Obtaining good colors from a film scan

Hi,

I’m interested in digitizing my color negatives. Unfortunately, I don’t own a film scanner, but I tried scanning them in two different places. The problem is that at each lab the scans come out partially good, but in complementary ways - one scan (done on a Noritsu minilab I think) has good colors (sometimes even a little over the top, but I like that), but apparently the machine boosts contrast which sometimes results in crushed shadows and blown highlights; the other scan captures the dynamic range of the negative very well, but the colors are dull.

Below is a typical example


and below that an even more extreme one (both are Portra 400 pushed one stop).


Is there a way of recreating the vibrant colors of the minilab scan by, I don’t know, applying some known .icc color profile to the TIFF of the flat scan? I’m aware that one can try to play with the contrast and colors by e.g., manipulating RGB channels separately “to taste”, but 1) this seems like a lot of work, 2) it’s hard to make this work consistently across different frames, 3) in any case, I’d have no idea how to reproduce the lovely quality of colors typically associated with films like Portra (in all my attempts the colors are still somewhat off).

1 Like

The first scans seem fine to me. If they are consistent you could try to make a style and apply this to all scanned images.

I like the first in each pair.

These are JPEGs, so only 8 bits per channel, and probably lossy compressed. Can the lab supply images that have not been through JPEG compression? If they can, you can do more with the results.

You can certainly give the images more “punch”, using any editing tools. If you don’t take it too far, I expect you can come up with a formula that can be used for other scans. For example: an S-curve to increase contrast and saturation in mid-tones, and an unsharp mask to increase local contrast and saturation.

magick 486327f85834f8389a80d5fbd95f324477f1a276.jpeg -sigmoidal-contrast 5,50% -unsharp 0x10+0.5+0 x.jpg

I have taken this too far for my taste, but it shows what we easily can do with very primitive processing.

2 Likes

Recreating the same effect over multiple images seems easy. Just make a ‘style’ of the edit, which you can then apply to a whole collection in the light table view and export.

The flat scans seem fine for this since they allow you to do the saturation and contrast like you want.

Remember, film has no ‘correct’ look. Only the look you want / like.

But doing a quick vibrancy / saturation edit in color balance rgb in darktable seems easy enough, and there is a contrast slider there as well.

More technical, but you can also use ImageMagick or another tool to merge all the scans into one collage / contactsheet, of small versions of the scans.

You then have one image to edit and you can see what your changes are doing to all the frames of the roll. Save presets of your edits, apply them to the original scanned pictures and export.

the curve used in gimp fro the roughly match
mkmkfilm.7z (8.4 KB)

@mkmk have you considered digitizing with a DSLR or mirrorless camera so you have more control over the process?

2 Likes

@mkmk The first ones seem to be a much more faithful/neutral scan of the negatives. The second ones are way to heavily processed. In either case, like others say, you should try to get 16 bit TIFFs from the first company (just ask them about it).

Thanks a lot. Can you say a few words about how you arrived at the gimp curve? It seems to recreate the look fairly closely, although still seems more reddish than the lab scan. Sorry if this is a very newbie question ;).

Thanks everyone for the input! My initial question was not completely precise, so just a few words - I know there is no really such a thing as a “correct” look for a color negative scan, my question was more along the lines of “is there a way of reproducing, from a fairly neutral scan, color rendition (but not necessarily the overall contrast) typically associated with a Noritsu lab scan?” (other than trying to match by a custom-made curve for each film roll). The lab scans posted are much more heavily processed, but consistently so (I’ve scanned many rolls of film at the lab) and have generally pleasing color.