Open strogonoff opened 4 years ago
I think you can get virtually identical results, within the error margins of the calibration.
@Thanatomanic Thanks for responding! So this works in theory and I suppose the case is closed…
Also just found a page in RawPedia about reusable processing profiles with some relevant points.
Don't take my word as gospel :-) I would love to see a real world test of your scenario.
I’ll try and post results, will be matching shots from APS-C vs. a full-frame camera and with different lenses to boot so it’d be interesting to see.
I don’t yet have the proper setup for taking a flat field so that’d be another confounding variable.
Going to use a simple single-illuminant profile and specific lighting though
On 24 Jul 2020, at 7:58 PM, Thanatomanic notifications@github.com wrote:
Don't take my word as gospel :-) I would love to see a real world test of your scenario.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
(This is just a question so I would understand if this is not the right place to ask it.)
Say I have a PP3 profile A that specifies:
1) Well-calibrated DCP profile for camera X 2) Various additional settings on top of that (curves, Lab, CIE CAM, exposure adjustments, a Hald CLUT)
When I apply that profile to DNGs from camera X under specific lighting, I get certain results consistently.
Now, let‘s say in that PP3 I replace DCP profile with one calibrated for camera Y but keep all other aspects and settings as is, thus obtaining PP3 profile B.
In theory, if I apply PP3 profile B to a DNG from camera Y, can I expect it to produce identical results as if I applied PP3 profile A to a DNG from camera X?
Or does this involve too many assumptions that it can rarely be achieved in environments with limited choice of gear?
Perhaps there is a subset of settings that could be safely assumed to produce identical results, while other settings would have to be adapted to specific camera models similar to DCP profile?
I could obtain another camera and test this myself, but I’m interested how this works from profiling theory standpoint. (Perhaps I can accidentally achieve this with those two cameras but it won’t work with others, or perhaps it should work but I am missing some key assumption.)
Assumptions
I suppose the following would have to hold: