I’m developing a hobby project to simulate projected motion picture film images by mapping linear raw pixel values to (positive) print film transmittance values, using a composite of negative density curves and print density curves that come from the actual curves in the manufacturer’s datasheets.
It uses RawTherapee-CLI with a linear profile to do the demosaicing, denoising, and white balance and then my Python code takes over. I could have used rawpy but I don’t know how to make the denoising work and it only has two white balance options. It also doesn’t have Amaze.
I’ve got it working well for simulating Kodak Double-X and Plus-X negative b&w film along with 2302 b&w print film, at arbitrary development times (I’m interpolating between the multiple curves-per-film given in the data sheets).
By “working well” I mean that I can create good, filmic color images by applying this process to the RGB channels individually, and the parameters (to simulate varying the negative/positive development time scene exposure & darkroom exposure) work in a way that makes sense. I do not have any actual film images with these stocks to do a comparison with, and probably never will.
But I’m interested in extending the model to color reversal film and then if that’s a success I can try color negative composited with a positive print. I have the real world data for 5245, 5298, and 5386 ready to go, but the color process is a much more involved simulation (reversal should be easy though).
One thing I’d like to do is simulate different development times with color film like I’m doing with b&w, but very few color film data sheets include multiple density curves the way b&w ones usually do, not even a “contrast index/gamma vs. time” inset. I’ve found one, a Portra film, with curves for different push stops. I was hoping for a motion picture stock. Google is no help.
Anyway, I thought I should ask if anybody knew of a source for this data, either from a manufacturer or a paper in a journal or a hobbyist who’s made their own, etc.
Thanks.
