On Mon, Jan 7, 2019, at 06:49, Jim Robinson wrote: > I am trying to determine the vignetting corrections for a new lens (Laowa > 7.5mm F2.0 MFT) using using the lens_calibrate.py program and images taken > through a translucent sheet. I appreciate that with a very wide angle lens > like this it might overestimate the vignetting, but that is not the > problem. I run the program and get the lensfun.xml file which I then add > to the lensfun database (there is already an entry there for this lens for > chromatic abberation and distortion). If I then use darktable to apply this > correction to either the .pbm or .pgm files produced by lens_calibrate the > correction is very good (as it is for a 16bit tiff exported directly from > the demosaiced raw without any other processing) However if I apply it to > the original raw file (Olympus .orf) the correction is very poor with only > about half the degree of correction required being made (though it does > seem to be consistently out by proportionally the same amount). Thinking > this might be a darktable problem I then tried rawtherapee and ufraw, but > they both had the same problem with the raw file. Does anyone have any > ideas what might be going on, or has anyone seen anything similar?
It sounds like the script isn't doing the calibration on the linear RAW RGB color values... I haven't tried this new darktable-based script due to not having Gnome installed, which it needed last time I checked. A quick look at the script seems to confirm this. I don't know offhand the darktable commandline parameters for this, but the dcraw command from the original calibrate.py script uses: dcraw -4 -M -o 0 to generate such files. > Obviously I will raise this with darktable developers (my preferred > application), but given that rawtherapee and and ufraw behave in the same > way there may be a generic problem. I wouldn't bother, they'll just send you back here. ;) You should probably report this here: https://pixls.us/articles/create-lens-calibration-data-for-lensfun/ That seems to be the place for discussion related to that script. That aside, getting good vignetting data from wide lenses like that one can be tricky, so once you get the software issues fixed, you may still have some adventures ahead... be sure to test the results on some real-world blue-sky images or similar, to show up any over-correction (the likeliest problem). Good luck! -- jys _______________________________________________ Lensfun-users mailing list Lensfun-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lensfun-users
