color or curve adjustments. White balance set to light table. I just wanted to get an idea about the LEDs ability to reproduce the slide film colors. The film in these scans is expired Sensia 100 so there are some color shifts in the film. To my eye it’s not too far of from how the film looks but slides always look better projected.

35mm C41

Converted using Negative Lab Pro. For most people it’s just not possible to convert C41 negatives using manual curve manipulations in Photoshop or similar software. It would require a very keen eye, the controls are not made for this purpose and not sensitive enough to achieve anything near correct color balance.

I still think the colors look a bit wonky but it’s probably possible to improve the color balance by small adjustments. I don’t expect the metro picture to look natural as the film is daylight balanced.

35mm B&W

Converted in Capture One using linear tone curve and a heavy S shaped custom curve. Not really the correct way to invert negatives but I find that it produce quite pleasing results in practice. I will probably do a comparison with Negative Lab Pro conversion in the future.

645 B&W

Same conversion technique as for 35mm B&W negatives. I’m quite blown away by the APX 100 negative scan. It clearly resolves 27 megapixels although it was developed in Rodinal. The Bronica RF645 standard lens is really good.


The new light source produce higher contrast scans then the old one. I think there must have been some problem with reflected light causing a veil of light over the whole film area with the old light source. It now seems slightly easier to lay the curves down to produce an end result with deeper blacks; the highlights also seems to carry slightly more detail. Color results are encouraging enough to investigate further.


Published: 2020-03-01