From Fri, 7 Sep 2007 10:49:50 +0200 Cyrille Berger <[EMAIL PROTECTED]> wrote:
> First comment, it's just formating of the doc :D , it's easy to get > lost between the C++ API and the C API, I was a little bit confused > at first going from a C API then having C++ classes. Not a big deal. Hmm, ok, I'll try to split the API into the C++ API group and the C API group and will see if that helps. But alas, doxygen does not support embedded groups, so it will be a flat strucure: "lenses for C", "lenses for C++", "cameras for C", "cameras for C++" and so on. UPD: I tried and imho it looks even worse :( I'll leave it as it is now, until a more clever idea comes into my head. > Reading http://cs.ozerki.net/zap/lensfun/doc/structlfLens.html , what > is the best way to get the lens calibration data ? I would have like > to have a function which either return me the closest solution or an > interpolation, instead going throught the list in CalibDistortion and > pick the closest to the focal. I wasn't planning to make these functions public, but if you need them I can put them into the public interface. But my main idea was to have some "black box" object, this one: http://cs.ozerki.net/zap/lensfun/doc/group__Correction.html which will contain all the correction algorithms inside. When you initialize it, you hand it a lfLens and it will interpolate everything for you as well as it will prepare precomputed tables etc which will help during image correction. I can make some of these fields public (notably the interpolated correction coefficients), so that if you want you may use them on your own. But implementing 5 distortion algorithms in every client application is not something I was meaning, I rather would like to see them in one central place where they could be optimized (maybe even implemented in asm for some platforms). -- Andrew
signature.asc
Description: PGP signature
_______________________________________________ CREATE mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/create
