Working with scanned negatives

I’d really love to use Filmulator more but I have a specific use case. I have lots of scanned negatives in raw format, but as Filmulator doesn’t have an invert feature, I have to use other software to do the inversion. The trouble then is that I usually save the inverted photo as a TIFF, but Filmulator can’t import TIFFs.
Is there a workaround for this or any planned features that might help with this situation?

It is possible to extract the raw image using dcraw's document mode, process the file (in your case invert it) and then repackage it in a DNG using exiftool, which Filmulator may accept.

Have you considered using “Film Negative” in RawTherapee? I think it’s only in the development version. I do a lot of scanning with a DSLR in raw, mostly colour transparencies, B&W negatives and some colour negs. I’ve only recently tried Film Negative and am impressed.

At one point Filmulator took in linear TIFFs, but when I made the GUI I never added that functionality back in. It would be possible to add that back.

I’m not sure what I want to do about inverting DSLR scans of film. Part of me wants to shout “You already shot film, why do you want to simulate film on it again!?” The other thing is that a lot of the look of Filmulator is really baked in and not adjustable, and I’m not sure how well it will work on something very different from the extreme linear response of a camera sensor.

Personally for now I think I’ll leave inversion out, and add linear TIFF importing back in. Shouldn’t be too hard, plus it would let you import panoramas and such, though it’s cumbersome to work with linear files.

Could you upload some samples of your inverted negative TIFFs for me to try out? Ideally, exported with linear gamma in 16-bit integer.

Thanks Carlos. I understand why you wouldn’t want to add an invert feature to Filmulator, it kind of goes against the philosophy of it. I’m building a digital archive of photos going back decades and some of those shots were taken on terrible cameras and with poor photography skills, but I still want to keep many of them for memories. I’m drawn to Filmulator because of it’s simplicity (in terms of workflow) and it will be much quicker than using RawTherapee or Darktable. I really enjoy those two other programs, but I’ll use them more for when I really want to take time on a high-quality photo that I’ll use for more than just archive purposes.

Adding linear TIFF to Filmulator sounds like it would be a nice option for this. I’m happy to provide samples. Are you wanting just the straight inverted TIFF, or with white balance, levels, crop, etc. applied? In other words, what’s the baseline you want to work with? I can send several samples in different states if it helps you.

Hi. Yes, I have tried the Film Negative feature in RT and it’s probably going to be my solution going forward, at least for the initial inversion. However, I’m still fairly new to RT and I find the features a bit cumbersome for quick developing. I’m much more familiar with darktable, but I don’t like the Invert module and colour management in darktable for this use case.
I’m liking the idea of Filmulator for quickly developing hundreds of negatives for archive purposes.

Thanks for the tip, I’ll look into this. Just out of curiosity, is this linux only or would it work on Windows?

Ask @snibgo. He gave me the instructions on how to do it.

Instructions on what?

I’ve never worked with scanned negatives, or Filmulator. If I had raw DSLR photos of negatives, I would start by demosaicing them with dcraw, and see what they look like. They might need transfer curves in addition to negation; I would do that in ImageMagick.

Filmulator look interesting, but designed to make DSLR look like images from film. In this case, we already have images from film, so Fimulator might over-egg the pudding. However, we are in a sense simulating the printing of the negative on paper.

I would start with black-and-white negatives. Colour negatives might need different processing for three colours, and of course have an orange mask.

The processing could be done before demosaicing, using dcraw and ImageMagick and exiftool to build DNG files that are demosaiced. I don’t know if that would give better results. For initial experiments, I would keep it simple: dcraw to demosaic, and ImageMagick to process.

I would use Windows but it could be done on any shell. I expect any scripts would be simple.

Sorry, it was a reply to a reply. Context:

I @-ed you for your PM on this raw to DNG process.

I have published Processing Bayer pixels. This shows how to use dcraw to extract mosaic data, and exiftool to re-build a raw DNG file. It also shows how we can split the mosaic data into four images (R, G0, G1, B), and how we can process this with ImageMagick, with a simple de-noising example.

The page uses Windows BAT commands. They are trivially translatable to bash.

2 Likes

After white balance and levels, probably.

Here’s a selection of 16-bit TIFFs: https://drive.google.com/open?id=1rGblg5F49BFiEUGOHsASss83nDnQjZ_j

I used RawTherapee to rotate, crop, invert with Film Negative, and then do a rough white balance. I didn’t do anything else because the idea is to do most of the developing in Filmulator. If I’d spent much more time in RT, then it would probably make more sense to do everything in RT.

So basically, there’s still a fair amount of developing still needed.