Releases: rigaya/QSVEnc
Releases · rigaya/QSVEnc
7.75
- Fix --dolby-vision-rpu causing error from QSVEnc 7.74 when reading rpu file.
- Fix wrong parsing of
grain_yandgrain_cfor --vpp-libplacebo-deband. - Now --dolby-vision-rpu and --dhdr10-info can be used together.
7.74
- Remove rate control mode limitation for --dolby-vision-rpu. ( #222 )
- Fix invalid value not returning error when using --dolby-vision-profile. ( #222 )
- Add option to set active area offsets to 0 for dolby vision rpu metadata. ( --dolby-vision-rpu-prm crop, #222 )
- Change log behavior.
7.73
- Improve encode performance when using --dolby-vision-rpu copy for a long input file. ( #216 )
Previously, the encode speed kept on going slower when using --dolby-vision-rpu copy, but now encode speed should be stable. - Fix muxer error copying PGS subtitles (using --sub-copy) when the input has unsorted subtitle packets.
Now is able to avoid "Application provided invalid, non monotonically increasing dts to muxer" error. - Improve AV1 output when using --dhdr10-info.
7.72
- Changed implementation of --dhdr10-info to work on Linux. ( #216 )
Changed to use libhdr10plus instead of hdr10plus_gen.exe. - Fixed --dhdr10-info not working on raw output. ( #216 )
- Fixed crush when --dolby-vision-rpu is specified when there is no dovi in the input file. ( #216 )
- Fix input SAR not passed to output in QSVEnc 7.71.
7.71
- Add custom shader filter using libplacebo. (--vpp-libplaceo-shader)
- Add deband filter by libplacebo. (--vpp-libplacebo-deband)
- Add tone mapping filter by libplacebo. (--vpp-libplacebo-tonemapping)
- Fix memory leak when using the resize filter by libplacebo (--vpp-resize libplacebo-xxx).
- Now --dolby-vision-rpu copy will automatically convert to dolby vision profile 8 when input files is dolby vision profile 7 using libdovi.
- Fix --dhdr10-info not working. ( #216 )
7.70
- Update libvpl to support API 2.13.
- Add libplacebo resize filters for Windows x64 build (--vpp-resize).
- Update ffmpeg libraries. (Windows)
- ffmpeg 7.0 -> 20240822
- dav1d 1.4.1 -> 1.4.3
- libvpl 2.11.0 -> 2.12.0
- libvpx 2.14.0
- Add MMT/TLV demuxer patch to support mmts files.
- Fix help of --vpp-smooth showing wrong value range for qp option.
7.69
- Add RGB output feature. (--output-csp rgb)
- Please note that some filters do not support this feature.
- RGB output is only supported in HEVC FF mode.
- Rename --output-csp values to match those of NVEnc.
- i420 → yuv420, i422 → yuv422, i444 → yuv444
- previous values shall still work but removed from documents.
- Add feature to copy Dolby Vision profile from input file. (--dolby-vision-profile copy)
Supported on Windows and Ubuntu 24.04. - Add feature to copy Dolby Vision rpu metadata from input HEVC file. (--dolby-vision-rpu copy)
- Avoid "failed to get header." error on some HEVC input files. ( #196 )
When H.264/HEVC header cannot be extracted, it can be now retrieved from the actual data packets. - Fix problem in audio encoding that 5.1(side) could not be encoded.
- Fix --vpp-transform causing illegal memory access error when width or height cannot be divided by 64.
- Fix seek issue caused by audio getting muxed to a different fragment than the video at the same time, due to insufficient buffer for audio mux.
7.68
- Switch from oneVPL to libvpl.
- Add package for Ubuntu 24.04.
- Add support for API 2.11.
- Add per channel options to --vpp-tweak. ( #203 )
- Add test implementation for AI Super resolution. (--vpp-resize mfx-ai-superres)
Please note this is test implementation and may not work, as the driver (at least at 31.0.101.5593 @ Arc A380) seems to not support yet.