GAMUT MAPPING FOR HDR (DE)CODING

To enable good quality HDR image decoding, as corresponding SDR images which are typically defined in a narrower Rec. 709 gamut, the inventor found a pragmatic method of converting a high dynamic range image (HDR_2020) comprising a pixel having a first color (C_in) having a first color saturation, into a standard dynamic range image (SDR_709) having a peak luminance of 100 nit, in which that pixel has a second color (C_out) having a second color saturation which is lower than the first color saturation, the converting comprising: - performing a color space conversion (601) comprising applying a color matrix which maps a first RGB representation of the first color defined according to first red, green and blue primaries to a second RGB representation (RGB_sec) of that pixel color according to second RGB primaries, yielding an intermediate image (HDR_IM); - applying a reversible gamut mapping to the second RGB representation, yielding a mapped color (C_map); and - applying a luminance mapping (TOM) to the luminance of the mapped color, to obtain the low dynamic range image (SDR_709) with the pixel having an output color (C_out) which has a different luminance (L_Out) than the luminance of the mapped color, and outputting the standard dynamic range image (SDR_709), characterized in that the reversible gamut mapping comprises applying a respective mapping function (F1, F2, F3) to each of the three R, G, and B color components of the second RGB representation to respectively obtain output R, G, and B color components (Ro, Go, Bo) which define the mapped color, which respective three mapping functions comprise two linear segments (501, 502), of which one of the segments (501) corresponds to an identity transform which segment begins at a threshold value (TP) of respectively the input red, green or blue color component of the second RGB representation, and the second linear segment connects from the threshold value (TP) to a most negative value of the respective component of the second RGB representation, and wherein the threshold value (TP) is calculated by multiplying the values of the other two components of the second RGB representation than the color component for which the threshold value (TP) is calculated by respective constants (kRG, kRB) and summing those two multiplied contributions.