Photometric measurements show that, as a group, nearby Type Ia supernovae follow similar light curves and reach similar peak magnitudes (Branch & Tammann 1992). Thus, these supernovae may serve as standard candles or calibrated candles at cosmological distances. Magnitudes of local and distant supernovae, both in the same filter band, are compared using a K correction to account for the different spectral regions incident on that filter. A generalized approach compares magnitudes in different bands for the nearby and distant supernovae, bands that are selected to give sensitivity in corresponding regions of the unredshifted and redshifted spectra. Thus, R magnitudes for supernovae at z ~ 0.5 are compared with $B$ magnitudes of local supernovae. We compute these generalized K corrections over a range of redshifts and bandpass pairs and discuss their advantages over the traditional single-band K correction. In particular, errors near maximum light can be kept below 0.05 mag out to at least z = 0.6, whereas the traditional K correction is less accurate and can be difficult to determine beyond z > 0.2.
When comparing the broad-band fluxes of low- and high-redshift objects, we have to account for the fact that observed filters are sensitive to different wavelength regions of the rest-frame emission. If the redshift difference is large enough, it makes more sense to compare the blue-filter flux of a low-redshift object with the red-filter flux of a high-redshift object, then one also has to account for the different calubrations of the two filters.