- Apr 19, 2020
-
-
Aaron Boxer authored
it is zero by definition, since this is the first pass
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
For multi-tile decode of all tiles, clear output image in case not all tiles are decoded (for example, there should be multiple tiles in file, but there is only one tile stream. fixes #65
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
broken.jpc has been removed from test suite, as it takes a very long time to decode (although Grok can decode it)
-
Aaron Boxer authored
opj_j2k_decode_tile: if marker is found after end of tile part data, and marker is not EOC or SOT, then only return false if there are more tiles to read: otherwise don't worry about it. fixes #58
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
Corrected all rate calculations for TERMALL mode - now we don't need to increment pass rates for TERMALL.
-
Aaron Boxer authored
-
Aaron Boxer authored
correction should be zero for TERMALL, since all necessary bits are output from registers during flush
-
Aaron Boxer authored
If we don't do this, then setting from previous file will affect next file.
-
Aaron Boxer authored
2. add one to pass rate at end of block encode (except for TERMALL), to ensure that it matches actual number of encoded bytes 3. use correct lazy length computation for pass rate correction factor (p 498 from T+M)
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-
Aaron Boxer authored
-