> > On 4 Dec 2003 at 18:03, John Francis wrote: > > > Sure, you don't get full resolution for every possible combination of colours. A > > regular pattern of pure Red (or Blue) and black lines will resolve far less > > lines/mm than a black-and-white pattern. You can also get nasty aliasing > > artifacts with images with a high-frequency chroma component (just like on TV). > > > > But for real-world scenes (less than fully-saturated colours, and no chroma- > > only boundaries), or for black-and-white test patterns, algorithms that favour > > luminance over chroma will be able to resolve better than 2-pixel-wide lines. > > And this is one of the main reasons that chromatic aberrations at high contrast > boarders are so much more visible when shooting on digital media, very real- > world.
Not really. If that high-contrast edge were being properly handled, you wouldn't see the chroma bleed across the edge. In fact what you are seeing there is an algorithm that isn't making use of the luminance signal to control chroma shift.
