FYI I just extracted all the data from the tri-x spec sheet. TX/TXT/TXP, all development times, small tank - large tank, etc. That’s 14 film profiles.
Plus all 7 grades of polymax fine art paper (g24).
You might want to use it. The json files are here. Though the easiest way is to run the import script
./import-to-dt.sh $PATH_TO_DARKTABLE
The code that generates it, if you want to vet it or build on top of it, is here. Yep, AI slop, but the results are legit.
Tested in darktable.
A note on TXT/TXP: borderline unusable for the average photo, but for indoor portraits with somewhat controlled lighting, its declared use case, they beat the better known TX hands down.
I think there should be an option to choose paper grade where available. makes a big difference and it’s a real pivotal choice during the film development process too.
I also think that if processing entire spec sheets like this is possible, and the result is a dozen or more film profiles, that will need some UI/UX changes. It can’t all just go in a flat list…
Sounds interesting but maybe target upsteam instead of a darktable module implementing upstream. I don’t think we should offer features the official spektrafilm code does not.
I am, there is a separate file for spektrafilm proper. i didnt test that, but the QA plots are generated from it, and if the darktable one works, i dont expect issues.
Btw, the whole split could be unnecessary, if collapse_to_darktable_pack() wouldn’t " widen BW fields from spektrafilm’s genuinely single-channel source format (log_sensitivity/channel_density/density_curves_model all 1-column) to the 3-column shape darktable’s C reader (spektra_sim.c) always expects regardless of channel_model." (source: AI)
sound like an unnecessary and perhaps unintended deviation.
I don’t think I am proposing that. Having e.g. a dropdown for paper grades or film subtype, iso, etc. doesn’t change anything, it all remains 100% spektrafilm if it was 100% spektrafilm before.
It’s just a UI/UX change. Anyway, up to you of course, Im fine maintaining a private fork that serves my sense of ergonomics. I just think it’s a good idea
I must confess that I do not understand what you have done from this forum post alone and the AI slope readme is waaaay to long to expect anyone to read…
Are you intending to provide a spektrafilm PR with the new data? And a PR or clear example of what to change in darktable that already exists in spektrafilm? Because i think that is the proper way forward.
well, I’m offering. I did it for my own purposes, but it’s evident that it might have value for others, so I’m putting it out here. Whether I want to go through the PR process I wasn’t thinkign about yet. But, as said above, because of the nature of the data, I think spektrafilm too would benefit from it most if it would accommodate for it UX-wise.
Maybe it’s too early for that, with one film stock family and one paper family (i.e. different grades.) and it will make more sense once i have more stuff. Either way, questions that are not up to me.
They are ready to go as is. The links pointing to the files are there, just ignore the rest of the repo I just threw it to a project that already did some different digitization, which might be confusing.
For darktable, they just need to be copied to where you keep data for the spectrafilm module. Actually, i think some other minor action was needed to register them, but you know what needs to be done on that front better than i do.
(that being said, now after your fix/change the darktable files probably won’t work anymore, unless you use them with the HEAD^ version. hopefully the spectrafilm ones will. I’m starting to work now so I won’t be able to regenerate them right now)
paper now conform to ISO 6846:1992 / ANSI PH2.2-1972 paper speed criterion
now it can be seen why such comprehensive digitization needs an accommodating UX rather than a flat list of options. It’s still just half dozen of films (if we treat the Tri-X subtype as one)
Ideally, if possible, it seems that some of these properties (like paper grade) should be simulated and controllable via sliders in the UI, rather than added as separate profiles. As you imply, the list gets pretty crowded.
