|
lensfun
0.3.99.0
|
Lens distortion calibration data. More...
#include <lensfun.h>
Public Attributes | |
| enum lfDistortionModel | Model |
| The type of distortion model used. | |
| float | Focal |
| Nominal focal length in mm at which this calibration data was taken. | |
| float | RealFocal |
| Real focal length in mm for this nominal focal length. More... | |
| cbool | RealFocalMeasured |
| Whether the real focal length was actually measured. | |
| float | Terms [5] |
| Distortion coefficients, dependent on model (a,b,c or k1 or k1,k2) | |
| lfLensCalibAttributes | CalibAttr |
| Calibration settings (currently unused). | |
Lens distortion calibration data.
Lens distortion depends only of focal length. The library will interpolate the coefficients values if data for the exact focal length is not available.
| float lfLensCalibDistortion::RealFocal |
Real focal length in mm for this nominal focal length.
When Lensfun speaks of “focal length”, the nominal focal length from the EXIF data or the gravure on the lens barrel is meant. However, especially for fisheye lenses, the real focal length generally differs from that nominal focal length. With “real focal length”, Lensfun means the focal length in the paraxial approximation, see http://en.wikipedia.org/wiki/Paraxial_approximation. Note that Hugin (as of 2014) implements the calculation of the real focal length wrongly, see http://article.gmane.org/gmane.comp.misc.ptx/34865. It is needed for accurate geometry transformations, e.g. from fisheye to rectilinear.
The default value usually is the nominal focal length. Only for Hugin-based distortion models, the default is the nominal focal length multiplied by the "d" parameter (i.e. linear in r) in the Hugin polynomial. This is because it is this focal length that Hugin assumes in its fit.
In practice, its effect is mostly negligible. When converting to rectilinear, it merely results in a magnification (which probably is reverted by autoscaling). When converting to a fisheye projection besides stereographic, the degree of distortion is not detectable by the human eye. Moreover, most non-fisheyes have quite accurate nominal focal lengths printed on the lens. Thus, the only use case left is the conversion from non-stereographic fisheye to stereographic. This maps perfect circled to perfect circles, so it is noticeable if the nominal focal length is rather off.
1.9.1