they can do all that can be encoded in a VLog 3D lut, thus the result of the full spektrafilm computation for a fixed set of paramters, minus non-local and stochastic effects (halation, scatteing, grain, diffusion filters, diffusion of couplers, etc…). i would say it is like having the ART implementation in camera but only with fixed presets. white balance works fine i would say, because by design spektrafilm was designed to preserve 18% midgray input->ouptut.
2 Likes
Ofcourse this was along the lines of what @Thomsen suggested . I even saw a very WIP port of filmsim and Spektrafilm but we shall see if the dev decides to finish it and publish it . (Should be foss that one ofcourse). I believe those Androids have benefited a lot from all these amazing projects… pretty much Magic Lantern territory.
This looks really nice! mind sharing your json file of this one?
Looks really interesting, unfortunately I haven’t been able to get your LUT generator or VKDT working properly on my mac. It’s probably user error 
I just get this message despite having the proper permissions, etc.
very sorry but i don’t know how to do this. i basically never use the terminal unless i have commands to paste into it.
(post deleted by author)
To create a persistent installation:
uv tool install git+https://github.com/andreavolpato/spektrafilm.git@dev
To start it:
spektrafilm
To upgrade it:
uv tool upgrade spektrafilm
Just copy and paste those into your terminal.
3 Likes
Try this in terminal:
cd /path/to/your/downloaded/Spektrafilm-LUT-Generator
chmod +x launch_mac.command
Then try run again 
Alternatively
conda activate spektrafilm
python “/path/to/Spektrafilm-LUT-Generator-main/spektrafilm_state_to_lut_gui.py”
thank you so much! i can’t believe it was just that easy.
how do I export as a high resolution file? when I press the save button all I get are very low resolution files.
did you press scan before to calculate the high resolution image?
That is the order of things I would recommend (if one needs to use Photoshop), yes.
okay, so far this works as expected
no wait. the sensitivities are windowed, both the cmf and the film ones. so by windowing some more, you’re changing the ratio of r/g/b response (the window falls off really soft in the ir range).
fitting this extra error correction to some specific spectral shape seems arbitrary, but you’re using quite a relevant set of spectra here. and your plots with errors mostly in the dip-shape regions are kinda convincing. in the interest of reducing the amount of extra data/correction terms we have floating around here… would it make sense if i tried to introduce some fixed/static windowing and re-optimise the spectral lut with that in the loop? i’m assuming it wouldn’t change the behaviour of the xyz roundtrip much but would provide us with windowed spectral upsampling. this wouldn’t address metamer mismatch, but now i’m curious…
It was said above, but it’s becoming a bit hard to follow the different discussions in this thread.
Perhaps Spektrafilm — and its implementations in ART and vkdt — soon deserves its own category, @paperdigits and @hanatos? People who are not so close to development and Linux seem to be discovering Spektrafilm (which is great!) and a support thread, for solving installation issues and things like that, might be a good idea.
3 Likes
oh sweet I was totally missing that part. Is there a way to save it as 16 bit tif or is JPG the only option right now. when I try to save it I don’t see an option to save as tif only jpg. thanks for your help on the scanning part!
Sure, if @arctic would find it useful.
1 Like
The TIFF option is gone. If you want lossless, use PNG (8-bit only, I believe) or OpenEXR (more bits, probably).