avcodec/jpeg2000: Move gainb handling into the quantization code
authorMichael Niedermayer <michaelni@gmx.at>
Sun, 21 Jun 2015 23:30:33 +0000 (01:30 +0200)
committerMichael Niedermayer <michaelni@gmx.at>
Sun, 21 Jun 2015 23:36:05 +0000 (01:36 +0200)
commit1c495b0bf690995c45f79f4f19500921e14ec78a
treed5ef046b1289363a74f1d54e8ee609ec8a6649a4
parent6504047f82866fdc0e2a5422ad7fecb44740e9d6
avcodec/jpeg2000: Move gainb handling into the quantization code

thats how the specification defines it, this also improves numerical
accuracy of the integer wavelet implementation. It otherwise should
be equivalent, in case of overflows this can be reverted.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
libavcodec/jpeg2000.c
libavcodec/jpeg2000dwt.c
tests/ref/fate/j2k-dwt
tests/ref/fate/jpeg2000-dcinema
tests/ref/vsynth/vsynth1-jpeg2000-97
tests/ref/vsynth/vsynth2-jpeg2000-97
tests/ref/vsynth/vsynth3-jpeg2000-97
tests/ref/vsynth/vsynth_lena-jpeg2000-97