Let me say first of all that I'm just one of the monkeys who makes profiles from the contributed images when I have time, and don't have a profound grasp of the optics or solving algorithms, so take this with a grain of salt, but... based on empirical experience trying to fit corrections to images from a fair number of lenses, my hunch is that while "pretty close" approximations of the center offset might produce subjectively acceptable results for distortion and vignetting (particularly where the corrections are not large), this might not be the case for TCA. Good correction for this seems to be sensitive to a correctly centered image circle (or, I suppose, an accurate offset adjustment). An interesting flipside to this which occurs to me is that it would be theoretically possible to use this as a method of *obtaining* accurate offset values (where the amount of TCA is sufficient), but I don't know if this is implemented in any existing software. Other than that, I'll leave an authoritative answer to somebody with a better understanding of both the theory and real-world behavior of shift lenses.
-- jys On Wed, Apr 4, 2018, at 18:22, Robin Kramer wrote: > There is a center parameter in lensfun, it seems to work OK to just > create a version of the lens that I had previously calibrated the > distortion parameters then add a y value. > > Hugin lens calibration will also calculate the parameter eg "d", however > you have to give it a starting point that is pretty close. It's not > intuitive, but it seems to be in pixel units. > > I think just using the lensfun with a couple of choices will be good > enough, because in practice, I won't know the precise shift values, so > would be guessing anyway. > > Conceptually, should the other parameters be the same? > > Sincerely yours, > > Robin > > > On Apr 4, 2018, at 12:11 AM, junkyardspar...@yepmail.net wrote: > > > > > > > >> On Tue, Apr 3, 2018, at 15:38, Robin Kramer wrote: > >> Is there any support for lenses with movements. I'm especially > >> interested in shift. I have a lens with a shift adapter. When I load > >> the straight image, the hugin gui calibrates the lens just fine, but > >> when I load the shifted calibration chart images(all with the same > >> movement), it seems to hang, or at least take a very long time to > >> converge. I'm not sure how to interface this with lensfun, since I > >> suspect another parameter is needed. > >> > >> I use shift often it's a very nice effect, for making trees look > >> straight, or capturing the sky without looking up. > > > > Unfortunately, these types of lenses don't play nicely with lens > > corrections. There's some discussion of this here: > > > > https://github.com/lensfun/lensfun/issues/299 > > > > -- > > jys ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Lensfun-users mailing list Lensfun-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lensfun-users
