]> git.sesse.net Git - ffmpeg/commit
libswcale/input: use more accurate planer rgb16 yuv conversions
authorMark Reid <mindmark@gmail.com>
Sat, 3 Oct 2020 23:31:58 +0000 (16:31 -0700)
committerMichael Niedermayer <michael@niedermayer.cc>
Tue, 6 Oct 2020 15:56:52 +0000 (17:56 +0200)
commita48adcd1369131ebadd1c928fe790c72795182a7
treea30fec676173eeaa7e5fc7b89655e3d9456a101e
parent214998c55ff99586b57c1ccd34a7e6f15c4fcdb8
libswcale/input: use more accurate planer rgb16 yuv conversions

These conversion appears to be exhibiting the same rounding error as the rgbf32 formats where.
I seperated the rounding value from the 16 and 128 offsets, I think it makes it a little more clear.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
13 files changed:
libswscale/input.c
tests/ref/fate/filter-pixfmts-scale
tests/ref/fate/psd-rgb48
tests/ref/fate/psd-rgba64
tests/ref/fate/sws-floatimg-cmp
tests/ref/vsynth/vsynth1-ffv1-v3-rgb48
tests/ref/vsynth/vsynth1-r210
tests/ref/vsynth/vsynth2-ffv1-v3-rgb48
tests/ref/vsynth/vsynth2-r210
tests/ref/vsynth/vsynth3-ffv1-v3-rgb48
tests/ref/vsynth/vsynth3-r210
tests/ref/vsynth/vsynth_lena-ffv1-v3-rgb48
tests/ref/vsynth/vsynth_lena-r210