search

intel / libva

Libva is an implementation for VA-API (Video Acceleration API)
http://intel.github.io/libva/
Other
650 stars 299 forks source link

Setting RC_Quality has no effect when encoding VP9 with RC_MODE ICQ on DG2 #776

Open danielsmakandra opened 9 months ago

danielsmakandra commented 9 months ago

I try to encode videos to the VP9 codec while using the ICQ rc_mode with ffmpeg and vaapi directly.

Hardware

Intel Arc A380 on Archlinux VM

Software

Problem

The RC quality setting isn't respected by libva

Running sudo ffmpeg -an -v verbose -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format vaapi -i ~/videos/V2023CW0079.mp4 -c:v vp9_vaapi -global_quality 100 /tmp/V2023CW0079_icq_100.webm
Sets the RC_Quality setting correctly as indicated by ffmpeg's output

Running with quality setting 1 also sets the RC_Quality correctly: sudo ffmpeg -an -v verbose -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format vaapi -i ~/videos/V2023CW0079.mp4 -c:v vp9_vaapi -global_quality 1 /tmp/V2023CW0079_icq_1.webm

But both commands produce output files with almost the same filesize and similar quality:

4,0M    /tmp/V2023CW0079_icq_100.webm
4,1M    /tmp/V2023CW0079_icq_1.webm

As a check to see if I got the ffmpeg parameters wrong I tried the same command with the h264 encoder. With RC_Quality set to 1 and 100. It produces files with vastly different filesizes:

3,5M    /tmp/V2023CW0079_icq_100.mp4
81M /tmp/V2023CW0079_icq_1.mp4

Because of this I suspect that the Bug is within libva.

Logs

with ffmpeg -global_quality 100: 

ffmpeg version n6.1 Copyright (c) 2000-2023 the FFmpeg developers
  built with gcc 13.2.1 (GCC) 20230801
  configuration: --prefix=/usr --disable-debug --disable-static --disable-stripping --enable-amf --enable-avisynth --enable-cuda-llvm --enable-lto --enable-fontconfig --enable-gmp --enable-gnutls --enable-gpl --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libdav1d --enable-libdrm --enable-libfreetype --enable-libfribidi --enable-libgsm --enable-libiec61883 --enable-libjack --enable-libjxl --enable-libmodplug --enable-libmp3lame --enable-libopencore_amrnb --enable-libopencore_amrwb --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librav1e --enable-librsvg --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtheora --enable-libv4l2 --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpl --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxcb --enable-libxml2 --enable-libxvid --enable-libzimg --enable-nvdec --enable-nvenc --enable-opencl --enable-opengl --enable-shared --enable-version3 --enable-vulkan
  libavutil      58. 29.100 / 58. 29.100
  libavcodec     60. 31.102 / 60. 31.102
  libavformat    60. 16.100 / 60. 16.100
  libavdevice    60.  3.100 / 60.  3.100
  libavfilter     9. 12.100 /  9. 12.100
  libswscale      7.  5.100 /  7.  5.100
  libswresample   4. 12.100 /  4. 12.100
  libpostproc    57.  3.100 / 57.  3.100
[h264 @ 0x55822cfa8f80] Reinit context to 1920x1088, pix_fmt: yuv420p
Selecting decoder 'h264' because of requested hwaccel method vaapi
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '/home/daniel/videos/V2023CW0079.mp4':
  Metadata:
    major_brand     : mp42
    minor_version   : 0
    compatible_brands: mp42mp41
    creation_time   : 2023-09-04T15:21:56.000000Z
  Duration: 00:01:05.68, start: 0.000000, bitrate: 14616 kb/s
  Stream #0:0[0x1](eng): Video: h264 (Main), 1 reference frame (avc1 / 0x31637661), yuv420p(tv, bt709, progressive, left), 1920x1080 (1920x1088) [SAR 1:1 DAR 16:9], 14291 kb/s, 25 fps, 25 tbr, 25k tbn (default)
    Metadata:
      creation_time   : 2023-09-04T15:21:56.000000Z
      handler_name    : ?Mainconcept Video Media Handler
      vendor_id       : [0][0][0][0]
      encoder         : AVC Coding
  Stream #0:1[0x2](eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 317 kb/s (default)
    Metadata:
      creation_time   : 2023-09-04T15:21:56.000000Z
      handler_name    : #Mainconcept MP4 Sound Media Handler
      vendor_id       : [0][0][0][0]
[out#0/webm @ 0x55822cfab300] No explicit maps, mapping streams automatically...
[vost#0:0/vp9_vaapi @ 0x55822cfcf800] Created video stream from input stream 0:0
[AVHWDeviceContext @ 0x55822d1322c0] libva: VA-API version 1.20.0
[AVHWDeviceContext @ 0x55822d1322c0] libva: Trying to open /usr/lib/dri/iHD_drv_video.so
[AVHWDeviceContext @ 0x55822d1322c0] libva: Found init function __vaDriverInit_1_20
[AVHWDeviceContext @ 0x55822d1322c0] libva: va_openDriver() returns 0
[AVHWDeviceContext @ 0x55822d1322c0] Initialised VAAPI connection: version 1.20
[AVHWDeviceContext @ 0x55822d1322c0] VAAPI driver: Intel iHD driver for Intel(R) Gen Graphics - 23.3.5 ().
[AVHWDeviceContext @ 0x55822d1322c0] Driver not found in known nonstandard list, using standard behaviour.
File '/tmp/V2023CW0079_icq_100.webm' already exists. Overwrite? [y/N] y
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> vp9 (vp9_vaapi))
Press [q] to stop, [?] for help
[h264 @ 0x55822d121340] Reinit context to 1920x1088, pix_fmt: vaapi
[graph 0 input from stream 0:0 @ 0x55822d33fb80] w:1920 h:1080 pixfmt:vaapi tb:1/25000 fr:25/1 sar:1/1
[vp9_vaapi @ 0x55822cffdf40] Using input frames context (format vaapi) with vp9_vaapi encoder.
[vp9_vaapi @ 0x55822cffdf40] Input surface format is nv12.
[vp9_vaapi @ 0x55822cffdf40] Using VAAPI profile VAProfileVP9Profile0 (19).
[vp9_vaapi @ 0x55822cffdf40] Using VAAPI entrypoint VAEntrypointEncSliceLP (8).
[vp9_vaapi @ 0x55822cffdf40] Using VAAPI render target format YUV420 (0x1).
[vp9_vaapi @ 0x55822cffdf40] RC mode: ICQ.
[vp9_vaapi @ 0x55822cffdf40] RC quality: 100.
[vp9_vaapi @ 0x55822cffdf40] RC framerate: 25/1 (25.00 fps).
[vp9_vaapi @ 0x55822cffdf40] Driver does not report any additional prediction constraints.
[vp9_vaapi @ 0x55822cffdf40] Using intra and P-frames (supported references: 3 / 0).
[vp9_vaapi @ 0x55822cffdf40] All wanted packed headers available (wanted 0, found 0x10).
Output #0, webm, to '/tmp/V2023CW0079_icq_100.webm':
  Metadata:
    major_brand     : mp42
    minor_version   : 0
    compatible_brands: mp42mp41
    encoder         : Lavf60.16.100
  Stream #0:0(eng): Video: vp9 (Profile 0), 1 reference frame, vaapi(tv, bt709, progressive, left), 1920x1080 (0x0) [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 1k tbn (default)
    Metadata:
      creation_time   : 2023-09-04T15:21:56.000000Z
      handler_name    : ?Mainconcept Video Media Handler
      vendor_id       : [0][0][0][0]
      encoder         : Lavc60.31.102 vp9_vaapi
Automatically inserted bitstream filter 'vp9_superframe'; args=''eed=N/A    
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x55822cfa8040] EOF while reading input7.9kbits/s speed=6.75x    
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x55822cfa8040] Terminating demuxer thread
[vist#0:0/h264 @ 0x55822cfaabc0] Decoder thread received EOF packet
[vist#0:0/h264 @ 0x55822cfaabc0] Decoder returned EOF, finishing
[vist#0:0/h264 @ 0x55822cfaabc0] Terminating decoder thread
No more output streams to write to, finishing.
[out#0/webm @ 0x55822cfab300] All streams finished
[out#0/webm @ 0x55822cfab300] Terminating muxer thread
[AVIOContext @ 0x55822d0b9fc0] Statistics: 4174581 bytes written, 2 seeks, 17 writeouts
[out#0/webm @ 0x55822cfab300] Output file #0 (/tmp/V2023CW0079_icq_100.webm):
[out#0/webm @ 0x55822cfab300]   Output stream #0:0 (video): 1642 frames encoded; 1642 packets muxed (4161507 bytes); 
[out#0/webm @ 0x55822cfab300]   Total: 1642 packets (4161507 bytes) muxed
[out#0/webm @ 0x55822cfab300] video:4064kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.299315%
frame= 1642 fps=169 q=-0.0 Lsize=    4076kB time=00:01:05.64 bitrate= 508.7kbits/s speed=6.77x    
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x55822cfa8040] Input file #0 (/home/daniel/videos/V2023CW0079.mp4):
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x55822cfa8040]   Input stream #0:0 (video): 1642 packets read (117336895 bytes); 1642 frames decoded; 0 decode errors; 
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x55822cfa8040]   Total: 1642 packets (117336895 bytes) demuxed
[AVIOContext @ 0x55822cfb0840] Statistics: 120004418 bytes read, 0 seeks

With ffmpeg -global_quality 1:

ffmpeg version n6.1 Copyright (c) 2000-2023 the FFmpeg developers
  built with gcc 13.2.1 (GCC) 20230801
  configuration: --prefix=/usr --disable-debug --disable-static --disable-stripping --enable-amf --enable-avisynth --enable-cuda-llvm --enable-lto --enable-fontconfig --enable-gmp --enable-gnutls --enable-gpl --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libdav1d --enable-libdrm --enable-libfreetype --enable-libfribidi --enable-libgsm --enable-libiec61883 --enable-libjack --enable-libjxl --enable-libmodplug --enable-libmp3lame --enable-libopencore_amrnb --enable-libopencore_amrwb --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librav1e --enable-librsvg --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtheora --enable-libv4l2 --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpl --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxcb --enable-libxml2 --enable-libxvid --enable-libzimg --enable-nvdec --enable-nvenc --enable-opencl --enable-opengl --enable-shared --enable-version3 --enable-vulkan
  libavutil      58. 29.100 / 58. 29.100
  libavcodec     60. 31.102 / 60. 31.102
  libavformat    60. 16.100 / 60. 16.100
  libavdevice    60.  3.100 / 60.  3.100
  libavfilter     9. 12.100 /  9. 12.100
  libswscale      7.  5.100 /  7.  5.100
  libswresample   4. 12.100 /  4. 12.100
  libpostproc    57.  3.100 / 57.  3.100
[h264 @ 0x561b429b4f80] Reinit context to 1920x1088, pix_fmt: yuv420p
Selecting decoder 'h264' because of requested hwaccel method vaapi
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '/home/daniel/videos/V2023CW0079.mp4':
  Metadata:
    major_brand     : mp42
    minor_version   : 0
    compatible_brands: mp42mp41
    creation_time   : 2023-09-04T15:21:56.000000Z
  Duration: 00:01:05.68, start: 0.000000, bitrate: 14616 kb/s
  Stream #0:0[0x1](eng): Video: h264 (Main), 1 reference frame (avc1 / 0x31637661), yuv420p(tv, bt709, progressive, left), 1920x1080 (1920x1088) [SAR 1:1 DAR 16:9], 14291 kb/s, 25 fps, 25 tbr, 25k tbn (default)
    Metadata:
      creation_time   : 2023-09-04T15:21:56.000000Z
      handler_name    : ?Mainconcept Video Media Handler
      vendor_id       : [0][0][0][0]
      encoder         : AVC Coding
  Stream #0:1[0x2](eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 317 kb/s (default)
    Metadata:
      creation_time   : 2023-09-04T15:21:56.000000Z
      handler_name    : #Mainconcept MP4 Sound Media Handler
      vendor_id       : [0][0][0][0]
[out#0/webm @ 0x561b429b7300] No explicit maps, mapping streams automatically...
[vost#0:0/vp9_vaapi @ 0x561b429db800] Created video stream from input stream 0:0
[AVHWDeviceContext @ 0x561b42b3e2c0] libva: VA-API version 1.20.0
[AVHWDeviceContext @ 0x561b42b3e2c0] libva: Trying to open /usr/lib/dri/iHD_drv_video.so
[AVHWDeviceContext @ 0x561b42b3e2c0] libva: Found init function __vaDriverInit_1_20
[AVHWDeviceContext @ 0x561b42b3e2c0] libva: va_openDriver() returns 0
[AVHWDeviceContext @ 0x561b42b3e2c0] Initialised VAAPI connection: version 1.20
[AVHWDeviceContext @ 0x561b42b3e2c0] VAAPI driver: Intel iHD driver for Intel(R) Gen Graphics - 23.3.5 ().
[AVHWDeviceContext @ 0x561b42b3e2c0] Driver not found in known nonstandard list, using standard behaviour.
File '/tmp/V2023CW0079_icq_1.webm' already exists. Overwrite? [y/N] y
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> vp9 (vp9_vaapi))
Press [q] to stop, [?] for help
[h264 @ 0x561b42b2d340] Reinit context to 1920x1088, pix_fmt: vaapi
[graph 0 input from stream 0:0 @ 0x561b42d4bb80] w:1920 h:1080 pixfmt:vaapi tb:1/25000 fr:25/1 sar:1/1
[vp9_vaapi @ 0x561b42a09f40] Using input frames context (format vaapi) with vp9_vaapi encoder.
[vp9_vaapi @ 0x561b42a09f40] Input surface format is nv12.
[vp9_vaapi @ 0x561b42a09f40] Using VAAPI profile VAProfileVP9Profile0 (19).
[vp9_vaapi @ 0x561b42a09f40] Using VAAPI entrypoint VAEntrypointEncSliceLP (8).
[vp9_vaapi @ 0x561b42a09f40] Using VAAPI render target format YUV420 (0x1).
[vp9_vaapi @ 0x561b42a09f40] RC mode: ICQ.
[vp9_vaapi @ 0x561b42a09f40] RC quality: 1.
[vp9_vaapi @ 0x561b42a09f40] RC framerate: 25/1 (25.00 fps).
[vp9_vaapi @ 0x561b42a09f40] Driver does not report any additional prediction constraints.
[vp9_vaapi @ 0x561b42a09f40] Using intra and P-frames (supported references: 3 / 0).
[vp9_vaapi @ 0x561b42a09f40] All wanted packed headers available (wanted 0, found 0x10).
Output #0, webm, to '/tmp/V2023CW0079_icq_1.webm':
  Metadata:
    major_brand     : mp42
    minor_version   : 0
    compatible_brands: mp42mp41
    encoder         : Lavf60.16.100
  Stream #0:0(eng): Video: vp9 (Profile 0), 1 reference frame, vaapi(tv, bt709, progressive, left), 1920x1080 (0x0) [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 1k tbn (default)
    Metadata:
      creation_time   : 2023-09-04T15:21:56.000000Z
      handler_name    : ?Mainconcept Video Media Handler
      vendor_id       : [0][0][0][0]
      encoder         : Lavc60.31.102 vp9_vaapi
Automatically inserted bitstream filter 'vp9_superframe'; args=''eed=N/A    
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x561b429b4040] EOF while reading input7.9kbits/s speed=6.75x    
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x561b429b4040] Terminating demuxer thread
[vist#0:0/h264 @ 0x561b429b6bc0] Decoder thread received EOF packet
[vist#0:0/h264 @ 0x561b429b6bc0] Decoder returned EOF, finishing
[vist#0:0/h264 @ 0x561b429b6bc0] Terminating decoder thread
No more output streams to write to, finishing.
[out#0/webm @ 0x561b429b7300] All streams finished
[out#0/webm @ 0x561b429b7300] Terminating muxer thread
[AVIOContext @ 0x561b42ac5fc0] Statistics: 4218012 bytes written, 2 seeks, 18 writeouts
[out#0/webm @ 0x561b429b7300] Output file #0 (/tmp/V2023CW0079_icq_1.webm):
[out#0/webm @ 0x561b429b7300]   Output stream #0:0 (video): 1642 frames encoded; 1642 packets muxed (4204938 bytes); 
[out#0/webm @ 0x561b429b7300]   Total: 1642 packets (4204938 bytes) muxed
[out#0/webm @ 0x561b429b7300] video:4106kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.296223%
frame= 1642 fps=169 q=-0.0 Lsize=    4119kB time=00:01:05.64 bitrate= 514.0kbits/s speed=6.76x    
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x561b429b4040] Input file #0 (/home/daniel/videos/V2023CW0079.mp4):
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x561b429b4040]   Input stream #0:0 (video): 1642 packets read (117336895 bytes); 1642 frames decoded; 0 decode errors; 
[in#0/mov,mp4,m4a,3gp,3g2,mj2 @ 0x561b429b4040]   Total: 1642 packets (117336895 bytes) demuxed
[AVIOContext @ 0x561b429bc840] Statistics: 120004418 bytes read, 0 seeks
XinfengZhang commented 5 months ago

it should related driver BRC algorithm, and from the comments in the driver issue, it is a feature request , and not fully implemented.

danielsmakandra commented 4 months ago

it should related driver BRC algorithm, and from the comments in the driver issue, it is a feature request , and not fully implemented.

Thanks for your reply, could you point me to the mentioned issue?