An adaptive circuit and method for separating luminance and chrominance components from a composite video signal by deriving three input lines from the composite video signal, determining whether any luminance similarity exists among the three input lines, and then selectively enabling a component filter based on any luminance similarity. If no luminance similarity exists among all three of the input lines, then a subtractive comb filter is enabled to maintain high vertical luminance resolution. If luminance similarity exists among all three of the input lines, then an additive comb filter is enabled. The additive comb filter performs three-line averaging when a high degree of similarity exists among all three consecutive input lines to minimize cross-chroma artifacts on lines that are similar. Chrominance similarity among the three input lines can also be determined by generating first and second chrominance values using different pairs of the three input lines, computing a difference of the first and second chrominance values, and comparing the chroma difference to a threshold value. If no luminance similarity exists between any of the three input lines or no chrominance similarity exists (i.e., there is vertical chroma transition), then a notch filter incorporated into the subtractive comb filter can be enabled.