Software » Useful Tools » QSVEnc 7.79

7.79
Fix some case that audio not being able to play when writing to mkv using --audio-copy.


2025.01.08 (7.78)
- Avoid the issue that the width of "tkhd: Track Header Box" (L-SMASH boxdumper) and "Visual Track layout" (mp4box -info) output by mp4 muxer becomes 0 when the SAR ratio is not set (e.g. 0:0).

2025.01.06 (7.77)
- Fixed an issue where some parameters did not work properly with --vpp-libplacebo-tonemapping.
- Fixed an issue where the trim position was shifted when encoding ts files etc. from the middle and OpenGOP was used.
When calculating the trim correction amount, the case where the previous frame comes after the first keyframe in OpenGOP was not taken into account.
- --trim now accepts Aviutl trim export plugin notation.

2025.01.03 (7.76)
- Fixed the "Fixed the issue where --dolby-vision-rpu was broken when reading from a file" in 7.75, which had not yet been fixed.
- Improved startup speed from the second time onwards after starting the encoder PC.
- Improved automatic GPU selection with --device auto.
To make more aggressive use of other GPUs.

2024.11.24 (7.75)
- Fixed --dolby-vision-rpu being broken when reading from a file.
- Fixed the issue that grain_y, grain_c were not read in --vpp-libplacebo-deband.
-Fixed grain_c help for --vpp-libplacebo-deband.
- Supports using --dolby-vision-rpu and --dhdr10-info together.

2024.11.22 (7.74)
- Fixed an issue where non-target profiles could be loaded with --dolby-vision-profile.
- Added an option to reflect the cropped letterbox part when using --dolby-vision-rpu. ( --dolby-vision-rpu-prm crop )
- Removed mode restrictions for --dolby-vision.
- Minor changes to log display, etc.

2024.11.12 (7.73)
- When using --dolby-vision-rpu copy and encoding for a long period of time, the encoding speed will now maintain a constant speed, which is a noticeable decrease in speed.
- Improved output when using --dhdr10-info when exporting AV1.
- An error occurred when the subtitle timestamps of the input file were out of order. This has been changed so that the error is resolved by re-sorting.
- Avoid crash when running with --vpp-tweak set to do nothing.

2024.11.02 (7.72)
[QSVEncC]
- Changed the implementation of --dhdr10-info to support operation on Linux.
Changed to use libhdr10plus instead of hdr10plus_gen.exe.
- Fixed an issue where an error would occur when specifying --dolby-vision-rpu if the input file did not contain dovi.
- Fixed --dhdr10-info not working when outputting raw.
- Fixed an issue where the SAR ratio of the input file was not reflected in 7.71.

2024.10.27 (7.71)
[QSVEncC]
- Added libplacebo debanding filter. (--vpp-libplacebo-deband)
- Added tone mapping filter by libplacebo. (--vpp-libplacebo-tonemapping)
- Added a filter using libplacebo's custom shader. (--vpp-libplacebo-shader)
- When using --dolby-vision-rpu copy, if the input file is dolby vision profile 7,
Automatically convert to dolby vision profile 8 using libdovi.
- Fixed --dhdr10-info not working.

2024.09.24 (7.70)
[QSVEncC]
- Updated libvpl to support 2.13.
- Added a resize filter using libplacebo. (Windows x64 version)
- Updated the ffmpeg library used. (Windows version)
- ffmpeg 7.0 -> 20240902
- dav1d 1.4.1 -> 1.4.3
- libvpl 2.11.0 -> 2.12.0
- libvpx 2.14.0
- Included MMT/TLV patch (allows mmts to be read)
- Fixed incorrect help for --vpp-smooth.

[QSVEnc.auo]
- Supports wav output longer than 6 hours.
- Supports output of audio longer than 6 hours using ffmpeg audio output.
- Added opus encoding using ffmpeg's libopus.
- Updated VC runtime installer.

2024.08.20 (7.69)
- Added RGB output function. (--output-csp rgb)
- Added Dolby Vision profile copy function. (--dolby-vision-profile copy)
- Added the ability to copy Dolby Vision rpu metadata. (--dolby-vision-rpu copy)
- If the H.264/HEVC header cannot be obtained, it is obtained from the first packet. ( #196 )
- Fixed a bug where decoding would fail if the AV1 in the mkv file had the header as is.
- Fixed an issue where the audio mux buffer was insufficient, causing the audio to be muxed to a different fragment than the video at the same time.
- Fixed an issue where invalid memory access would occur when using --vpp-transform and the frame size was not divisible by 64.

2024.07.04 (7.68)
[QSVEncC]
- Changed from oneVPL-based to libvpl-based.
- Added packages for Ubuntu 24.04.
- Updated to support API 2.11.
- Added AI Super Resolution added in API 2.11. (--vpp-resize mfx-ai-superres)
But it seems the driver is not yet supported.
- Added per-channel control to --vpp-tweak.

[QSVEnc.auo]
- Avoided the problem where the output extension was not set unintentionally and muxing was not possible depending on the status of Windows registered extensions.

2024.06.29 (7.67)
[QSVEncC]
- Fixed an issue where the top and bottom of frames were swapped when encoding from an RGB avi file.

[QSVEnc.auo]
- Updated Chinese translation, provided by Nsyw.
- When using Extended Editing, if the video and audio lengths are different, a warning will be displayed and the process will be paused, allowing you to decide whether to continue or not.

2024.06.08 (7.66)
- Added a new noise reduction filter (--vpp-fft3d).

2024.05.24 (7.65)
- Added a new deinterlace filter (--vpp-decomb).

2024.05.12 (7.64)
- Updated to ffmpeg 7.0 (Windows version)
- ffmpeg 6.1 -> 7.0
- libpng 1.4.0 -> 1.4.3
- expat 2.5.0 -> 2.6.2
-opus 1.4 -> 1.5.2
- libxml2 2.12.0 -> 2.12.6
- dav1d 1.3.0 -> 1.4.1
- libvpl 2.11.0 (new!)
- nv-codec-headers 12.2.72.0 (new!)
- You can now specify the decoder to use with avsw.
- When --audio-bitrate is not specified, leave it to the codec to decide the default bitrate instead of setting it.
- Don't process audio/subtitle/data tracks that aren't specified with --audio-bitrate or --audio-copy.
- Fixed an issue where side_data was not handled correctly in QSVEnc 7.62 and later, causing --master-display copy/--max-cll copy to not be performed correctly.

2024.04.28 (7.63)
- Added a new noise reduction filter (--vpp-nlmeans).
- Extend --audio-resampler to allow parameters to be set via a string.
- Added an option to --output-res to ignore the SAR ratio when calculating.
- Slightly speed up vpp-smooth/vpp-denoise-dct using sync on subgroup.

2024.03.17 (7.62)
- For HEVC, prefer FF when available.
- Fixed a bug in the AVX2 version of RGB3 to RGB4 conversion.
- Fixed a bug where the --vpp-smooth strength did not match when outputting 8-bit when outputting 10-bit.
- Changed the default value of --avsync from the unrealistic cfr to auto.
- Added option to specify audio quality. ( --audio-quality )
- Improved the message that appears when a device cannot be opened due to insufficient permissions in a Linux environment.
- Fixed an issue where outputting to a non-existent drive would cause an abnormal termination.

2024.02.20 (7.61)
- When encoding with Arc GPU with Resizable BAR disabled, the GPU clock drops.
Avoid the problem of encoding slowing down significantly. (--avoid-idle-clock)

2024.02.16 (7.60)
- Fixed an issue where getWorkSurf could fail in 7.59.
- Fixed an issue where --disable-d3d did not work properly on 7.56 and later.
- Other code cleanups.

2024.02.11 (7.59)
- Added noise reduction filter (--vpp-denoise-dct)
- Speed ​​up processing other than avhw.
- Added time measurement function for each task. (--task-perf-monitor)

2024.01.04 (7.58)
- Fixed a bug where spline-related resize might not work properly.

2023.12.08 (7.57)
- Updated ffmpeg to 6.1. (Windows version)
- ffmpeg 5.1 -> 6.1
- libpng 1.3.9 -> 1.4.0
-opus 1.3.1 -> 1.4
- libsndfile 1.2.0 -> 1.2.2
- libxml2 2.10.3 -> 2.12.0
- dav1d 1.0.0 -> 1.3.0
- libaribcaption 1.1.1 (new!)

- --caption2ass is now removed.
The same processing can be achieved with --sub-codec ass#sub_type=ass,ass_single_rect=true.

2023.12.03 (7.56)
- Improved progress display when using --seek.
- When --disable-opencl is specified, OpenCL information is used to not display GPU information, which increases speed.
- By specifying this when checking for DX11, you can suppress the DX9 check and speed up session initialization.
- Fixed an issue where an abnormal termination would occur if the target file was empty when --option-files was specified.

2023.11.28 (7.55)
- AVX2 instructions were mixed into the initialization code, which prevented it from working in environments that do not support AVX2.
- Improved session initialization speed (Windows).
- Changed --audio-delay to be passed as a decimal point.
- Improved the process of copying aac when outputting ts.

2023.11.19 (7.54)
- Fixed a bug where the --dynamic-rc option added in 7.53 did not work properly.

2023.11.18 (7.53)
- Added option to dynamically change rate control mode (--dynamic-rc).
- --vpp-afs now enables rff considerations by default.
- Fixed an issue where when converting to bob using --vpp-yadif/--vpp-nnedi, if an rff video was processed without --vpp-rff, calculations around the timestamp would go wrong and the program would terminate with an error.
- When using mfx deinterlacing, avoid GPU busy issue when field order is changed midway.
- If deinterlacing is specified but interlace is not set, it will automatically behave like --interlace auto.
- Suppress large amounts of messages that slow down encoding.
- Enhanced logging output for --check-features/--check-hw.

2023.11.04 (7.52)
[QSVEncC]
- Added a filter to expand rff. (--vpp-rff)
- Reverted --b-pyramid from being disabled by default in QSVEnc 7.50.

2023.10.28 (7.51)
[QSVEncC]
- ICQ mode is now the default.
- --fallback-rc is now enabled by default.

[QSVEnc.auo]
- Fixed an issue where AV1 could not be output due to muxer limitations when using an external audio encoder.
The use of external muxer has been discontinued and changed to using internal muxer.
- AV1 CQP limit increased to 255.
However, we generally recommend using fixed quality (ICQ).

2023.10.18 (7.50)
[QSVEncC]
- Fixed an issue where an abnormal termination would occur with --thread-audio > 1 when audio filter switching occurs during encoding.
- Added "quiet" to --log-level.
- Improved handling of --check-features.
- Supports the new AVChannelLayout API (Windows version).

2023.10.01 (7.49)
[QSVEncC]
- Supports device selection with --device on Linux.
- Improved error messages for --vpp-afs, --vpp-nnedi, --vpp-yadif, --vpp-pad.
- Fixed --vpp-decimate not properly calculating the last frame timestamp, causing a crash.

2023.08.08 (7.48)
[QSVEncC]
- Disable --open-gop when --hyper-mode is on.
- Improved operation when --hyper-mode on is specified in a multi-GPU environment.
- If --video-tag is not specified, "hvc1" will be used as the default to improve playback compatibility for HEVC.
(libavformat defaults to "hev1")

2023.07.24 (7.47)
[QSVEncC]
-Fixed --check-features not working properly on 4575 driver.
- --audio-stream is now supported when reading avs.
- Added option to enable perceptual pre encode filter. (--vpp-perc-pre-enc)

2023.06.24 (7.46)
[QSVEncC]
- Fixed auto_xx etc. of --vpp-denoise to work.
- Fixed an issue where hdr2sdr in --vpp-colorspace did not work properly when the input colorprim was not specified.
- Added --max-framesize-i, --max-framesize-p.

[QSVEnc.auo]
- Fixed an issue where audio would cut out intermittently when processing faw.

2023.06.20 (7.45)
[QSVEncC]
Added --tile-row, --tile-col, --max-framesize.

[QSVEnc.auo]
- Fixed a problem with QSVEnc 7.43's faw processing that could cause it to terminate abnormally.

2023.06.xx (7.44)
- Avoid error "AAC bitstream not in ADTS format and extradata missing" when using --audio-copy with ts files.

2023.06.04 (7.43)
[QSVEncC]
- Improved output when specifying colormatrix etc. in HEVC encoding.
- When selecting a GPU from multiple devices, the priority of devices that fail to initialize OpenCL will be reduced.

[QSVEnc.auo]
- Now you can process faw with built-in functions even if faw2aac.auo/fawcl.exe is not present.

2023.06.02 (7.42)
- Avoid errors in environments where OpenCL is not properly installed.
- Expanded log output adjustment function.

2023.05.18 (7.41)
- Fixed an issue where encoding was not possible with the 4369 driver.

2023.05.14 (7.40)
[QSVEncC]
- Supports dshow camera input, etc.
- Changed the command line separator for --audio-source and --sub-source.
- --vpp-colorspace adjustments.

2023.05.10 (7.39)
[QSVEncC]
- Changed to continue processing even if OpenCL initialization fails.
- Added definition for RGY_PRIM_ST432_1.
- Parallel audio processing per track.

2023.05.02 (7.38)
[QSVEncC]
- Added the ability to specify file format etc. with --audio-source/--sub-source.
- Added support for libavdevice.
- Fixed an issue where audio with a timestamp not starting at 0 could not be synchronized correctly with the video when loaded with --audio-source.
- Changes to dependent packages for Linux version (Ubuntu 22.04).
libmfx-gen1.2 -> libmfxgen1

2023.04.12 (7.37)
[QSVEncC]
- Supports oneVPL API 2.9.
- Added option to tune encoding quality (--tune).

2023.03.29 (7.36)
[QSVEncC]
- Avoid the problem of AV1 in mkv not being decoded by hw and ending with a Failed to DecodeHeader error.
- Avoid strange colors in y4m files that do not contain color space information.
- Avoid the error "pgs_frame_merge not found" on Linux.
- Improved the accuracy of bilinear interpolation when shrinking and added bicubic.
- OpenCL build log tweaks.
- Prevented negative values ​​from being entered into the audio/subtitle timestamp.
- Removed the 4GPU limit when selecting a device.
- Added --gpu-copy option.

[QSVEnc.auo]
- Added a warning when the length of the output video is too short.
- When checking if QSV is available, use the 64-bit version of QSVEncC on 64-bit OS.

2023.03.07 (7.35)
[QSVEncC]
- Added an option to change the color tone according to a specified curve. (--vpp-curves)
- Changed --vpp-overlay to the last filter.
- If OpenCL initialization fails, continue as much as possible without OpenCL.
- Improved the way GPU usage is calculated.
- Avoid an error termination when --ctu is specified on a generation of GPU that does not support it.
- Improved handling of --open-gop to make it easier to seek through files.

[QSVEnc.auo]
- Supports output when a file is dragged and dropped from the Object Explorer.

2023.02.13 (7.34)
[QSVEncC]
- Added option to set frame time from timecode file (--tcfile-in).
- Added option to specify time precision (--timebase).
- Display a warning if --audio-profile could not be changed.
- Fixed an issue where --check-hw would terminate abnormally if information could not be obtained properly.

2023.02.09 (7.33)
[QSVEncC]
- Fixed an issue where copying PGS subtitles was not working properly from 7.30.

2023.02.07 (7.32)
[QSVEncC]
- From 7.30, when using --vpp-resize spline16, spline36, spline64, unintended lines and noise were introduced. Fixed this issue.

2023.02.06 (7.31)
[QSVEncC]
- Fixed a bug where the RFF section would sometimes become 12 fps due to a false positive when using the --vpp-afs rff option.

2023.02.05 (7.30)
[QSVEncC]
- Updated ffmpeg library (Windows version)
ffmpeg 5.0 -> 5.1
libpng 1.3.8 -> 1.3.9
expat 2.4.4 -> 2.5.0
libsndfile 1.0.31 -> 1.2.0
libxml2 2.9.12 -> 2.10.3
libbluray 1.3.0 -> 1.3.4
dav1d 0.9.2 -> 1.0.0
- Avoid the problem that PGS subtitles could not be muxed properly when loaded with --sub-source.
- Added rff option to --vpp-afs.
- --check-hw now also displays the names of codecs that are supported for encoding.
- The log output of --check-hw and --check-features can now be controlled with --log-level.

2023.01.30 (7.29)
[QSVEncC]
- Added image burn filter (--vpp-overlay).
- Tweaked mux behavior for lowlatency.
- Added option to attach a file to a video file. (--attachement-source)
- Fixed the issue where bitrate information, etc. was not output when using --perf-monitor.
- Fixed an issue where the audio encoding thread (--thread-audio 1) was no longer working.

2023.01.22 (7.28)
[QSVEncC]
- Fixed typo in --scenario-info.

2023.01.22 (7.27)
[QSVEncC]
- Added option to --vpp-decimate to drop multiple frames.
- Improved AV1 master-display handling.
-Fixed an issue where AV1 encoding was emitting incorrect data when either maxcll or mastering display was missing.
- Fixed an issue where specifying --audio-copy by language did not work properly.

2023.01.21 (7.26)
[QSVEncC]
- Added an option to pass scenario information to optimize image quality. (--scenario-info)
- Fixed an issue where a strange value would be set when specifying a value of 65535 or more for vbvbufsize.
- Changed the behavior so that the specified chromaloc etc. takes priority even when using dolby-vision-profile.
- Do not change IdrInterval when OpenGOP.
- Remove the minimum lookahead depth restriction.

2023.01.11 (7.25)
[QSVEncC]
- Fixed an issue where initialization failed when using remote desktop.
- Enhanced debug logging.
- Lookahead depth may be effective even with VBR/CBR, so display it in the log.
- Changed trellis to display only when encoding in H.264.

[QSVEnc.auo]
- Fixed garbled error messages.

2022.11.12 (7.24)
[QSVEncC]
- Added an option (--frames) to specify the number of frames to encode.
- Added --fps to force input frame rate.
- Added swapuv option to --vpp-tweak.
- Added an option to --vpp-subburn to burn only subtitles with the forced flag.
- Output will be raw even if the output file name is .raw.

2022.11.01 (7.23)
[QSVEncC]
- Fixed a bug in 7.22 where the order of rate control modes in --check-features was swapped.

[QSVEnc.auo]
- Fixed inability to select the appropriate rate control mode in 7.22.

2022.10.30 (7.22)
[QSVEncC]
- Fixed an issue where --vpp-afs did not work properly when combined with AV1 encoding.
- Automatically select the appropriate GPU based on the specified options.

[QSVEnc.auo]
- Added a setting field to specify AV1 bit depth.

2022.09.21 (7.21)
[QSVEncC]
- Added support for AV1 encoding for --master-display/--maxcll/--dhdr10plus.
- Added new options.
- --repeat-headers
- --intra-refresh-cycle
---hevc-gpb
- Added --atcsei info to logs.
- AV1 CQP/ICQ cap increased to 255.
- Changed AV1 GopRefDist default to 8.
-Fixed OpenCL compilation error with --vpp-deband.
- Fixed an error when combining OpenCL filters with non-avhw import on Arc GPU.

[QSVEnc.auo]
- Fixed some suspicious behavior on the settings screen.

2022.09.21 (7.20)
[QSVEncC]
- Added --gop-ref-dist parameter for AV1. Previously, it was fixed at "1", but by increasing this to "4" or other values, the compression ratio can be significantly improved.
- Changed the default value of AV1's --gop-ref-dist to 4 (previously it was fixed at "1").

2022.09.19 (7.19)
[QSVEncC]
- Change the default for --hyper-mode to off.
This is because errors may occur in non-supported environments and it may be difficult to avoid them at this time, especially with HEVC encoding.

[QSVEnc.auo]
- Fixed typos.
- Updated Chinese translation.

2022.09.18 (7.18)
[QSVEncC]
- Fixed an issue where QSVEnc 7.17 would abnormally terminate when --profile was specified.

[QSVEnc.auo]
- Added Chinese support provided by Nsyw, who supports Chinese in Aviutl.
Thank you for translating!

2022.09.17 (7.17)
[QSVEncC]
- Fixed an issue where VP9 encoding was no longer possible in some environments after 7.08.
- Added a check to see if aud/pic-struct is available.

[QSVEnc.auo]
- Added tooltips to the settings screen.
- Supports English display.

2022.09.07 (7.16)
[QSVEncC]
- Fixed an issue where HEVC encoding was not possible in QSVEnc 7.15 on pre-Icelake environments.

2022.09.05 (7.15)
[QSVEncC]
- Adjusted to enable HyperMode in HEVC.
- Fixed an issue where an unintended device would be grabbed when automatically selecting a device.
- Improved stability of OpenCL filters.

2022.09.01 (7.14)
[QSVEncC]
- Improved stability of --ssim, --psnr.
- Improved information collection such as GPU usage.

[QSVEnc.auo]
- AuoLink function has been discontinued.

2022.08.25 (7.13)
[QSVEncC]
- --audio-stream processing now also handles cases where the audio channel changes midway.

2022.08.24 (7.12)
[QSVEncC]
- Fixed an issue where --vpp-yadif would sometimes cause an error to occur on the last frame.

2022.08.23 (7.11)
[QSVEncC]
- Added more deinterlacing filters (--vpp-yadif)
- Fixed a bug where GPUs other than the primary one could not be detected depending on the environment.

[QSVEnc.auo]
- Fixed a bug where FAW 1/2 modes could not be processed correctly when using AVX2.

2022.08.17 (7.10)
[QSVEncC]
- Fixed an issue where the system would not function properly when libmfx-gen1.2 was installed in a Linux environment.

2022.08.14 (7.09)
[QSVEncC]
- Fixed an issue where an error would occur when using --disable-opencl.
- Fixed an issue where the GPU name could not be obtained with --check-device on Linux.
- Fixed an issue where processing would be interrupted unintentionally when reading from standard input on Linux.
In addition, due to this change, we have changed the system so that terminating a program with 'q' or 'Q' from the console on Linux is no longer supported (use Ctrl+C instead).

[QSVEnc.auo]
- Added --gapless-mode 2 to qaac/fdkaac command line.

2022.08.08 (7.08)
[QSVEncC]
-Fixed HyperMode detection.
- Adjust parameters so that HyperMode can be used as much as possible when --hyper-mode on is specified.
- Added check for B frames.
- Extend --check-features to check separately when --fixed-func is enabled (FF) and disabled (PG).
- --check-features and --check-environment can now obtain information on the device specified by --device.
- Lifted the --async-depth restriction.

[QSVEnc.auo]
- Added HyperMode setting field.
- Added device selection field.

2022.08.01 (7.07)
- Fixed an issue where the app would crash when selecting a device other than the default device.
- Added option to display a list of available devices (--check-device).
- Fixed the issue where AV1 level and profile were swapped.
- Disable B-frame specification in AV1.
When the B frame setting was specified, a file that could not be played properly was created.

2022.07.03 (7.06)
- Added option to enable Adaptive CQM added in API 2.2. (--adapt-cqm)
- Added option to enable AdaptiveRef added in API 2.4. (--adapt-ref)
- Added option to enable Adaptive LTR added in API 2.4. (--adapt-ltr)
- Fixed a problem where AV1 hw decoding was not possible from QSVEnc 6.10.

2022.07.02 (7.05)
[QSVEncC]
- Fixed an issue in QSVEnc 7.02 where color abnormalities were observed on the left edge of the screen when using --vpp-afs and --fixed-func.
- Removed --adapt-ltr, which was deprecated in oneVPL.
- Reduced use of other deprecated functions.

2022.06.28 (7.04)
[QSVEncC]
- Updated oneVPL to support API 2.6.
- HyperMode (DeepLink) is now available. (--hyper-mode)
- Reduced use of deprecated AVPacket functions.
- Fixed an issue where black horizontal lines would appear when using crop when loading yuv422.

[QSVEnc.auo]
- Change the default audio bitrate.
- Change preset audio bitrate.
- If the executable cannot be found in exe_files, it will search in plugins\exe_files.

2022.06.17 (7.03)
[QSVEncC]
- Fixed a bug where --vpp-colorspace lut3d did not work properly.
- Prevent garbled characters when using --check-features-html.

[QSVEnc.auo]
- Fixed garbled characters in the encoding completion log.

2022.06.14 (7.02)
[QSVEncC]
- Added support for 3D LUT in vpp-colorspace. (--vpp-colorspace lut3d)
- Added 3D noise reduction filter (--vpp-convolution3d).
- Added error check for ldr_nits and source_peak when using hdr2sdr in vpp-colorspace.
- Added option to resize to a specified resolution while maintaining aspect ratio.
-Fixed a problem when vpp-smooth does not support fp16 or subgroup.
- Fixed an issue where avhw could not be decoded correctly when using hevc 10bit.
- Fixed stripes appearing on duplicated frames when using vpp-afs with YUV420.
- Added packages for Ubuntu 22.04.

[QSVEnc.auo]
- Adjusted the drawing of the settings screen when using the Black Window Plugin.
- Use -aac_coder twoloop in ffmpeg (AAC).
- Changed to display an error message when running the easy installer directly.
- When you're running out of disk space, it now shows you which drive is missing.
- When using an external muxer, processing is now done using the remuxer as much as possible.
- Don't use ScrollToCaret().
- Fixed an issue where the temporary audio output destination was no longer reflected.

2022.04.16 (7.01)
[QSVEncC]
- Fixed an issue where encoding would not start and the program would freeze depending on the environment.
- Fixed audio delay when the audio start time is not 0 and the video and audio timebases are different.

2022.04.07 (7.00)
[QSVEncC]
- Migrated to Visual Studio 2022.
- AV1 encoding support (-c av1)
- Added option to select device to use (--device).

[QSVEnc.auo]
- Migrated to .NET Framework 4.8.
- Changed the package folder structure.
- Installation using the simple installer has been discontinued.
- If no path is specified, search for the executable in exe_files and use it.
- Output the path of the executable file used in the log.
- Save paths as relative paths by default.
- Improved behavior when using Advanced Editing when video and audio lengths are different.
After loading audio in the extended editing, if you load audio with a different sampling rate into Aviutl itself,
The number of audio samples remains the same, but the sampling rate changes, which can result in a change in the duration of the audio.
If you suspect that there is a discrepancy between the length of the video and audio when using extended editing, try the following:
Fix the number of samples in the audio.
- Added an error message when the number of frames to encode is 0.
- Fixed an exception that occurred when failing to save the log.
- If log saving fails, the reason will be displayed.
- Show more details about some causes of mux errors.
Added an error message when selecting an unsupported audio encoder for mp4 output.
- Error message
"x264 terminated unexpectedly. Invalid parameters (options) may have been passed to x264."
Show some of the causes in detail.
1. Added an error message when encoding fails due to insufficient disk space.
2. Added an error message when trying to output a file name or folder name that contains environment-dependent characters.
3. Added an error message when attempting to output to a folder to which you do not have access, such as a folder protected by Windows.

2022.03.06 (6.10)
[QSVEncC]
-Updated ffmpeg-related dlls. (Windows version)
ffmpeg 4.x -> 5.0
expat 2.2.5 -> 2.4.4
fribidi 1.0.1 -> 1.0.11
libogg 1.3.4 -> 1.3.5
libvorbis 1.3.6 -> 1.3.7
libsndfile 1.0.28 -> 1.0.31
libxml2 2.9.10 -> 2.9.12
libbluray 1.1.2 -> 1.3.0
dav1d 0.6.0 -> 0.9.2

[QSVEnc.auo]
- When outputting, do not overwrite files that Aviutl has open.
- Fixed H.264 features not being recognized on the settings screen.

2022.02.26 (6.09)
[QSVEncC]
-Fixed the issue where b-pyramid could not be used in 6.08.

[QSVEnc.auo]
- When outputting, do not overwrite files that Aviutl has open.
- If no settings have been made, the settings output last time will be loaded.
- Chapters can now be handled correctly even when outputting using multiple Aviutl processes.
- Fixed an issue where log output could not be output correctly when changing the log output mode.

2022.02.08 (6.08v3)
2022.02.08 (6.08v2)
- Avoid problems in environments where SetThreadInformation cannot be used.

2022.02.06 (6.08)
- Supports VP9 encoding. (-c vp9)
- Added an option to read and reflect Dolby Vision rpu. (--dolby-vision-rpu)
- Added option to specify Dolby Vision profile. (--dolby-vision-profile)

2022.01.27 (6.07)
- Fixed the problem that the bit rate drops drastically when encoding for a long time when using a bit rate specification mode such as VBR mode.

2021.12.1 (6.06)
- Added options to specify thread priority and power throttling mode. (--thread-priority, --thread-throttling)
- Addressed an issue in environments where d3d11 memory cannot be used.
- Fixed a crash that occurred in some environments when displaying the QSVEnc.auo settings screen if QSVEncC64.featureCache.txt exists.

2021.11.01 (6.05)
- Fixed a bug where using a negative resolution when specifying --dar would result in the resolution being calculated as sar and resulting in an unintended resolution.
- Supports --vpp-denoise mode specification in API v2.05.
It seems that it cannot be used with the 9955 driver either.

2021.10.14 (6.04)
- Added Windows 11 to the expected operating environment.
- Added detection for Windows 11.
- Added option to specify thread affinity (--thread-affinity).
- Added an option to display the time on each line of the log (for debugging). (--log-opt addtime)
- Added option to copy dynamic hdr10plus metadata. (--dhdr10-info)
-Speed ​​up bitstream header search using AVX2/AVX512.
- Fixed a green line appearing in the top left corner of the screen when converting 12-bit depth to 10-bit depth.

2021.09.25 (6.03)
- Slightly faster startup speed.
-Fixed the issue where --caption2ass could not be used.
- Added option to display OpenCL information. (--check-clinfo)
- When quality=0 is set with --vpp-smooth, prec=fp16 cannot be used, so it will automatically be switched to prec=fp32.
- Added an option to display the time on each line of the log. (--log-opt addtime)

2021.09.19 (6.02)
・Updated VPL to 2021.6.
- Added --check-impl, which uses the VPL implementation to enumerate available modes.
- Optimized --vpp-resize lanczosx. 50% faster on 11700K.
- Optimized --vpp-smooth. 25% faster on 11700K.
- Optimized --vpp-knn. About 2x faster on 11700K.
- Added an option for measuring the performance of OpenCL filters. (--vpp-perf-monitor)
- Added --audio-bsf, which applies a bitstream filter to audio.

2021.09.06 (6.01)
- Fixed the issue where --d3d9 and --disable-d3d did not work in 6.00.
- Fixed an issue where a green line would appear on the last line depending on the resolution when using --vpp-colorspace.
This was not fully fixed in 6.00.

2021.09.05 (6.00)
- Transitioned from MediaSDK to oneAPI Video Processing Library (VPL) and now supports API 2.04.
API 2.xx should be available for Rocketlake/Tigerlake (Windows version).
- Fixed the issue where the horizontal resolution to be specified when changing from yuv444 to nv12 was incorrect. This fixed the problem of green lines appearing on the last line.
(It was not fully fixed in 5.06)
-Fixed the issue where audio-delay was not working.

2021.08.12 (5.07)
-Updated font types available in vpp-subburn.
-Fixed the issue where audio-delay was not working.
-Fixed an issue where a green line would appear on the last line when using --vpp-colorspace.

2021.07.26 (5.06)
- Added the ability to calculate ssim/psnr. (--ssim/--psnr)
- Fixed the issue where the horizontal resolution to be specified when changing from yuv444 to nv12 was incorrect. This fixes the problem of green lines appearing on the last line.
-Fixed color shift from yuv444 to p010 avx2 version.
-Fixed issue where rgb reading and OpenCL filters could not be combined.
-Fixed an issue where --disable-va did not work on Linux.
-Fixed an issue where --vpp-colorspace did not work on Linux.

2021.06.12 (5.05)
- Improved the efficiency of processing when loading non-avhw files and OpenCL filters are connected.
-Exits with an error if the input file and output file are the same.
-Fixed an issue where an abnormal termination could occur when using --vpp-decimate.
- Fixed a bug where the specified interlace settings were sometimes not reflected when loading y4m.
- Avoid error termination that occurs in some Avisynth environments.

2021.05.29 (5.04)
- Changed to always prioritize d3d11 when possible.
Similar changes were made to the MediaSDK sample, so we made the necessary changes.
https://github.com/Intel-Media-SDK/MediaSDK/commit/c4fbaedd8a827ec36ee312e978e993d3f938201c
-Fixed the issue where the message "Failed to find d3d9 device" would appear in 5.02.
-Fixed an issue in 5.01 where the frame order would be swapped when loading a file other than avhw and passing it directly to the OpenCL filter.
-Fixed the issue where color components were shifted when cropping with OpenCL from 5.01.
- Avoided the issue where vpp-mpdecimate/decimate would freeze in environments prior to Broadwell.

2021.05.23 (5.03)
- Fixed an issue where an abnormal termination would occur when trying to output raw or log data to the current directory.
-Fixed decoding error on Sandybridge on Win8.1.
-Fixed the issue where color components were shifted when using crop and OpenCL filter together.

2021.05.16 (5.02)
[QSVEncC]
- In 5.01, d3d11 was used preferentially even in situations where it was not necessary. This will be reverted to using d3d9 instead.
- In 5.01, when using --avsync forcecfr, it was not possible to insert more than 16 consecutive frames. This has been changed to allow insertion of 18,000 frames (actually unlimited).
An abnormal termination occurred when attempting to insert 16 or more consecutive frames.
- In 5.01, when using --vpp-mpdecimate or --vpp-decimate, resizing was incorrectly performed. Fixed this.
-Fixed the issue where incorrect resolution operations/changes were made when using --vpp-pad and --crop in 5.01.
- In 5.01, --vpp-afs, --vpp-mpdecimate, --vpp-decimate caused an abnormal termination (Failed to acquire OpenCL interop). Fixed this.
- Fixed an issue where --vpp-deinterlace bob would cause an abnormal termination (Application provided invalid, non monotonically increasing dts to muxer) in 5.01.
- Avoid the problem in 5.01 where decoding is not performed properly (a green image appears) in a Win7 Sandybridge environment.
-Fixed an error that occurred when using the OpenCL filter with -c raw in 5.01.
-Improved log messages for debugging.

[QSVEnc.auo]
- Fixed a bug where resizing had no effect even when specified from the settings screen.

2021.05.08 (5.01)
-Fixed Linux build that did not work from 5.00 beta1.
- Added an option to set the upper limit of the file analysis size when reading a file with avsw/avhw. (--input-probesize)
- Expanded --input-analyze to allow specifying decimal points.
- Added an option to output information about the packets read. ( --log-packets )
・For data streams only, packets without a timestamp will be forwarded as is.
- Added the ability to load a file containing options. ( --option-file )
- Added an error message when video information cannot be obtained.
- You can now select the track to copy by codec name.
- Fixed a bug where the process for converting subtitles was not enabled when it was necessary.
-Fixed the issue where raw output was not performed even when -c raw was specified in 5.00beta.
- Fixed an issue where an abnormal termination would occur when using --vpp-subburn with subtitles of size 0.
- When using OpenCL filters or AV1 decoding, prioritize d3d11 mode.

- --videoformat cannot be easily obtained from the input, so "auto" has been removed.
- Improved behavior when multiple --audio-source/--sub-source are specified.
-Fixed an issue where subtitle metadata was being output twice.
- Copy metadata from input file even when --sub-metadata or --audio-metadata is specified.

- Added the following OpenCL vpp filter.
- --vpp-afs
---vpp-colorspace
--vpp-deband
- --vpp-decimate
--vpp-edgelevel
- --vpp-mpdecimate
--vpp-nnedi
--vpp-pad
--vpp-pmd
--vpp-smooth
--vpp-tweak
- --vpp-unsharp
- --vpp-warpsharp

- Added AVX2/SSE2 version of yuv444 → nv12/p010/ayuv/y410 conversion.

Known Issues
・The vpp filter may not work with YUV422/YUV444.

2021.04.07 (5.00 beta2)
- Now works in environments without OpenCL, such as SandyBridge.
- Fixed an issue where the program would no longer work on pre-Broadwell environments due to the error "Failed to find d3d9 device."
-Fixed OpenCL compilation error in pre-Broadwell environments.
- The unstable CPU version of --vpp-delogo has been abolished and the OpenCL version of --vpp-delogo has been implemented.
- Supports AV1 HW decoding on Rocketlake. However, --d3d11 must also be specified.
- Fixed the bug that YUV444 encoding cannot be performed when --fixed-func is specified.
- Changed the default value of --async-depth to 3. This was using a lot of memory unnecessarily.
-Fixed a bug that could cause an abnormal termination when using the OpenCL filter when using -c raw.

Known Issues
・Cannot be built in a Linux environment.
・The vpp filter may not work with YUV422/YUV444.

2021.03.30 (5.00 beta1)
New Features
-MediaSDK updated to support API 1.35.
- Expanded support for IceLake.
-Initial implementation of Rocketlake support.
- The internal implementation has been revamped, making it possible to incorporate OpenCL filters.
・Added --vpp-knn.
・Added --vpp-transpose.
- Supports HEVC YUV422/YUV444 decoding. (Icelake/Rocketlake)
- Supports HEVC YUV444 encoding. (--output-csp, Icelake/Rocketlake)
- Supports VP9 YUV444 decoding. (Icelake/Rocketlake)
- Added information about color spaces supported by HW decoding to --check-features.
- Added option to specify resizer algorithm. (--vpp-resize/--vpp-resize-mode)
- Added H.264 Level 6, 6.1, 6.2.

Known Issues
・Cannot be built in a Linux environment.
・--vpp-delogo does not work.
- Operation may be slow when using a reader other than avhw.
・The vpp filter may not work with YUV422/YUV444.

abolition
- vpp-half-turn has been abolished. --vpp-transform etc. can be used instead.

2021.02.17 (4.13)
-Added Unicode support for Avisynth.
- Added long path support for Windows 10.
- Enhanced to allow specifying metadata with --audio-source / --sub-source.
- Supports selection of audio and subtitles by language.
- Changed the rounding method when lowering bit depth.
- Fixed a bug where the ms value could not be obtained correctly when loading a chapter.

2020.11.23 (4.12)
- Workaround for the issue where extbrc could not be used in VBR mode after updating to the 9025Beta driver.
- Fixed a bug where the output would become abnormal when chromaloc was set.

2020.11.22 (4.11)
- Fixed a bug where the ms value could not be obtained correctly when loading a chapter.
-Fixed Linux build when AVX/AVX2 is not available.
- Check if the rate control mode is in target with WinBRC.

2020.11.19 (4.10)
-Fixed the issue where B pyramid and Pyram QP Offset could not be used in 4.09.
- Added extbrc option.

2020.11.12 (4.09)
- Compatible with Media SDK 1.34.
-Convert from yuv422 to yuv420 on the CPU.
It seems that in environments with dGPUs, conversion from yuv422 to yuv420 on the GPU may not work properly.
- Fixed an issue where the performance monitor could not be obtained correctly.

2020.09.30 (4.08)
-Fixed inability to decode Apple prores.
- Added option to specify color space when reading raw data. (--input-csp)
Supports 8-16bit reading of yuv420/422/444.
- Change the return value of the program depending on the success or failure of --check-lib.
- When encoding HEVC, automatically use main10 when --output-depth 10 is specified.
Until now, encoding at 10-bit depth was not possible unless it was used in conjunction with --profile main10.
- Updated Linux build to make QSV encoding easier on Intel iGPUs from Broadwell onwards.
- Added information on how to build on Linux.

2020.08.06 (4.07)
- Added an option to specify Avisynth.dll to load. (--avsdll)

2020.08.02 (4.06)
[QSVEncC]
-Fixed a bug that could cause an abnormal termination in some cases.

2020.07.29 (4.05)
[QSVEncC]
・Updated Media SDK to 2020 R1.
-Updated ffmpeg related dll.
This allows muxing of PGS to ts/m2ts.
-Fixed the issue where --audio-stream stereo did not work.
- Default duration is now set when outputting mkv.
- Implemented a workaround for not being able to use --chromaloc.

2020.06.16 (4.04)
[QSVEncC]
- Some HEVC files may not be decoded correctly, but we have made it possible to decode them with the SW decoder if possible.
・You can now specify delay with --audio-source.
- Get more detailed Avisynth version when reading avs.
- vpy loading has been single-threaded since 4.02
Reverted to multi-threaded operation.

[QSVEnc.auo]
- It is now possible to specify special values ​​(negative values) for --output-res in the QSVEnc.auo settings screen.

2020.05.31 (4.03)
[QSVEncC]
- Fixed an issue where the end of audio was not encoded with some --audio-filters that involved delay.
-Fixed the issue where lowlatency could not be used.
-Fixed an issue where the program would abnormally terminate when --video-tag was specified.
- Added a set of options to control the metadata to be output.
--metadata
--video-metadata
--audio-metadata
--sub-metadata
- Added options to specify the disposition of the stream. (--audio-disposition, --sub-disposition)
- Fixed a bug where file names could not be obtained properly with --audio-source/--sub-source.
・Added the following options that were not listed in --help.
--video-tag
--keyfile
--vpp-smooth
- Added option to display option list. (--option-list)

2020.05.06 (4.02)
[QSVEncC]
- Fixed a memory access error when converting yuv444 to yv12/p010 in multi-threaded mode.
- Added a mode to minimize latency. (--lowlatency)
The encoding throughput itself will be reduced, so there may not be much use for it.

[QSVEnc.auo]
- Fixed an issue where an abnormal termination would occur when performing audio encoding "simultaneously" when using an external encoder.

2020.04.15 (4.01)
[QSVEncC]
- Avoided the issue of not being able to use vpp in IvyBridge environment from 3.33.

[QSVEnc.auo]
-Fixed the issue where the default audio encoder settings were not reflected.

2020.04.05 (4.00)
[QSVEncC]
・If the options specified for the audio decoder or encoder are incorrect,
Changed to issue a warning and continue instead of terminating abnormally on an error.
- Avoided the issue of not being able to use vpp in SandyBridge/IvyBridge environments from 3.33.
- Fixed the issue where --chapter only worked when using avsw/avhw.

[QSVEnc.auo]
- Added a mode to use the internal encoder in QSVEnc.auo.
This behavior will be the default and using an external encoder will be optional.
- Fixed an issue where the program would abnormally terminate when trying to open the settings screen in an environment without QSV.

2020.03.07 (3.33)
[QSVEncC]
- Added an option to specify input options when reading avsw/avhw. (--input-option)
- Added test code that uses the Media SDK color filter. (--vpp-colorspace)
- Improved the issue where some audio, such as trueHD, could not be muxed properly.
-Fixed the issue where the IceLake generation was not detected correctly.
- Changes have been made to accommodate the corrections to QSVEnc.auo.

[QSVEnc.auo]
- Fixed an issue where the frame would become distorted when minimizing or restoring the Aviutl window when outputting from QSVEnc.auo.

2020.02.29 (3.32)
[QSVEncC]
-Fixed a case where caption2ass did not work properly.
・Review of help.
-Fixed the issue where --cqp did not work properly in 3.31.

[QSVEnc.auo]
- Modifications aimed at making the simple installer more stable.
You can choose to download and install the executable file you need,
Change the installation method to extract and install the included executable file.
- Changed the default audio encoder to AAC using ffmpeg.
- Adjusted the order of tab transitions on the QSVEnc.auo settings screen.

2020.02.20 (3.31)
[QSVEncC]
- Improved error messages when command line is specified incorrectly.
- Reviewed mux processing to improve symptoms such as difficulty seeking.

[QSVEnc.auo]
- Bitrate limit lifted.

2020.02.02 (3.30)
[QSVEncC]
・vpp-sub has not been working reliably recently, so it has been disabled.
- Added a function to copy information such as colormatrix from the input file.
--colormtarix auto
--colorprim auto
--transfer auto
--chromaloc auto
--colorrange auto
- VUI information, mastering display, and maxcll information will be displayed in the log.
-Fixed an issue where the program would sometimes terminate with an error.
- Always show output file name in log.
- Display VUI information, mastering dsiplay, and maxcll information in the log.

[QSVEnc.auo]
- Changed the implementation for linking with QSVEncC.
Hopefully they've addressed the issue of the occasional green frame appearing(?)...

2020.01.18 (3.29)
[common]
・Changed the operating environment.
-Updated Media SDK to 2019 R1.
- Uses process GPU usage information.

[QSVEncC]
- HDR related meta information can now be copied from input files.
(--master-display copy, --max-cll copy)
-Updated ffmpeg related dll.
Enables AV1 software decoding.
libogg-1.3.3 -> 1.3.4
twolame-0.3.13 -> 0.4.0
wavpack-5.1.0 -> 5.2.0
libxml2-2.9.9 -> 2.9.10
dav1d-0.5.2 !new!

2019.12.24 (3.28)
[QSVEncC]
- Fixed memory leak in audio processing.
-Improved error messages when encoding audio.
-Fixed the issue where copying subtitles etc. no longer worked.
-Fixed a bug where audio decoding could fail when using trueHD in mkv.
- Added option to add delay to audio. ( --audio-delay )
-Fixed an issue where encoding a cut video from VC-1 containing mkv would fail.

[QSVEnc.auo]
- Updated the simple installer.
-Fixed an error in the AVX2 version of yuy2 to nv12i conversion.

2019.11.23 (3.27)
[QSVEnc.auo]
-Fixed an issue where profiles could not be saved.

2019.11.19 (3.26)
[QSVEnc.auo]
-Fixed the inability to resize.
-Fixed an issue where vpp-deinterlace bob did not work properly.

[QSVEncC]
- You can now specify negative values ​​for either the width or height in output-res.
Resize to fit one side while maintaining the aspect ratio, but make the number divisible by the negative value.
--output-res -4x1080

2019.11.15 (3.25)
[QSVEnc.auo]
- Speed ​​up inter-process communication between QSVEnc.auo and QSVEncC.
- Changed QSVEnc.auo output to mp4/mkv output, reducing the number of mux steps, especially when using auto field shift.
Also, by using the QSVEncC muxer, you can list QSVEncC as the library that created the container.

[QSVEncC]
- Exclude VC-1 from hardware decoding.
Since 3.04, it has become impossible to code with VC-1, but it has not been possible to recover it.
-Fixed an issue that could cause a deadlock under high load.
-Fixed a bug where the CPU operating frequency could not be obtained properly.
- Added option to load subtitle file. (--sub-source)
- Expanded the method for specifying --audio-source.
-Allows audio reading from avs.
-Fixed a bug where audio encoding did not work properly.
- Prevents master-display and max-cll information from being garbled when muxing.

2019.06.26 (3.24)
・Revised the behavior of --sub-copy asdata.
- From 3.21, if you use -c raw or --disable-d3d, you will get the error "Failed to SynchronizeFirstTask: unknown error".
Fixed an issue where the program would terminate with an error.

2019.06.26 (3.23)
[QSVEncC]
-Add an option to copy data streams. (--data-copy)

2019.06.23 (3.22)
[QSVEncC]
- Added a mode to copy subtitles as data with --sub-copy.
--sub-copy asdata
- Added the ability to specify decoder options to --audio-codec.
--audio-codec aac#dual_mono_mode=main
-Fixed the issue where RGB reading was not possible.

2019.06.15 (3.21)
[QSVEncC]
-Fixed the issue where deinterlace preservation was enabled with --vpp-deinterlace none.
・--chapter supports matroska format.
- Added/updated ffmpeg and related dlls.
- [Added] libxml2 2.9.9
- [Added] libbluray 1.1.2
- [Added] aribb24 rev85
- [Updated] libpng 1.6.34 -> 1.6.37
- [Updated] libvorbis 1.3.5 -> 1.3.6
- [Updated] opus 1.2.1 -> 1.3.1
- [Updated] soxr 0.1.2 -> 0.1.3

2019.04.26 (3.20)
[QSVEnc.auo]
- When using bob/it etc. for deinterlacing, the frame rate is not reflected properly.
Fixed an issue where the sound was out of sync.

[QSVEncC]
-Fixed the issue where --mbbrc did not work in 3.19.

2019.04.19 (3.19)
[common]
・Transitioned to VC++2019.

[QSVEnc.auo]
- Updated the simple installer (compatible with VC++2019).

[QSVEncC]
- Fixed an issue where decoding would stop for some H.264 streams.
We have received reports that this issue occurs on SKY PerfecTV Premium and other services.
-Fixed an issue where some TrueHD audio was not copied correctly.
- Added option to enable Adaptive LTR (--adapt-ltr).
Only works in some modes such as CBR and VBR.

2019.03.24 (3.18)
[QSVEnc.auo/QSVEncC common]
- When encoding H.264 with B frames in d3d11 mode in a Haswell environment, the image may be distorted.
In that case, disable B frames.

[QSVEncC]
- Added option to specify video codec tag. (--video-tag)
- Improved handling of timestamps when encoding audio, reducing audio lag during VFR.

2018.12.17 (3.17)
[QSVEncC]
-Fixed a bug where --master-display did not work properly.

2018.12.11 (3.16)
[QSVEnc.auo]
- Fixed the issue where the frame acquisition time from Aviutl was not calculated correctly when encoding was interrupted.

2018.12.10 (3.15)
[QSVEnc.auo]
- Fixed a bug that could cause a crash due to an out-of-range access exception if width was not divisible by 32 when using automatic field shift.

2018.12.04 (3.14)
[QSVEncC]
-Fixed the issue where benchmark mode did not work properly.

[QSVEnc.auo]
-Fixed the issue where the audio encoder could not be used in AuoLink mode.

2018.11.24 (3.13)
[QSVEncC]
- Fixed a bug where normal processing was not possible when using protocols such as UDP for reading.
-Fixed a bug where --audio-file did not always work properly.

2018.11.18 (3.12)
[QSVEncC]
- Implemented subtitle extraction processing using Caption.dll. (--caption2ass)
-Fixed issue where Avisynth would not work properly when using older versions.

[QSVEnc.auo]
- Updated the simple installer.
- Fixed the issue where Apple dll could no longer be downloaded.
- Automatically use system proxy settings.

2018.10.19 (3.11)
[common]
・When using vpp with SandyBridge or IvyBridge,
A fix has been implemented to address the issue where encoding could not be performed.

[QSVEnc.auo]
- Fixed the issue where wav output was not possible from the QSVEnc.auo settings screen.
The specified video encoder does not exist. [ ] error message appeared.
- Added settings for AAC encoding using ffmpeg and dual mono separation to QSVEnc.ini.
- Added batch processing after audio encoding when using faw2aac.
(However, batch processing before audio encoding when using faw2aac is not performed.)

[QSVEncC]
- Added --vpp-mctf check to --check-features.

2018.10.12 (3.10)
[common]
-Updated to Intel Media SDK 2018 R2 (API v1.27).
- Added Motion Compensate Temporal Filter (MCTF). (--vpp-mctf)

[QSVEnc.auo]
- Support for specifying relative path of temporary folder.
- Fixed a bug where muxer.exe would terminate with an error when handling multiple audio streams.

[QSVEncC]
- Added --vbv-bufsize.
- Fixed an issue where 10-bit HEVC input files in some mp4/mkv containers could not be decoded properly.
- Addressed the issue of audio being out of sync with some video files.

2018.08.01 (3.09)
[QSVEncC]
-Trim now taken into account when progressing.
- Avoid crashes in environments where OpenCL does not work properly.
Apparently it's a rare occurrence.
-Fixed issue where pipe output was not possible in version 3.00 and later.

2018.07.10 (3.08)
[QSVEncC]
- Ability to pass options to audio encoder.
Example: --audio-codec aac:aac_coder=twoloop
- You can now specify a profile when encoding audio. (--audio-profile)
- Slightly reduced memory usage at high bitrates.
- Improved the issue where the frame rate was incomplete when using variable frame rates.
- Improved synchronization when audio starts first.
- Added option to specify HEVC tier. (--tier)

2018.07.05 (3.07)
[QSVEncC]
-Fixed the issue where --audio-file was not working properly.
- Added missing options such as --colorprim and --transfer.
- Improved the effect of --input-analyze.
- Improved the issue where --vpp-deinterlace did not work when outputting raw.

2018.06.10 (3.06)
[QSVEncC]
- Improved speed of --check-features.
- Fixed a bug where --avsync forcecfr/vfr did not always work properly.
- Fixed a bug where audio encoding options would sometimes behave unintendedly.

2018.06.03 (3.05)
[QSVEnc.auo]
-Fixed an issue where plugins were not recognized in 3.04.

[QSVEncC]
-Fixed a crash when using avs/vpy/y4m readers.

2018.06.02 (3.04)
[QSVEncC]
-Updated ffmpeg and related library dlls.
・Use commands such as --audio-codec / --audio-bitrate / --audio-samplerate / --audio-filter.
If no track is specified, all audio tracks in the input file will be processed.
- Supports encoding with vfr preserved. (--avsync vfr)
- Improved compatibility when using --max-cll / --masterdisplay.
- Added an option to specify the chroma location flag.
- When outputting while muxing using interlace preserving encoding, mux on a frame-by-frame basis instead of on a field-by-field basis.

2018.05.14 (3.03)
[QSVEncC]
- Improved handling of HDR related metadata.
- Improved video and audio synchronization.
- Explicitly set the process locale to match the system locale.

2018.04.23 (3.02)
[QSVEnc]
- Fixed the behavior when double-clicking the command display field on the settings screen.
-Fixed the issue where resizing was not reflected.
-Fixed suspicious behavior when using AuoLink.

2018.04.21 (3.01)
[QSVEnc]
- Fixed a problem where the settings screen was based on 120dpi, and the layout would be distorted when displayed at 96dpi.

2018.04.21 (3.00)
[common]
- Supports Intel Media SDK 2018 R1 (API v1.26).
・VQP mode was abolished. Its role was outlived by the appearance of ICQ and other systems.
- The extbrc option has been abolished.
- Scene change detection has been removed. This did not have a noticeable effect.

[QSVEnc]
- Encoder is no longer built into the plugin, but is now piped to QSVEncC.
Reduce memory usage of the Aviutl main process.

[QSVEncC]
- Added HEVC-related options added in API 1.26 (--tskip, --sao, --ctu).
Basically, it is for upcoming CPUs (cannot be used with the Kabylake generation).
- Added options to set HDR related metadata (--max-cll, --master-display).
- Fixed a crash when trying to output to an output file name containing "%".
- Fixed a crash when specifying --log with an output file name containing "%".
-Fixed the issue where AVX2 was not used in yv12(10bit)->p010[AVX2].
- Limit the number of threads for avsw decoder to 16.
- Copy rotation metadata if it exists in the input file.

2018.01.13 (2.74)
[common]
- Weightb can now be used for HEVC encoding.
- For Kabylake and later, HEVC 10bit depth is forcibly enabled.
Kabylake and later should support HEVC 10-bit depth, but this may not be correctly determined by query.
- Improved log output.

[QSVEncC]
- When --audio-copy/--audio-codec/--sub-copy is specified, do not terminate with an error even if the input file does not have an audio/subtitle track.
-Fixed the issue where it was not possible to build on Linux.
- Supports high bit depth reading of yuv420/yuv422/yuv444 from avs.
However, it does not support the so-called high bitdepth hack.

2017.08.22 (2.73)
[QSVEncC]
-Fixed y4m reading of yuv422 with 9/12/14/16bit depth.

2017.08.16 (2.72)
[QSVEncC]
-Fixed the issue where raw output did not work properly after 2.63.
- Fixed a bug where color components were incorrect when reading high bit depth yuv422/yuv444 from y4m.
-Help fixes.

2017.07.01 (2.71)
[common]
-Fixed an issue where 2.70 would sometimes not work properly in environments with dGPU.
-Fixed slow startup issue in 2.70.
-Fixed the issue where bitrate was not displayed in la/la-hrd.

2017.06.20 (2.70)
[common]
-Fixed an issue where the app did not work properly in some environments, such as Braswell.

2017.06.18 (2.69)
[QSVEncC]
-Fixed a bug where files that could not be played were created under certain conditions when using --audio-stream.

2017.06.17 (2.68)
[common]
- Disable FadeDetect on generations prior to Kabylake.
As expected, before Kabylake, it seems to be unstable and encoding will end midway or freeze...

[QSVEnc.auo]
-Fixed the issue in 2.67 where the image would be distorted when encoding in HEVC 10bit.

2017.06.17 (2.67)
[common]
- Supports Intel Media SDK 2017 R1 (API v1.23).
・Enable fade-detect.
-Fixed the issue where weightb/weightp no longer worked.

[QSVEncC]
- Added the --repartition-check option (only when encoding with H.264).
- Supports yuv422 reading with avsw/y4m/vpy.
However, d3d9/d3d11 memory mode cannot be used and will switch to sw memory mode.
- Supports RGB reading with avsw.
- Fixed issue where dual mono separation using --audio-stream did not work properly.
- --check-features now displays the decoder's features.

2017.06.12 (2.66)
[QSVEnc.auo]
-Fixed a color shift that occurred when using resolutions that are not divisible by 16.

[QSVEncC]
- When loading from avs/avi to YUY2, color shifts occurred when the resolution was not divisible by 16. Fixed.

2017.06.11 (2.65)
[QSVEncC]
- Fixed an issue where the image would break when passing high bit depth to y4m.

2017.06.10 (2.64)
[QSVEnc.auo]
-Fixed the issue where AuoLink mode did not work properly in 2.63 due to an exception 0xc0000094.

[QSVEncC]
-Fixed an issue where an error would occur when trying to load an avi file in 2.63.

2017.06.08 (2.63)
[common]
-Enables 10-bit depth encoding even in d3d11 mode.
-Displays Windows build version in log.
- Fixed a bug where videos with heights that are not divisible by 32 could not be encoded with interlace preserved.

[QSVEnc.auo]
- Updated the simple installer.

[QSVEncC]
-Updated ffmpeg and related library dlls.
-Fixed the issue where disabling GPB for HEVC was no longer available.
- Improved stability when decoding QSV.
- Fixed an issue where the program would freeze when trying to interrupt encoding while using the vpy reader.
-YUV444 source can be read by reading avsw.
-Fixed a bug where subtitles were not copied correctly.
- Added an option to specify the number of threads to be used by Intel Media SDK. (--mfx-thread)
You can specify a value of 2 or more, but you cannot specify 0 or 1. The default is automatic (= the number of logical processors).
If you want to reduce CPU usage as much as possible, setting --mfx-thread 2 may reduce CPU usage slightly.

2017.01.08 (2.62)
[QSVEncC]
- Supports Kabylake HEVC 10bit encoding.
-Fixed a bug where GPU information could not be obtained properly.

2017.01.05 (2.61)
[QSVEncC]
-Fixed vpy reader not working properly after 2.57.

2016.12.19 (2.60)
[QSVEncC]
・HEVC encoding with mkv as input
Fixed a deadlock that occurs immediately after starting encoding.

[QSVEnc.auo]
- Updated the simple installer.

2016.12.05 (2.59)
[QSVEncC]
- Added the --chapter-no-trim option to not reflect trim when loading chapters.

[QSVEnc.auo]
- Updated the simple installer.

2016.11.06 (2.58)
[QSVEncC]
-Fixed a bug that caused avsync forcecfr to not work properly from 2.55.

[QSVEnc.auo]
- Updated the simple installer.

2016.09.29 (2.57)
[QSVEncC]
- Supports 10-bit reading with avsw/vpy reader.
- Added option to add logo. (--vpp-delogo-add)
SSE4.1 version only.
-Fixed a bug where --audio-source did not always work as expected.
- Disable the option "--fade-detect" which makes encoding unstable.
- Avoids the rare occurrence of freezing when encoding is complete.
-Improved error tolerance in audio processing.
- When deinterlacing is specified, it will be treated as --tff by default.

2016.09.11 (2.56)
[QSVEnc.auo]
-Fixed a bug where encoding was always done at 29.97fps when using AuoLink.

2016.09.03 (2.55)
[common]
- Added option to add aud/pic_struct.

[QSVEncC]
- Supports Skylake HW HEVC 10bit decoding.
-Updated ffmpeg dll.

2016.07.09 (2.54)
[QSVEnc]
- Do not obtain or display information about unused switches.

[QSVEncC]
- Added an option to specify the format of the input file when reading with avqsv/avsw reader. (--input-format)
・When encoding for a long time (6 hours 37 minutes or more) when outputting flv, the timestamp will overflow.
"Application provided invalid, non monotonically increasing dts to muxer in stream"
Fixed an issue where muxing would not work properly due to the following error.
- Fixed an issue where the program would terminate with an error message "Failed to get free surface for vpp." when using --avsync forcecfr after 2.46.
- Added a mode to use ffmpeg's sw decoder. (--avsw)

2016.06.23 (2.53)
[common]
-Fixed an issue where vpp-rotate was not available on drivers compatible with API 1.19.

[QSVEncC]
-GPU usage can now be displayed more easily.
You can display it simply by right-clicking on "Display GPU usage.bat" and running it as administrator.
-Fixed Linux build.

2016.06.18 (2.52)
-Fixed an error message.
- Updated the simple installer.

2016.06.12 (2.51)
[common]
- Compatible with API v1.19.
- Added an option to specify the quality of resizing by vpp.
--vpp-scaling
Simple, fine
- Added option to mirror image using vpp.
--vpp-mirror
h, v
"v" for vertical, "h" for horizontal.
- When encoding HEVC, added --qp-offset, which specifies the QP offset value for each pyramid reference layer.

2016.05.19 (2.50)
[common]
-Fixed an exception that would cause the program to crash when deinterlacing in 2.49.

2016.05.18 (2.49)
[common]
- Fixed a bug that caused the program to die with "Failed to initialize encoder. : invalid video parameters." even in some environments.
-Fixed the issue where --scenechange and VQP were not working properly after 2.46.

2016.05.04 (2.48)
[QSVEncC]
-Fixed the issue where mp4 files that could not be played on some players, such as Windows 10 videos, were output after 2.46.
- Added option to select video tracks when there are multiple tracks (--video-track, --video-streamid).
--video-track selects 1,2,3... from the highest resolution track, or -1,-2,-3,... from the lower resolution tracks.
The default is --video-track 1, which targets the highest resolution.
--video-streamid is specified by the stream id of the video stream.

2014.04.29 (2.47)
[QSVEncC]
- --vpp-sub now allows you to burn in image-type subtitles.
Supports AVX2 / AVX / SSE4.1 / SSE4.1 pshufb slow.
- --vpp-sub can now also run in GPU memory mode, making it significantly faster.
- Significantly faster processing when there are no subtitles to burn with --vpp-sub.
・--vpp-sub default parallel number changed from 2 to 3. It is faster.
- The default value of --async-depth has been slightly reduced. This was causing memory consumption.
- Supports --avsync forcecfr + trim.
-Added VP8/VP9 decoding supported by Skylake. (Apparently hybrid)

2016.04.24 (2.46)
[QSVEncC]
- Added the ability to burn in subtitles using libass. (--vpp-sub
or
)
If an integer is specified, the specified subtitle track from the input video file will be extracted and burned.
If a string is specified, a subtitle file separate from the input video file is read and burned in.
The optimal mode is automatically selected from four modes: AVX2 / AVX / SSE4.1 / SSE4.1 pshufb slow.
Please note the following three points.
・Only text-format subtitles are supported.
・Cannot be used in conjunction with --sub-copy.
-Since system memory mode is required, it cannot be used in conjunction with functions that require d3d11 mode (such as --vpp-rotate).
- Added an option to set libass shaping. (--vpp-sub-shaping)
Either simple (default) or complex.
- Added an option to specify the character code for subtitles. (--vpp-sub-charset)
If you do not specify, it will be automatically set. For Japanese, it may be okay to set it automatically to some extent.
When specifying, see below.
https://trac.ffmpeg.org/attachment/ticket/2431/sub_charenc_parameters.txt
- --avsync forcecfr can now be used even with H.264 input.
- Fixed a bug where --check-features output HTML in some cases when it should have output text.

2016.04.20 (2.45v2)
- Updated the simple installer.

2016.04.15 (2.45)
[QSVEncC]
- Added a workaround for errors when using --audio-copy.

2016.04.03 (2.44)
[QSVEncC]
- Improved the vpp-detail-enhance effect, which was fixed at 50 on Haswell and later.
- In the x64 version, you can now get MFX/GPU usage. This cannot be done during remote desktop.
-Fixed an issue where it was not possible to build on Linux.
- Improved error messages on command line.

2016.03.31 (2.43)
[QSVEncC]
- Improved the format of audio-related logs and added filter information.
- Audio filters can now be specified for each track.
-Fixed an issue where unnecessary logs were displayed when applying the voice filter.
- Added a batch file that checks whether QSV is available and suggests solutions if it is not available.

2016.03.27 (2.42)
[QSVEncC]
- Enables audio filtering. (--audio-filter)
dll and update it.
To change the volume, use "--audio-filter volume=0.2", for example.
The format is the same as ffmpeg's -af. It should be usable if it is a simple filter (1 stream in 1 stream out).
- avsync forcecfr + trim cannot be used together, so it will terminate with an error.
-Muxing is now possible when encoding HEVC.
- The default number of consecutive B frames when encoding HEVC has been changed from 2 to 3.
In addition, a warning will no longer be displayed when the number of consecutive B frames is three or more.
There doesn't seem to be a problem with the latest drivers.

2016.03.19 (2.41)
[QSVEncC]
-Fixed the audio delay that was not fully fixed in 2.40.
- Reduce the default output buffer size (--output-buf) to 8MB.
There have been reports that this actually reduces performance.

2016.03.17 (2.40)
[QSVEncC]
-Fixed an issue where audio was out of sync with some input files.
-Improved frame rate estimation.
-Added the input file name to the benchmark result file.
・Added information that can be obtained with --perf-monitor.

[QSVEnc]
- Updated the simple installer.

2016.03.13 (2.39)
[common]
-Fixed an issue where the encoding would freeze at the start of encoding and not proceed after 2.37.
-Fixed the error message when an audio track was not found. No impact on operation.
-Fixed a bug where the log at the end of encoding was sometimes inaccurate.

[QSVEnc]
- Updated the simple installer.

[QSVEncC]
-Fixed an issue where the encoding would freeze at the start of encoding and would not proceed after 2.38.
- Improved issue where command line parse errors were sometimes difficult to understand.
-Fixed benchmark mode not working.

2016.03.08 (2.38)
[common]
- Fixed an issue where feature information acquisition (--check-features) was inaccurate on machines with API 1.4 or lower.

[QSVEncC]
- Added an option to continue encoding without terminating with an error if there is no audio track. (--audio-ignore-notrack-error)
・When an unavailable rate control mode is specified, the program will not terminate with an error.
Added option to automatically fall back to more commonly supported rate control modes (--fallback-rc).
If you specify a bitrate, you will ultimately use vbr; if you specify a quality, you will ultimately use cqp.
- Added an error message that --avsync forcecfr cannot be used in conjunction with --vpp-deinterlace it/bob.
- Display a warning if parameter values ​​are automatically rounded.
- Clearly displays if a rate control mode or codec is not supported.
- Added a thread for reading. --input-thread

It can be turned on and off with . - The initialization process has been significantly changed and simplified. Initialization is now much faster. 2016.02.29 (2.37) [QSVEncC] - Fixed memory release leaks. - Fixed an issue where the program would sometimes terminate with errors such as "Failed to SynchronizeFirstTask" and "Failed to get free surface for vpp pre." after 2.27. 2016.02.24 (2.36) [QSVEncC] - Fixed a problem where --seek did not work in some cases. - --avsync forcecfr only works when decoding MPEG, but in other cases, it will not terminate with an error, but will display a warning and continue after disabling it. 2016.02.21 (2.35) [QSVEncC] - Fixed a problem where the sound would sometimes be out of sync from the beginning when using --avsync forcecfr. - --avsync forcecfr is no longer available when decoding H.264. (For the time being, only MPEG2 decoding) - Fixed a crash during H.264 decoding. 2016.02.20 (2.34) [Common] - Fixed a bug where HEVC encoding was not possible after 2.32. [QSVEncC] - Added a setting for when you want to set an upper limit on the QSV processing speed. (--max-procfps) Default is 0 (no limit). This option is for when you are encoding multiple QSVs and do not want one stream to take up all the CPU/GPU power. - Added an option to keep video and audio synchronized. (--avsync) Currently, only through and forcecfr modes are implemented. Through (default) assumes the input is CFR as before and does not check the input pts. In forcecfr, the input pts are looked at and frames are padded or thinned to match CFR, so that synchronization with the audio can be maintained. This is mainly a measure to address the problem of audio being out of sync when the input is vfr or RFF. When using with vfr, please explicitly specify the fps you want to encode with --fps. - Added an option to seek before encoding. (--seek) Encoding starts after performing a fast but imprecise seek. If you want to specify a precise range, use --trim as before. The format is hh:mm:ss.ms. "hh" and "mm" can be omitted. - Improved information display when an error occurs on the command line. - Further improved input fps judgment. No misjudgment even with RFF. - Speeded up processing mainly for low resolution. 2016.02.15 (2.33) [QSVEnc] - Fixed a division by zero exception (0xc0000094). [QSVEncC] - Audio decode errors are ignored and processing continues. Errors are replaced with silence. Specifically, the number of consecutive decode errors is counted, and if it is within a threshold, the error is ignored and processing continues. The threshold can be set with --audio-ignore-decode-error

Set it with . The default is 10. If 0, if even one decoding error occurs, the process will be stopped and an error will occur. 2016.02.13 (2.32) [Common] - Improved the speed of the function check at the start of encoding by using an initialized session. - Fixed an error message being displayed when using VPP with d3d memory on Sandybridge even though it worked properly. [QSVEnc] - Fixed the progress display being 200% when bobifying. - Fixed an issue where the fps was not doubled when bobifying. [QSVEncC] - Fixed an issue where the --audio-stream option was not treated as the same track when specifying multiple audio-related options. - Fixed an exception that occurred when using --audio-stream when the ffmpeg dll does not exist. - Fixed an error that occurred when the output file folder does not exist. - Fixed an issue where --fade-detect was not working. - Organized help. - Added the range of values ​​for vpp-denoise and vpp-detail-enhance to help. - Fixed the inability to compile for Linux. 2016.02.10 (2.31) [Common] - Fixed a crash at the end of encoding when executed in a place without ffmpeg dll from 2.29 onwards. Aviutl's QSVEnc.auo crashed with 0xc06d007e exception (unknown application exception) [kernelbase.dll]. 2016.02.09 (2.30) [QSVEncC] - Fixed m3u8 files not being output correctly in hls output. - Updated ffmpeg_dll. Fixed a problem with udp reading. --avqsv-analyze is no longer required to be specified. It will work without problems if you do it like this: -i udp://127.0.0.1:1234?pkt_size=262144^&fifo_size=8000000 -o test.mp4 - Added option to load chapter file. (--chapter

) Supports Nero and Apple formats. Cannot be used with --chapter-copy. 2016.02.05 (2.29) [QSVEncC] - Updated ffmpeg_dll. Switched to speed optimization (-O3) to speed up audio encoding, etc. Also built with network and protocols enabled. This allows real-time encoding of UDP, etc. -i udp://127.0.0.1:1234?pkt_size=262144^&fifo_size=8000000 -o test.mp4 --output-thread 0 -a 1 --avqsv-analyze 10 - Added option to display dll version. (--check-avversion) - Added option to display supported protocols. (--check-protocols) - Added option to pass option parameters when muxing. (-m

:

) This is mainly used to pass the parameters required for HTTP Live Streaming output. For example, if you want to output for HLS, set it as follows. -i

-o test.m3u8 -f hls -m hls_time:5 -m hls_segment_filename:test_%03d.ts --gop-len 30 ・Error messages from libavcodec/libavformat can now be written to a log file. Previously, they were only displayed on the console. ・Added a function to convert the audio sampling frequency. (--audio-samplerate [

?]

) - Added an option to switch the engine used when converting audio sampling frequency. (--audio-resampler

) The options are "swr" (default = swresampler) and "soxr" (libsoxr). - Added an option to specify tracks and separate/merge audio channels. (--audio-stream [

?][

]) Typically, for dual mono audio, it is separated by using --audio-stream FR,FL, etc. At the same time, it can also be used to specify the number of audio channels, and by using --audio-stream stereo, the audio can always be converted to 2ch. It separates and merges audio channels. Audio tracks with --audio-stream specified are always encoded. By separating with a comma (,), multiple tracks can be generated from the same input track. Format:

Specify the track to be processed.

Specifies the channel to use as input. If omitted, all input channels are used.

Specifies the output channel format. If omitted,

Uses all channels from. Example 1: --audio-stream FR,FL This is the function that is likely to be most needed. Splits the left and right channels from dual mono into two mono audio tracks. Example 2: --audio-stream :stereo Converts any audio to stereo. Example 3: --audio-stream 2?5.1,5.1:stereo This generates a track that downmixes the second track of the input file to stereo, while encoding the 5.1ch audio as 5.1ch. I'm not sure if you'll ever actually use this, but it's a good example to introduce the format. Available symbols mono = FC stereo = FL + FR 2.1 = FL + FR + LFE 3.0 = FL + FR + FC 3.0(back) = FL + FR + BC 3.1 = FL + FR + FC + LFE 4.0 = FL + FR 4.0 = FL + FR + FC + BC quad = FL + FR + BL + BR quad(side) = FL + FR + SL + SR 5.0 = FL + FR + FC + SL + SR 5.1 = FL + FR + FC + LFE + SL + SR 6.0 = FL + FR + FC + BC + SL + SR 6.0(front) = FL + FR + FLC + FRC + SL + SR hexagonal = FL + FR + FC + BL + BR + BC 6.1 = FL + FR + FC + LFE + BC + SL + SR 6.1(front) = FL + FR + LFE + FLC + FRC + SL + SR 7.0 = FL + FR + FC + BL + BR + SL + SR 7.0(front) = FL + FR + FC + FLC + FRC + SL + SR 7.1 = FL + FR + FC + LFE + BL + BR + SL + SR 7.1(wide) = FL + FR + FC + LFE + FLC + FRC + SL + SR 2016.01.25 (2.28) [Common] - Fixed a problem where memory usage was greatly inflated when the video encoding speed was faster than the audio encoding speed. - Fixed a problem where synchronization during muxing could occur. [QSVEncC] - Changed perf-monitor-plot to be based on pyqtgraph. Faster than previous matplotlib-based. Requires python3.4 or later + pyqtgraph (+ numpy, PySide). 2016.01.17 (2.27) [Common] - Log information and error information for obtaining GPU device and GPU memory can now be obtained in detail. [QSVEnc] - Fixed the issue where AuoLink-related tabs were visible even when AuoLink was unavailable. - Fixed the issue where QuickTime could not be downloaded with the simple installer. [QSVEncC] - Added an option to specify the output buffer size. (--output-buf) Specifies the output buffer size in MB. Default is 64, maximum is 128. Set to 0 to not use. Allows you to change the size from 64MB, which was always reserved up until now. - Added an option to not use the output thread. (--no-output-thread) While the output thread speeds up encoding, it consumes a fair amount of memory. Therefore, to save memory, an option to not use the output thread has been added. The output thread is used by default. - Added an option to minimize memory usage (--min-memory). Currently, this is the same as "-a 1 --no-output-thread --output-buf 0 --input-buf 1". 2015.12.30 (2.26) [QSVEncC] - Fixed a bug where the SAR ratio was not reflected when outputting mkv files. - GPU usage can now be obtained if GPU-Z is running. - --trim can now be used when using --audio-source. - --trim can now be used with readers other than avqsv. 2015.12.24 (2.25) [QSVEncC] - Now reads at the fps specified by the avqsv reader. - Fixed a problem where a normal file could not be output because processing was interrupted when an error occurred in the audio header parser due to an incomplete packet such as a ts end. 2015.12.20 (2.24) [QSVEnc] - Now allows the use of avqsv in combination with AuoLink. [QSVEncC] - Fixed a bug where error codes for errors that occurred in the encoding pipeline were not properly retrieved. 2015.11.24 (2.23) [QSVEncC] - Fixed an issue where HEVC could not be decoded on Haswell. 2015.11.20 (2.22) [Common] - Supports API v1.17. - Supports fade detection. (--fade-detect) - Supports image rotation. Supports 90°, 180°, and 270°. (--vpp-rotate) - Fixed an issue where Skylake Pentium was detected as Broadwell... maybe. [QSVEnc] - Fixed an issue where CBR could not be selected on the settings screen. [QSVEncC] - Fixed an issue where seeking did not work properly (sound did not play for a while when seeking) when the time resolution was coarse or not divisible for mkv files, etc. - Fixed a typo in the help. 2015.11.15 (2.21) [Common] - Fixed an issue where the program would crash when qpmin/qpmax was specified. [QSVEncC] - Fixed an issue where the program would crash when using --check-environment in the x64 version. - It is now possible to change the quality settings to be measured during benchmarking. Also, the default has been changed from "All" to "1,4,7". - Added an option to perform performance analysis. (--perf-monitor) - Significantly improved the speed of audio encoding. (Windows only) 2015.11.02 (2.20) [Common] - Fixed a 1-bit incorrect flag position for rdseed. Broadwell was being detected as Haswell. - Fixed an issue where progress would stop at 100%. [QSVEnc] - Added audio delay cut parameters to QSVEnc.ini for fdk-aac (ffmpeg). - Removed the libmfxsw32.dll check, which was still in progress. [QSVEncC] - Fixed a problem where the program would freeze if the final packet was incomplete when encoding audio using ts, etc. 2015.10.30 (2.19) [QSVEnc] - An error when loading libmfxsw32.dll has become an issue, so encoding with sw has been disabled for the time being. - More detailed log display. [QSVEncC] - Fixed a problem where command line parsing did not work properly. - Fixed the inability to decode VC-1 in m2ts/ts. - Improved the handling of subtitles in m2ts etc. - Fixed the problem where subtitles were not muxed when audio was not muxed. - Improved debug log output. - Various internal fixes. 2015.10.16 (2.18) [QSVEnc] - No updates [QSVEncC] - Fixed a bug that sometimes caused block noise in the video when extracting, copying and encoding audio. 2015.10.11 (2.17) [QSVEnc] - Fixed a strange behavior in the setting screen for weighted B frames. - Fixed an exception that sometimes occurred when closing and reopening the setting screen. [QSVEncC] - No update 2015.10.10 (2.16) [Common] - Supports API 1.16. - Added options for weighted B frames and weighted P frames. --weightb, --weightp. [QSVEncC] - Added the ability to select subtitles to copy. Use --sub-copy to specify the subtitle number as 1, 2, 3, .... If nothing is specified, all subtitles will be copied as before. - Added VC-1 hw decode. - Fixed the issue where --disable-d3d was no longer effective. - When memory is specified automatically, if the output codec is raw, system memory will be automatically used. This is overwhelmingly faster. 2015.10.03 (2.15) [QSVEncC] - Added support for UTF-8 plain text subtitles. 2015.10.03 (2.14) [QSVEncC] - Fixed a bug where it was checking for old dlls and did not work. -Fixed the issue where H.264 was displayed twice in the level and profile notation in the help. 2015.10.03 (2.13) [Common] -Added H.264 Level 5.2. [QSVEncC] -Added the function to copy chapters. --chapter-copy -Added the function to copy subtitles. --sub-copy -Copy video meta information. -Can now display specifiable levels and profiles in the help. -Other fixes to the help. 2015.09.02 (2.12) [Common] -Migrated to VC++2015. -When encoding HEVC, the default "Number of consecutive B frames" is now 2. When set to 3 or more, block-like collapse occurs with some decoders. -Improved and fixed some log displays. [QSVEnc] -Updated the simple installer to coincide with the transition to VC++2015. [QSVEncC] - Fixed the problem that --check-features was not output correctly in the x64 version of 2.11. 2015.08.26 (2.11) [Common] - Supports Skylake HEVC. (HW encoding) However, it may still be unstable. Also, output while muxing is not supported at the moment. [QSVEncC] - Tried to support VP8, but it didn't work because it said there was no plugin. - Improved --check-features. By specifying --check-features <output file name>, it outputs to the specified file, and after output, it opens it with the default application. If the output file name is ".html", it will be output in html format, if it is ".csv", it will be output in csv format, and otherwise it will be output in the conventional txt format. - Added --audio-source. When used in combination with --audio-copy, it can mux external audio files. 2015.08.20 (2.10) [Common] - Added an option to perform full HW encoding using Skylake's Fixed Func. QSVEnc: FixedFunc, QSVEncC: --fixed-func 2015.08.13 (2.09) [Common] - Fixed an issue where the program would crash with exception 0xc0000005 when executed on Win10. - Added a function to detect Skylake generation. - Improved the display when GPU information could not be obtained via OpenCL. [QSVEncC] - Fixed an issue where the program would abnormally terminate when using vpp-delogo in d3d11 mode. - Fixed an issue where some --level options could not be specified correctly. - Added the --dar option. - Fixed an issue where raw output was not possible even when --format raw was specified. 2015.08.03 (2.08) [Common] - Changed the method for checking the OS version. [QSVEnc] - Fixed an issue where the log window would sometimes be hidden behind other windows. [QSVEncC] - Added --avsync-depth option. - Changed the option name from --lookahead-ds to --la-quality. The other lookahead options are la-xxx, but this one was lookahead, which was bad. - Added a mode to output without encoding. Specify -c raw. It is possible to run only QSV decode, QSV VPP, QSV decode + VPP, etc. and output with y4m. This allows you to run QSVEncC.exe -iSpecify with. If nothing is specified, all subtitles will be copied as before. - Added VC-1 hw decode. - Fixed the problem that --disable-d3d did not work. - When memory is specified automatically, if the output codec is raw, system memory will be automatically used. This is overwhelmingly faster. 2015.10.03 (2.15) [QSVEncC] - Added support for UTF-8 plain text subtitles. 2015.10.03 (2.14) [QSVEncC] - Fixed the problem that it did not work because it was checking old dll. - Fixed the problem that H.264 was displayed twice in the level and profile notation in the help. 2015.10.03 (2.13) [Common] - Added H.264 Level 5.2. [QSVEncC] - Added the function to copy chapters. --chapter-copy - Added the function to copy subtitles. --sub-copy &#12539;Modifies the copy of video meta information. &#12539;Added the ability to display the specifiable level and profile in the help section. &#12539;Other fixes to the help section. 2015.09.02 (2.12) [Common] &#12539;Migrated to VC++2015. &#12539;When encoding HEVC, the default for "Number of consecutive B frames" is now 2. When set to 3 or more, block-like collapse occurs in some decoders. &#12539;Improved and fixed some log display. [QSVEnc] &#12539;Updated the simple installer to coincide with the transition to VC++2015. [QSVEncC] &#12539;Fixed the issue where --check-features was not output correctly in the x64 version of 2.11. 2015.08.26 (2.11) [Common] &#12539;Supports Skylake HEVC. (HW encoding) However, it may still be unstable. Also, output while muxing is not supported at the moment. [QSVEncC] &#12539;I tried to support VP8, but it didn't work because I was told there was no plugin. - Improved --check-features. By specifying --check-features <output file name>, it will output to the specified file and open it with the default application after output. If the output file name is ".html", it will output in html format, if it is ".csv", it will output in csv format, otherwise it will output in the conventional txt format. - Added --audio-source. By using it together with --audio-copy, you can mux external audio files. 2015.08.20 (2.10) [Common] - Added an option to perform full HW encoding using Skylake's Fixed Func. QSVEnc: FixedFunc, QSVEncC: --fixed-func 2015.08.13 (2.09) [Common] - Fixed an issue where the program would crash with exception 0xc0000005 when executed on Win10. - Added a function to determine the Skylake generation. - Improved the display when GPU information cannot be obtained via OpenCL. [QSVEncC] - Fixed an issue where using vpp-delogo in d3d11 mode would cause an abnormal termination. - Fixed an issue where some --level options could not be specified correctly. - Added the --dar option. - Fixed an issue where raw output was not possible even when --format raw was specified. 2015.08.03 (2.08) [Common] - Changed the method for checking the OS version. [QSVEnc] - Fixed an issue where the log window would sometimes be hidden behind other windows. [QSVEncC] - Added the --avsync-depth option. - Renamed the option name from --lookahead-ds to --la-quality. The other options related to lookahead are la-xxx, but this one was lookahead, which was not a good idea. - Added a mode to output without encoding. Specify -c raw. It is possible to run only QSV decode, QSV VPP, QSV decode + VPP, etc. and output in y4m. This allows you to run QSVEncC.exe -iSpecify with. If nothing is specified, all subtitles will be copied as before. - Added VC-1 hw decode. - Fixed the problem that --disable-d3d did not work. - When memory is specified automatically, if the output codec is raw, system memory will be automatically used. This is overwhelmingly faster. 2015.10.03 (2.15) [QSVEncC] - Added support for UTF-8 plain text subtitles. 2015.10.03 (2.14) [QSVEncC] - Fixed the problem that it did not work because it was checking old dll. - Fixed the problem that H.264 was displayed twice in the level and profile notation in the help. 2015.10.03 (2.13) [Common] - Added H.264 Level 5.2. [QSVEncC] - Added the function to copy chapters. --chapter-copy - Added the function to copy subtitles. --sub-copy &#12539;Modifies the copy of video meta information. &#12539;Added the ability to display the specifiable level and profile in the help section. &#12539;Other fixes to the help section. 2015.09.02 (2.12) [Common] &#12539;Migrated to VC++2015. &#12539;When encoding HEVC, the default for "Number of consecutive B frames" is now 2. When set to 3 or more, block-like collapse occurs in some decoders. &#12539;Improved and fixed some log display. [QSVEnc] &#12539;Updated the simple installer to coincide with the transition to VC++2015. [QSVEncC] &#12539;Fixed the issue where --check-features was not output correctly in the x64 version of 2.11. 2015.08.26 (2.11) [Common] &#12539;Supports Skylake HEVC. (HW encoding) However, it may still be unstable. Also, output while muxing is not supported at the moment. [QSVEncC] &#12539;I tried to support VP8, but it didn't work because I was told there was no plugin. - Improved --check-features. By specifying --check-features <output file name>, it will output to the specified file and open it with the default application after output. If the output file name is ".html", it will output in html format, if it is ".csv", it will output in csv format, otherwise it will output in the conventional txt format. - Added --audio-source. By using it together with --audio-copy, you can mux external audio files. 2015.08.20 (2.10) [Common] - Added an option to perform full HW encoding using Skylake's Fixed Func. QSVEnc: FixedFunc, QSVEncC: --fixed-func 2015.08.13 (2.09) [Common] - Fixed an issue where the program would crash with exception 0xc0000005 when executed on Win10. - Added a function to determine the Skylake generation. - Improved the display when GPU information cannot be obtained via OpenCL. [QSVEncC] - Fixed an issue where using vpp-delogo in d3d11 mode would cause an abnormal termination. - Fixed an issue where some --level options could not be specified correctly. - Added the --dar option. - Fixed an issue where raw output was not possible even when --format raw was specified. 2015.08.03 (2.08) [Common] - Changed the method for checking the OS version. [QSVEnc] - Fixed an issue where the log window would sometimes be hidden behind other windows. [QSVEncC] - Added the --avsync-depth option. - Renamed the option name from --lookahead-ds to --la-quality. The other options related to lookahead are la-xxx, but this one was lookahead, which was not a good idea. - Added a mode to output without encoding. Specify -c raw. It is possible to run only QSV decode, QSV VPP, QSV decode + VPP, etc. and output in y4m. This allows you to run QSVEncC.exe -i09) [Common] - Fixed an issue where the program would crash with exception 0xc0000005 when executed on Win10. - Added a way to detect Skylake generation. - Improved the display when GPU information could not be obtained via OpenCL. [QSVEncC] - Fixed an issue where the program would abnormally terminate when using vpp-delogo in d3d11 mode. - Fixed an issue where some --level options could not be specified correctly. - Added the --dar option. - Fixed an issue where raw output was not possible even when --format raw was specified. 2015.08.03 (2.08) [Common] - Changed the way to check the OS version. [QSVEnc] - Fixed an issue where the log window would sometimes be hidden behind other windows. [QSVEncC] - Added the --avsync-depth option. - Renamed the option name from --lookahead-ds to --la-quality. The other options related to lookahead are la-xxx, but this one was lookahead, which was not a good idea. - Added a mode to output without encoding. Specify -c raw. You can run only QSV decode, QSV VPP, QSV decode + VPP, etc. and output with y4m. This allows you to run QSVEncC.exe -i09) [Common] - Fixed an issue where the program would crash with exception 0xc0000005 when executed on Win10. - Added a way to detect Skylake generation. - Improved the display when GPU information could not be obtained via OpenCL. [QSVEncC] - Fixed an issue where the program would abnormally terminate when using vpp-delogo in d3d11 mode. - Fixed an issue where some --level options could not be specified correctly. - Added the --dar option. - Fixed an issue where raw output was not possible even when --format raw was specified. 2015.08.03 (2.08) [Common] - Changed the way to check the OS version. [QSVEnc] - Fixed an issue where the log window would sometimes be hidden behind other windows. [QSVEncC] - Added the --avsync-depth option. - Renamed the option name from --lookahead-ds to --la-quality. The other options related to lookahead are la-xxx, but this one was lookahead, which was not a good idea. - Added a mode to output without encoding. Specify -c raw. You can run only QSV decode, QSV VPP, QSV decode + VPP, etc. and output with y4m. This allows you to run QSVEncC.exe -i

-o - -c raw --tff --vpp-deinterlace bob | x264 --demuxer y4m -o <output mp4 file> - QSVEncC.exe --avqsv -i

-o - -c raw --tff --vpp-deinterlace bob | x264 --demuxer y4m -o <output mp4 file> - and so on can now be executed. -Fixed an error when --profile was specified. 2015.07.21 (2.07) [QSVEnc] -Fixed a bug where muxing would not occur unless audio was encoded. 2015.07.20 (2.06) [Common] -Supported API v1.15. -Tentative support for Skylake HEVC encoding. -In IvyBridge, pyramid lookup is now off by default, and a warning is now displayed even if it is specified as on. [QSVEncC] -Updated the build of ffmpeg_lgpl_dll to gcc 5.1.0. Added swresample-1.dll. -Supported MPEG2 hw encoding. -Created a compatibility table for each codec with --check-features. -Supported audio encoding. Accordingly, --audio-codec and --audio-bitrate have been added. - The help for --mv-scaling has been fixed. - Available formats and codecs can now be checked. --check-codecs, --check-encoders, --check-decoders, --check-formats - Fixed a bug where 32-bit integers overflowed when calculating avgDuration with a large timebase, resulting in strange fps values. - Fixed a bug where the log display was strange when using the y4m reader. 2015.07.11 (2.05) [Common] - Added debug output. [QSVEncC] - Fixed a problem where --audio-file could not be output correctly. 2015.07.06 (2.04) [QSVEnc] - Fixed a crash that occurred when selecting a profile. 2015.07.05 (2.03) [QSVEncC] - Fixed a problem where file names were garbled on the output side. 2015.07.05 (2.02) [QSVEncC] - The avqsv reader now supports unicode file names. - Fixed the issue where --vpp-delogo and --vpp-half-turn could not be used except when using the avqsv reader. 2015.07.05 (2.01) [QSVEnc] - Moved to .NET Framework 4.5. - Windows 10 can now be detected correctly. - Fixed the issue where ALAC mode could not be muxed with qaac. - Fixed the issue where the size of the setting screen sometimes became large. [QSVEncC] - Added support for reading from standard input of the avqsv reader. - The --mux-video option has been abolished. - --format

You can now specify the format when outputting while muxing. If not specified, it will be automatically determined from the file extension. By specifying "raw", it will output in H.264/ES as before. - When outputting while muxing, it is now possible to output to standard output. - Added a transparency logo filter as vpp. Specify the logo file with --vpp-delogo-file, etc. Supports ".lgd", ".ldp", and ".ldp2". For logo packs, specify the logo name with --vpp-delogo-select, or specify an ini file for automatic selection. Other options include: > Adjust the logo position with 1/4 pixel accuracy with --vpp-delogo-pos > Correct transparency with --vpp-delogo-depth > Correct each color component with --vpp-delogo-y, --vpp-delogo-cb, --vpp-delogo-cr Processing is performed by the CPU using SSE4.1, AVX, and AVX2. Like Aviutl, it processes in YC48-12bit, but chrominance components are processed in 4:2:0. Also, since it is finally rounded to 8bit, high calculation accuracy is not necessary, so some processing is omitted to speed up. 2015.06.28 (v2.00&#946;12) [QSVEncC] - Fixed a bug where the argument for --lookahead-ds in the help was written incorrectly. - Fixed a bug where the argument for --trellis in the help was written incorrectly. - Fixed a bug where the value of --lookahead-ds was reflected in trellis. - Windows 10 can now be detected correctly. - Fixed a problem where y4m could not be loaded. - Fixed a problem where progress was not displayed correctly for some mpeg files. 2015.06.22 (v2.00&#946;11) [QSVEncC] - Fixed a problem where it was not possible to encode while muxing with interlace preservation in 2.00&#946;10. -Fixed a problem where audio delays still occurred even in 2.00&#946;10. 2015.06.21 (v2.00&#946;10) [QSVEncC] -Fixed a problem where audio delays occurred when encoding mkv/flv. -Fixed a problem where some MPEG2 files could not be encoded correctly. 2015.06.14 (v2.00&#946;9) [QSVEncC] -SAR ratio is no longer automatically reflected when resizing. -Fixed a problem where audio delays were likely to occur when audio started halfway through. -Fixed a problem where audio was shortened to match the video even when trim was not specified when it was longer than the video. -Fixed a problem where audio delays occurred with mkv, etc. -Fixed a problem where file names could not be read correctly when the audio file was only numbers with --copy-audio. -Fixed a problem where input files such as flv could not be decoded. 2015.06.07 (v2.00&#946;8) [QSVEncC] - Improved compatibility when muxing interlaced output. - Added the ability to specify the audio track to copy with --audio-file, and to specify multiple tracks. - Added a message indicating that avqsv + VQP is not supported. - Added some error messages. - Improved the issue of not being able to properly extract ac3 audio from m2ts, etc. - Improved handling of PAFF from m2ts, etc. 2015.05.31 (v2.00&#946;7) [QSVEncC] - Improved compatibility when muxing videos. - Added the ability to copy even if there are multiple audio tracks with --copy-audio. Also, added the ability to select the audio track to copy with --copy-audio. - Improved frame rate estimation. - Improved error handling when inputting a codec that cannot be decoded by QSV. - Improved accuracy of progress display. 2015.05.24 (v2.00&#946;6) [QSVEncC] - Fixed a bug where all readers except the avqsv reader would not work in 2.00&#946;5. 2015.05.24 (v2.00&#946;5) [QSVEncC] - Added muxing to mp4/mkv/mov. Automatically enabled by output file extension. To force it, use --mux-video. - Use --copy-audio to mux audio as well. - Improved audio lag with ts files. (Except RFF, RFF is not supported!) - Fixed a bug where the frame rate would double when H.264 PAFF was inserted in 2.00&#946;3. 2015.05.21 (v2.00&#946;4) [QSVEncC] - Improved the accuracy of frame rate estimation (mainly ts). - Added support for reading high bit depth from vapoursynth for vpp 10bit to 8bit color conversion testing. - Fixed an issue where HEVC 10bit decoding did not work properly on Broadwell. 2015.05.18 (v2.00&#946;3) [QSVEncC] - Fixed an issue where the frame rate doubled when H.264 PAFF was inserted into the avqsv reader. - Fixed an issue where the program would crash with "undefined behavior" when using --crop with the avqsv reader. - Fixed an issue where an inappropriate command line was not properly handled as an error. - Improved acquisition of GPU information. - Improved the accuracy of frame rate estimation. - Fixed a problem where wav output of PCM audio was not performed correctly. 2015.05.16 (v2.00&#946;2) [QSVEncC] - Added support for HEVC in mp4 decoding. - Now supports wav output for PCM audio such as DVD-Video/Bluray. - Progress can now be displayed in the avqsv reader. 2015.05.12 (v1.34) [QSVEnc] - Added the ability to batch process before and after audio encoding. - Fixed the issue where the frame rate was halved when muxing bob-ed video after 1.31. 2015.05.10 (v2.00&#946;) [QSVEncC] - Enabled QSV to perform decoding and encoding in a single process. Supports MPEG2, H.264, and HEVC decoding (HEVC only supports raw format). Added --avqsv, --audio-file, and --trim. This is still in the experimental stage and is unstable in many ways, so please only test it. 2015.04.05 (v1.33) [Common] - Fixed the issue where deinterlacing was not possible after 1.31 with the 4156 driver. Canceled the update to 1.31 to address the issue. 2015.03.21 (v1.32) [QSVEncC] - Fixed the issue where the encoding result was not output. - Improved the display of b-pyramid in the log. - Fixed the issue where --vpp-deinterlace was no longer effective. 2015.03.07 (v1.31) [Common] - API v1.Supports 13. - Added direct mode optimization (--direct-bias-adjust) and MV cost adjustment (--mv-scaling). - Added new deinterlacing modes. (Auto, Auto (bob), 24fps (fixed)) - Improved the display of b-pyramid in the log. - Improved the display of upper/lower QP limits in the log. 2015.03.04 (v1.30) [Common] - Removed the upper limit of QSVEnc bitrate (65535kbps). 2015.02.19 (v1.29v2) [QSVEnc] - Fixed the issue where QuickTime extraction was not performed correctly with the simple installer. This was because 7z.exe/7z.dll was not found in the setup folder. 2015.02.16 (v1.29) [QSVEnc] - Fixed an issue where qaac would not work if VC++2005 Runtime was not installed when installing using the simple installer. [QSVEncC] - Fixed a problem where the program could freeze when using the vpy reader. 2015.02.08 (v1.28) [QSVEnc] - Reduced one mux step when not using automatic field shift. 2014.11.20 (1.27) [Common] Supported API v1.11. - Supported LA_HRD (Look-Ahead Rate Control (HRD compatible)) mode. (--la-hrd) - Supported QVBR (Quality-based Variable Rate Control) mode. (--qvbr) - Added "Window-based rate control" that can be used for look-ahead rate control. (--la-window-size) - Added Image Stabilizer to Vpp. The effect is unclear. (--vpp-image-stab) - Improved the display of function information. - Improved the log display. 2014.11.10 (1.26) [Common] - Added OS version information to the log. - Added CPU usage during encoding. [QSVEnc] - Followed up to x264guiEx 2.23 feature additions - Added the ability to change the default audio encoder. - Fixed a problem where pause was not reflected in the progress display on the taskbar. - Added a function to cut audio delay using edts for qaac and fdk-aac. - Added --file-format to the muxer command. Fixed an issue where Apple format chapters were not reflected when using FAW. - Added audio and muxer logs to be output. - Added a dummy chapter when there is no chapter at 0 seconds. Prevented the first chapter from being assigned to 0 seconds ignoring the time specification when embedding Apple format chapters. - Added the ability to change the compression rate for flac. - Display muxer/audio encoder version in log. - Added opusenc to audio encoder. [QSVEncC] - Measure and display actual memory speed during benchmark. - High-precision timer is now available from QSVEncC. On by default. If you really want to turn it off, use --no-timer-period-tuning. 2014.07.01 (1.25) [Common] - Encoding under the following unstable conditions is now automatically avoided. - Lookahead series + scenechange is unstable (freezes) from API v1.8 onwards - Lookahead series with bframes >= 10 + b-pyramid is unstable (freezes) - b-pyramid + scenechange is unstable (image corruption) 2014.06.27 (1.24) [Common] - Fixed an issue where encoding with interlace preservation was not possible on environments with APIs below v1.6, such as Sandybridge. 2014.06.25 (1.23) [Common] - Improved QSV feature check - Correctly determines whether ICQ is available - Checks whether each mode supports interlace - Improved encoding information display - Fixed bitrate being displayed during ICQ - Also displays GPU information and driver version [QSVEnc] - Displays GPU name and driver number in feature information - Maximum bitrate specification other than CBR, VBR is disabled [QSVEncC] - AVX2 support in color conversion - avs reader also supports YUY2/RGB24/RGB32 reading 2014.06.21 (1.22) [Common] - Scene change detection is now forcibly disabled when using Lookahead mode, as using both Lookahead mode and scene change detection may cause the program to freeze. [QSVEnc] - Added a function to convert nero format chapters to UTF-8. From Other Settings. 2014.04.01 (1.21) [QSVEnc] - Fixed an issue where settings saved in "Aviutl Profile" in 1.19 and later were no longer compatible with 1.18 and earlier. 2014.03.28 (1.20) [QSVEnc] - Fixed an issue where chapters were not muxed when one-shot muxing was possible using only the muxer, such as when using faw2aac. Also improved the mux information display accordingly. 2014.03.25 (1.19) [QSVEnc] - Added a function to cut audio encoding delay (audio cut only) [QSVEncC] - Improved benchmark log - Displays CPU Boost clock if possible - Improved GPU information acquisition - Displays GPU driver version 2014.03.07 (1.18) [Common] - Fixed an issue where files were not output correctly in 1.16 and later. [QSVEnc] - Added the ability to change the save location for auto-saving logs. 2014.03.06 (1.17) [QSVEncC] - Adjusted output file in benchmark mode. - Now displays GPU information. - Layout adjustment. - Fixed the issue where the memory usage was showing as the remaining memory. - Vapoursynth Reader now supports x64. - Fixed the issue where the vpy reader would crash when encoding was interrupted after 1.08. 2014.03.04 (1.16) [Common] - Fixed the issue where the QSV encoding function could not be used after 1.12 when using Win8 + dGPU and not outputting from the iGPU. The meaning of the flags related to the video memory mode changed in Intel Media SDK 2014, but this was not supported. [QSVEncC] - Added benchmark mode for CQP mode (--benchmark). - Fixed the issue where the version information visible in Explorer was showing as x64 even for the x86 version after 1.12. 2014.03.01 (1.15) [Common] - Fixed the problem that LookaheadDepth was not reflected correctly from 1.12 onwards. [QSVEnc] - From 1.12 onwards, API v1.We think we've fixed the problem of freezing when opening the settings screen in environments that do not support API v1.6. [QSVEncC] - Now flushes the progress information so that it can be read immediately even when redirected. 2014.02.22 (1.14) [QSVEnc] - Slightly improved the time it took to open the settings screen since 1.12. - Fixed an error that would terminate if the specified log save location did not exist. 2014.02.20 (1.13) [Common] - We think we've fixed the problem of not working properly in environments that do not support API v1.6, such as Sandy. 2014.02.18 (1.12) [Common] - Switched to Intel Media SDK 2014 base, compatible with API v1.8. - Updated libmfxsw32.dll / libmfxsw64.dll. - Switched to VC++ 2013 Express for Desktop as the development environment. - Improved check of features supported by environment - In QSVEnc, it is displayed in the feature display tab - In QSVEncC, it can be checked by --check-features. - Color setting is now possible even with hardware encoding - colormatrix / colorprim / transfer - Added OpenGOP option. - Added API v1.6 features. - Rate control per macroblock - Enhanced rate control (only in bitrate specification mode) - Added API v1.8 features. - Added rate control mode > Fixed quality mode > Fixed quality mode with lookahead > Video conference mode - Adaptive I frame insertion - Adaptive B frame insertion - B frame pyramid reference - Lookahead quality setting (3 levels) - I think I did a lot of other things, but I can't remember [QSVEnc] - Display elapsed time - You can now pause by right-clicking on the log window - Slightly improved stability when outputting batches. 2013.12.07 (1.11v3) - Updated the easy installer - Fixed the problem that some files fail to be copied when extracting the easy installer to the Aviutl folder in the installation destination. 2013.11.24 (1.11v2) [QSVEnc] - Updated the easy installer - Fixed the problem that L-SMASH could not be downloaded. - When the installation destination requires administrator privileges, a dialog box to obtain them is now displayed. 2013.10.19 (v1.11) [QSVEnc] - Fixed the problem that the changed font (normal &#8660; italic) was not saved. - Fixed the problem that the memo of the setting file was not saved. - Updated the easy installer - Intended to be compatible with Windows 8.1 - Added the option to overwrite and update the presets when updating. [QSVEncC] - No updates. 2013.09.12 (v1.10) [Common] - Fixed the problem that strange things would happen when setting VPP deinterlace when the input is not set as interlaced. [QSVEncC] - Supports Unicode (basically processes wide strings). - Added the ability to save logs to files. Append type. (--log <file name>) 2013.08.25 (v1.09) [Common] - Fixed the problem in 1.08 where the "don't try d3d11" rule was not followed thoroughly. - Since d3d11 is generally slower than d3d9, d3d9 is used unless necessary. [QSVEncC] - Added the --d3d9/--d3d11 option to forcibly set the mode (if available). 2013.08.21 (v1.08) [Common] - Don't try d3d11 on Win7. [QSVEncC] - Improved error handling for avs/vpy reader. - Added vpy multi-thread mode (--vpy-mt). Uses the number of threads indicated by VapourSynth (up to a maximum of 127). 2013.08.05 (v1.07) [Common] - Made it possible to use QSV even when not outputting from the iGPU on Win8 + dGPU. It can be used by setting [iGPU Multi-Monitor] to Enabled in the BIOS (UEFI). - Supported d3d11 mode. [QSVEncC] - Fixed the problem that it was not working properly via y4m after 1.05. Thank you for pointing it out. - Made it so that information up to the middle of the process is displayed even when interrupted with Ctrl + C. 2013.08.02 (v1.06) [Common] - Enabled scene change detection and VQP when deinterlacing "normal" or "Bob" with vpp. - Added a message that Lookahead mode cannot be used with interlace preserving encoding. [QSVEnc] - Revised distribution profile and added some setting examples. [QSVEncC] - Added a reader for Vapoursynth r19 to the x86 version. Automatically uses Vapoursynth Reader with vpy extension. Slightly (up to 5%) faster than vfw reading. Only for reading yv12. Automatically switches to avi (vfw) reading for files other than yv12 (yuy2, RGB, etc.). 2013.07.20 (v1.05v2) [QSVEncC] - Fixed help display. 2013.07.20 (v1.05) [QSVEncC] - Added Avisynth Reader. Automatically uses Avisynth Reader with avs extension. Slightly (up to 10%) faster than vfw reading. 2013.07.13 (v1.04) [QSVEnc] - Fixed an issue where the maximum bitrate could not be specified in Lookahead mode. [QSVEncC] - Supports Lookahead mode. 2013.07.08 (v1.03) - Fixed the issue where a debug log file was being output. 2013.07.07 (v1.02) [QSVEnc] - Fixed the issue where an error occurred when trying to save a new profile. Thank you for pointing this out. - Revised the profile a little. [QSVEncC] - No particular changes. 2013.07.03 (v1.01) [Common] - Transitioned to Intel Media SDK 2013 base, compatible with API v1.7. - Updated libmfxsw32.dll / libmfxsw64.dll. - Removed API v1.6 functions that have no effect even if specified. - Macroblock-based rate control - Enhanced rate control (only in bitrate specification mode) - Added API v1.7 functions. - Lookahead rate control - Distortion rate optimization (trellis) [QSVEncC] - Fixed the issue where the quality option could not be read correctly. 2013.07.01 (v1.00) [Common] - Development environment changed to VC++ 2012 Express for Desktop. - Changed to .NET Framework 4.0 Client at the same time. - Removed Win Vista from the operating environment. Intel Media SDK and driver compatibility is unclear. - AVX/AVX2 support for color space conversion and scene change detection. As always, the fastest one is automatically selected. - Moved to Intel Media SDK 2013 base, which makes it compatible with API v1.6. - Updated libmfxsw32.dll / libmfxsw64.dll. - Added the following API v1.6 features. - Macroblock rate control - Enhanced rate control (only in bitrate specification mode) - Expanded quality settings from 3 levels to 7 levels. - Updated the simple installer to match v1.00. - Fixed the issue of flickering in gradual fade scenes when scene changes are detected. [QSVEnc] - Added the functions "Play video" and "Open folder containing video" by right-clicking on the output file name in the log window. [QSVEncC] - Changed the --quality option. There are 7 levels: best, higher, high, balanced(default), fast, faster, fastest. 2013.05.23 (v0.23v2) [Common] - Added Bluray mode forcing. Add force_bluray=1 to the [QSVEnc] section of QSVEnc.conf. 2013.05.12 (v0.23) [Common] - Improved encoding information display. [QSVEncC] - Supports AVI (vfw) reading. - Enabled direct reading of files that can be read via vfw, such as avi/avs/vpy. - Automatically switches to avi reading for avi/avs/vpy extensions. Explicitly specify with --avi. - Supports YV12/YUY2/RGB24/RGB32 reading, YV12 recommended. - Interlace is supported for YV12/YUY2 reading. - RGB24/RGB32 reading is not compatible with interlace because vpp converts RGB32 to NV12. - Displays progress and remaining time. - Displays the time it took to encode at the end. *There are no plans to implement a method to directly call Avisynth's API for certain reasons. 2013.05.05 (v0.22) [Common] - Added bob (60fps) to vpp deinterlace. - By setting the maximum GOP length to "0", fps x 10 is automatically set. 2013.05.05 (v0.21) [QSVEnc] - Fixed an issue where temporary video files remained. [QSVEncC] - Read the aspect ratio when y4m is input. 2013.05.04 (v0.20) [Common] - Changed the default input frame buffer from 4 to 3 when encoding with HW. - Adjusted the number of internal pipeline buffers to speed things up a bit. [QSVEnc] - Slightly faster yuy2 to nv12 conversion. - Reflects updates up to x264guiEx 1.75. - Even if the free disk space cannot be obtained during muxing, a warning is displayed and the process continues. - Fixed the behavior when clicking "Default" on the settings screen. - Fixed the behavior of the pipe and 2pass checkboxes in the audio settings. - Added a setting to run pre- and post-encoding batch processing in minimized mode. [QSVEncC] - Speeded up the yv12 to nv12 conversion when loading frames by adjusting it. - Added crop option. - Fixed the display of the results after encoding being distorted. 2013.03.08 (v0.19v2) [Common][QSVEnc] None. [QSVEncC] - Added a format for specifying the color format of y4m. 2013.02.14 (v0.19) Thank you for reporting the problem. [Common] - Fixed the problem that encoding failed in Baseline Profile after 0.15. [QSVEncC] - Fixed sar being displayed twice in the help. 2013.01.31 (v0.18) [Common] - Fixed the problem that "undefined behavior" was displayed and encoding did not start in some environments in 0.17. Thank you for reporting it. - Further accelerated scene change detection and variable QP adjustment calculation for x86. 2013.01.26 (v0.17) [Common] - Fixed the problem that frames were periodically corrupted when encoding with interlace preservation was checked in 0.15 or later, or in 0.14 or earlier. Thank you for reporting the error. - Improved the speed of scene change detection and variable QP adjustment calculation. 2013.01.23 (v0.16) [Common] - Fixed the problem that scene change detection and variable QP were specifying different behavior than intended. Only the bottom half of the screen was being viewed. 2013.01.22 (v0.15) [Common] - Added forced I-frame insertion function by scene change detection. - Added variable QP mode. Both are only effective when the input is progressive (non-interlaced). - After encoding, the total size of each frame type is displayed. [QSVEnc] - Fixed an issue where automatic field shift was unstable. - Added a setting to improve timer accuracy to other settings. - Added a setting to run pre- and post-encoding batch processing in minimized mode. From other settings. 2012.12.26 (v0.14) [QSVEnc] - Added the ability to change the default output file type. From other settings. Aviutl must be restarted after the setting to reflect the change. 2012.12.22 (v0.13) [QSVEnc] - Supports automatic field shift. Please use the ini file for L-SMASH muxer (the one included with auo). It will not work with the ini file for mp4box. - Reflects the updates up to x264guiEx 1.65. - Saves the position of the log window. - Fixed display collapse at high DPI settings. - Fixed a possible crash at the start of encoding. 2012.11.15 (v0.12) [QSVEnc] - Reflects the updates up to x264guiEx 1.62. - Specify the color of the log window. - Fixed display collapse at high DPI. 2012.11.02 (v0.11) [QSVEnc] - Added fps specification to the muxer command. - Reflects the updates up to x264guiEx 1.61. - Imports audio encoding / muxer messages and displays errors. - Improved display when encoding video and audio simultaneously. - Specification of log window transparency. 2012.10.20 (v0.10) [Common] - v0.Since version 08, if the Intel iGPU was not the primary GPU (the GPU outputting to the main monitor), the video memory mode could not be used, causing the error: null pointer. Fixed. - The colormatrix, coloprim, and transfer settings do not work with HW encoding (QSV), so the settings screen and help display have been changed accordingly. [QSVEnc] - Fixed the Bluray compatible output checkbox being unchecked. - Reflects updates up to x264guiEx 1.59. - Added pre-encoding batch processing. - Added a link to the blog. 2012.10.13 (v0.09) [Common] - Added output settings for Bluray. [QSVEnc] - Added a preset for Bluray output. 2012.10.06 (v0.08) [Common] - Supports Intel Media SDK 2012 R3 (API v1.4). - It seems to be compatible with Windows8 + DirectX 11.1. - Doesn't it have any special meaning on Win7? [QSVEnc] - Added simple installer. - Reflects updates up to x264guiEx 1.57. - Added simultaneous video and audio processing mode. Audio processing order can be selected from "after", "before", and "simultaneous". - Right-click on the button that specifies the executable file to display help for the currently specified executable file. - In QSVEnc.ini, more than one audio/mux file name can now be written. Specifically, either can be used, like filename="ffmpeg.exe;avconv.exe". - For interlaced i420 conversion, use 3,1-weighted average instead of simple average as before. - Added warning when audio length and video length are significantly different. - Fixed 1/2 size FAWCheck misjudging AAC over 384kbps as non-FAW. - Added flac / fdk-aac (ffmpeg/avconv) settings to audio settings. [QSVEncC] - Nothing in particular. 2012.07.08 (v0.07) [Common] - The options added in v0.06 are now only valid for soft encoding. - Because it doesn't seem to be reflected in hardware encoding (QSV). - Omitted unused MVC-related code - Other small adjustments [QSVEnc] Needs to update QSVEnc.ini - L-SMASH support - muken has updated the L-SMASH muxer so that PAFF H.264/ES can be imported. Thank you! - QSVEnc will be changed to use the L-SMASH ini file as the base. Specify L-SMASH muxer / remuxer instead of mp4box. - Use L-SMASH rev600 or later. - I'll include the mp4box ini just in case... - Reflects updates from x264guiEx 1.46 to 1.49. - Support for faw2aac.auo > Display faw2aac progress in the log window. > If faw2aac.auo is available, FAW can be processed even without fawcl. - Support for L-SMASH (uses the same code as x264guiEx) - Minor bug fixes > Fixed a bug where the default folder on the "Open executable..." screen would sometimes be incorrect. > Fixed an error message that would be displayed when the root of the saved "stg setting file location" does not exist. - Display fixes > Fixed an error message that said "There is no video file" when muxing was not possible because there was no audio file. - Fixed the replacement of %{chpath} from the temporary file name to the output file name. [QSVEncC] - Nothing in particular. 2012.05.11 (v0.06) [QSVEnc] - Reflected some of the updates from x264guiEx 1.42. - Fixed an issue where relative paths were not obtained correctly on WinXP. On WinXP, the PathRelativePathTo function used to get a relative path sometimes returns "\dir" when it should. Although QSVEnc does not support XP, I added a workaround just in case. - Added a CVBR mode setting to the qtaaacenc/qaac settings in QSVEnc.ini. Since it is inserted between ABR and TVBR, which is the natural order, the index of the setting file may be shifted. [QSVEncC] - Fixed the problem that y4m + pipe does not work in v0.05. 2012.05.05 (v0.05) [Common] - Supports Intel Media SDK 2012 R2 (API v1.4). - Updated libmfxsw**.lib (API v1.4) - There are no significant updates in API v1.4. - Fixed the behavior of vpp. - Improved the display of information. Also added encoding information to display. - Added a setting item that was previously only automatically set in "Quality". However, this depends to some extent on the "quality" setting, so it may not always be exactly as expected. - MV search range and MV search accuracy settings. - CAVLC/CABAC settings. - Distortion rate optimization (RDO, requires CABAC) settings. - Block size settings for inter-frame prediction/intra-frame prediction. - Improved behavior when specifying a profile. - Many other small improvements and changes. I've already forgotten. [QSVEnc] * Requires updating QSVEnc.ini * Some configuration files are not compatible. (SAR ratio, etc.) Please set them again. - SAR is now specified at the time of encoding, not at the time of muxing. - Added an option to re-specify SAR ratio with mp4box. - Reflects the following updates from x264guiEx 1.34 to 1.41. - Folder structure can now be reflected in the display of configuration files (stg files). In addition, a folder can now be specified when saving new configuration files. - Improved the settings save process. - The simple bitrate calculator can now calculate the length of a video using the number of frames + frame rate. You can switch between hours, minutes, and seconds <-> number of frames with the button below. - When you select to mux a chapter file but the chapter file does not exist, muxing will still go through even if you use L-SMASH remuxer or mkvmerge. - When you click "Open File..."Added a setting to make the path obtained from " " etc. a relative path. From other settings. [QSVEncC] - Added options and added/modified help. [Current issues] (Intel 2696 driver from Windows Update, API v1.3) - Interlace preserving encoding is practically impossible. (Double frame count problem) - Colormatrix, colorprim, and transfer do not work with hw encoding (QSV). (colour_description_present_flag = 0 (h264_parse)) - Scene change detection is not possible with hw encoding (QSV). I implemented it and it worked in software mode, but it did not work well with hw encoding, so it has been disabled. Reference: http://software.intel.com/en-us/forums/showthread.php?t=103385 2012.02.27 (0.04v2) [QSVEncC] - Fixed some command lines not working properly. - You can now check whether the binary is x86 or x64 by right-clicking and selecting Properties. 2012.02.23 (v0.04) [Common] - Provisional support for Intel Media SDK 2012 (provisional support for API v1.3) - Added color settings - Added AVBR mode - Updated libmfxsw**.lib (API v1.3) - Detects and displays API version (QSVEnc in the settings screen, QSVEncC with --lib-check) [QSVEnc] - Reflects updates from x264guiEx 1.27 to 1.33 - Improved stability. - Added the ability to save the size of the log window. - Specify STAThreadAttribute. - Fixed a bug that caused "Post-encoding batch processing" to fail when muxing was not performed. - Added support for dragging and dropping the batch file specification field for post-recording batch processing. - Added support for simple WAV output for audio encoding. - Improved the process of saving settings files. - Improved file size acquisition. - Added the ability to change fonts in the settings screen and log window. - Fixed the oggenc2 command line in QSVEnc.ini. It was being forced to 44.1kHz. - Added settings for qaac/refalac to QSVEnc.ini. - Added settings for AnonEncoder to QSVEnc.ini. [QSVEncC] - Statically linked x86 version as well. (For QSVEncC only, VC++2008 Redistributable Package is not required) - -o - can be used to output to stdout (standard output). 2012.01.22 (v0.03v2) - Fixed a problem where the setting screen did not appear. 2012.01.22 (v0.03) Requires updating of ini file (ini file version 1 -> 2) - Loads frames in a separate thread. Should speed things up a bit. Added frame loading buffer function. Buffer size is 1-16 frames. Improves partial parallelization efficiency. In QSVEnc, you can specify the size with the read buffer size (third tab), and in QSVEncC, you can specify the size with --input-buf. Be careful, because if it's too large, it will slow down. (I think it's related to the cache size) Encoding speed Appropriate buffer size Up to 50fps 1-2 Up to 100fps 2-3 Up to 200fps 3-6 More than that 4-8 The faster the encoding speed, the more effective it should be. On the other hand, it doesn't have much effect when encoding full HD. Be careful, because if you set the buffer size too large when using QSV, it will not be able to secure GPU memory and will crash. -Incorporated additional features from x264guiEx v1.11-v1.26 and reflected bug fixes -Relative paths can now be used. -Suppressed tooltip help (from other settings) -Fixed the issue where "Start this window minimized" only took effect once and was then canceled. -Visual effects can now be turned off. -You can now leave notes in the profile. Displayed to the right of the profile. You can change it by double-clicking. -Fixed the problem that an exception occurs before an error message when QSVEnc.ini does not exist or is old. -Make it possible to output mp4/chapter compatible with Apple systems using mp4box. -Added log window related settings ("Transparent" and "Start minimized") to other settings. -Make it possible to read settings saved in the profile on the Aviutl side even if the size of the setting file is different. -Make it possible to turn on/off automatic deletion of chapter files. From other settings. -Make it possible to succeed in muxing even if the chapter file does not exist. -Make it possible to close the log window with the Esc key when not encoding. -Added batch file execution after encoding. -Added on/off setting of Esc key in the settings screen. -Mux error countermeasure. Fundamentally improved check during muxing. -Added x64 version to QSVEncC. Slightly faster...maybe. Basically not much different. Well, it means that you can build x64. -Prevent unnecessary memory release->reallocation. 2012.01.16 (QSVEncC_20120116) - Organized console output into stderr and stdout. Stderr as the default. 2011.10.02 (v0.02) - Enabled QSV to be used even when Intel GPU is not the primary GPU. Also fixed the problem that d3d memory mode could not be used in that case. (vpp performance drops when not in d3d memory mode) - Improved QSVEncC to not spit out incomprehensible error messages. 2011.09.27 (v0.01) - Fixed the problem that fps display was strange. No effect on encoding. 2011.09.26 (v0.00) - Released version 2011.09.26 - Added the option to specify the location of automatic log save. 2011.09.25 - Fixed many bugs. - Display the breakdown of encoded frame types in the log. 2011.09.24 &#12539;Added various messages. 2011.09.23 &#12539;Added slices setting item. &#12539;I tried making a command line version. QSVencC. Well, it's like BonTSDemuxC for BonTSDemux. I just wanted to try pipe input. But it should work with Avisynth. 2011.09.22 I don't understand vpp SceneChangeDetection. So it's not supported. 2011.09.20 Supported vpp deinterlacing (I don't know how to use IT) 2011.09.19 To be honest, I found out in 0.02 that color conversion (YUY2->NV12) only slows things down if you use hw vpp. I had no choice but to add other vpp (Resize,Denoise,DetailEnhancer) (What kind of direction is this going in...?) 2011.09.19 hw vpp (YUY2 -> NV12) support. Will it be faster with GPU EU processing, or will it be slower with increased memory copies? &#8594;It's slower orz 2011.09.18 Added B frame settings, GOP length, d3d mem mode, etc. The conditions that prevent seeking have been clarified. Improved YUY2->NV12 to direct conversion. 2011.09.18 It works. 2011.09.17 (on sample_encode.exe) Preserve interlace + hw encode is weird.