Microsoft patents video compression trick for smoother streaming

1. In a computer system that implements a video encoder, a method comprising: encoding a picture, thereby producing encoded data for the picture, wherein the encoding the picture includes: determining a set of multiple block vector (“BV”) predictor candidates for a current block of the picture, the current block being compressed with intra-picture coding in merge mode, the set of multiple BV predictor candidates including one or more actual BV values of previous blocks in a neighborhood around the current block, wherein the determining the set of multiple BV predictor candidates for the current block includes: determining that a given BV predictor candidate is redundant with one of the set of multiple BV predictor candidates for the current block; and replacing the given BV predictor candidate with another BV predictor candidate of the set of multiple BV predictor candidates for the current block; selecting one of the multiple BV predictor candidates to use for the current block; encoding the current block using intra block copy prediction with the selected BV predictor candidate, wherein the selected BV predictor candidate indicates a displacement to a region within the picture; and outputting, as part of a bitstream, the encoded data, wherein the encoded data includes a merge mode index value that indicates the selected BV predictor candidate, and wherein the encoded data further includes one or more syntax elements indicating that the current block has been compressed with intra-picture coding in merge mode.