Releases: rigaya/NVEnc
Releases · rigaya/NVEnc
3.25
3.24
- Reduce memory usage when using vpp filters.
- Fix bug of yv12(high)->p010 [AVX2] conversion.
- Add detection of 2:3/3:2-pulldown. ( #17 )
Now pure pulldown video will be detected as 23.976fps. This feature will not work when avsync, vpp-rff or vpp-afs is enabled. - Show detail of nvml initialization error. ( #16 )
3.23
3.22
3.21
3.20
- add vpp filter to sharpen edges and details. (--vpp-unsharp)
- add another vpp filter to sharpen mostly edges avoiding overshoot. (--vpp-edgelevel)
- now NVEncC will automatically select suitable GPU device to run on, in multi GPU environment.
- fix issue that crop was doubled when using crop & resize together ( #13 ).
- change "--transfer" flag names as they are in other software such as x264 ( #14 ).
smpte-st-2048 → smpte2048
smpte-st-428 → smpte428
3.18
-
now repeat field flag from avcuvid reader could be applied by using vpp-rff.
Note that this implementation is for rff=1 only. Also it could not be used with --vpp-deinterlace or --trim. -
add interlace support for --vpp-resize, --vpp-pmd, --vpp-gauss, --vpp-knn, --vpp-unsharp.
Filters will be automatically applied per field, when input is set as interlaced.
--vpp-deband, --vpp-delogo was already interlace aware. -
Fix possible crush when converting from high bit depth yuv420 -> p010.
-
add Auto Field Shift deinterlacer by CUDA, --vpp-afs.
3.17
3.16
3.15
- Add check fro NVIDIA driver version when possible.
- fix crush when run with a system without CUDA installed.
- some fixes on help. ( #11 )
- fix y4m reader corrupted when reading from pipe.
- fix y4m reader corrupted when reading high bit depth yuv422/yuv444.
- fix vpp-delogo not working on high bit depth processing.