Microsoft patents video encoding that switches color formats mid-stream

1. One or more non-transitory computer-readable media having stored thereon computer-executable instructions for causing one or more processing units, when programmed thereby, to perform operations for a video encoder, the operations comprising: encoding video in a sequence, thereby producing encoded data, wherein the encoding the video includes: determining that color space conversion operations between a first color space and a second color space are permitted, the second color space being different than the first color space; setting a value of a syntax element, for a parameter set, that indicates the color space conversion operations between the first color space and the second color space are permitted; and encoding multiple units of the video within the sequence, wherein the encoding the multiple units includes: setting one or more coding parameters that apply for a given unit, among the multiple units, when the given unit is in the first color space, including performing motion estimation to set, as one of the one or more coding parameters, a motion vector value or performing intra block copy estimation to set, as one of the one or more coding parameters, a block vector value; determining whether or not to switch color spaces for the given unit, including performing the color space conversion operations between the first color space and the second color space, wherein the one or more coding parameters are reused when evaluating the given unit in the second color space; setting a flag value for the given unit, the flag value indicating a selection between the first color space and the second color space for the given unit; and entropy coding the flag value using context-adaptive binary arithmetic coding; and outputting the encoded data in a bitstream, the encoded data including the parameter set and the flag value for the given unit.