Hi, I've done the simple vignetting calibration (focusing to infinity) for these lenses:
LUMIX G VARIO 12-32/F3.5-5.6 : Panasonic, Micro 4/3 System, 2, 4:3
LUMIX G VARIO 35-100/F4.0-5.6: Panasonic, Micro 4/3 System, 2, 4:3
It's the first time I tried that, so please do a sanity check on my values. I used a milky glass and put the camera underneath it pointing at the ceiling lamp (ca 2 meters above). The camera was handheld with lens pushing against the glass plate. Lens+Sensor image stabilization was on (no digital stabilization).
The tca and distortion are taken from the existing DB entries.
I used the calibration script from Windows, and it needed a major rearranging: when using multiprocessing on Windows, you need to move most of the "effective" code into if __name__ == "__main__":, but leave the function/class definitions outside. This is something that can be done quite easily, but I understand it'd affect readability of the script, when the functions are moved away from the places where they are used. There's also a system program called convert.exe which serves for converting FAT partitions to NTFS, and the calibration script is trying to call this one instead of the ImageMagick executable. Quite dangerous :) If you'd be interested in seeing the working script, I can share it.
Thank you, I included your data! Pointing towards a lamp is not optimal, however, because it may generate a non-uniform brightness behind the diffuser.
I can also include your calibrate.py enhancements.
Thanks, here's my enhanced calibrate.py. It probably needs some polishing for OS compatibility.
I'm afraid that the diff to the current revision of calibrate.py (at https://github.com/lensfun/lensfun/blob/master/tools/calibrate/calibrate.py) is way too large for me to review. Can you create a patch against this current version?
Done ;) https://github.com/lensfun/lensfun/pull/291