Recherche avancée

Sur d’autres sites (7560)

• arm : vp9itxfm : Skip empty slices in the first pass of idct_idct 16x16 and 32x32

  9 janvier 2017, par Martin Storsjö

  arm : vp9itxfm : Skip empty slices in the first pass of idct_idct 16x16 and 32x32
  This work is sponsored by, and copyright, Google.
  Previously all subpartitions except the eob=1 (DC) case ran with
  
  the same runtime :
                                       Cortex A7       A8       A9      A53
  
  vp9_inv_dct_dct_16x16_sub16_add_neon :   3188.1   2435.4   2499.0   1969.0
  
  vp9_inv_dct_dct_32x32_sub32_add_neon :  18531.7  16582.3  14207.6  12000.3
  By skipping individual 4x16 or 4x32 pixel slices in the first pass,
  
  we reduce the runtime of these functions like this :
  vp9_inv_dct_dct_16x16_sub1_add_neon :     274.6    189.5    211.7    235.8
  
  vp9_inv_dct_dct_16x16_sub2_add_neon :    2064.0   1534.8   1719.4   1248.7
  
  vp9_inv_dct_dct_16x16_sub4_add_neon :    2135.0   1477.2   1736.3   1249.5
  
  vp9_inv_dct_dct_16x16_sub8_add_neon :    2446.7   1828.7   1993.6   1494.7
  
  vp9_inv_dct_dct_16x16_sub12_add_neon :   2832.4   2118.3   2266.5   1735.1
  
  vp9_inv_dct_dct_16x16_sub16_add_neon :   3211.7   2475.3   2523.5   1983.1
  
  vp9_inv_dct_dct_32x32_sub1_add_neon :     756.2    456.7    862.0    553.9
  
  vp9_inv_dct_dct_32x32_sub2_add_neon :   10682.2   8190.4   8539.2   6762.5
  
  vp9_inv_dct_dct_32x32_sub4_add_neon :   10813.5   8014.9   8518.3   6762.8
  
  vp9_inv_dct_dct_32x32_sub8_add_neon :   11859.6   9313.0   9347.4   7514.5
  
  vp9_inv_dct_dct_32x32_sub12_add_neon :  12946.6  10752.4  10192.2   8280.2
  
  vp9_inv_dct_dct_32x32_sub16_add_neon :  14074.6  11946.5  11001.4   9008.6
  
  vp9_inv_dct_dct_32x32_sub20_add_neon :  15269.9  13662.7  11816.1   9762.6
  
  vp9_inv_dct_dct_32x32_sub24_add_neon :  16327.9  14940.1  12626.7  10516.0
  
  vp9_inv_dct_dct_32x32_sub28_add_neon :  17462.7  15776.1  13446.2  11264.7
  
  vp9_inv_dct_dct_32x32_sub32_add_neon :  18575.5  17157.0  14249.3  12015.1
  I.e. in general a very minor overhead for the full subpartition case due
  
  to the additional loads and cmps, but a significant speedup for the cases
  
  when we only need to process a small part of the actual input data.
  In common VP9 content in a few inspected clips, 70-90% of the non-dc-only
  
  16x16 and 32x32 IDCTs only have nonzero coefficients in the upper left
  
  8x8 or 16x16 subpartitions respectively.
  This is cherrypicked from libav commit
  
  9c8bc74c2b40537b0997f646c87c008042d788c2.
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavcodec/arm/vp9itxfm_neon.S
  • [DH] tests/checkasm/vp9dsp.c

• ffmpeg API : handle frame loss in hevc encoding

  8 janvier 2024, par Mario

  Everything works fine until the introduction of frame->pts increment due to frame loss.

  


  Below is the regular progression without frame->pts increments :

  


  


  frame->pts=8 pkt->pts=512 pkt->dts=-512 pkt->flags=1
  
frame->pts=9 pkt->pts=2560 pkt->dts=0 pkt->flags=0
  
frame->pts=10 pkt->pts=1536 pkt->dts=512 pkt->flags=0
  
frame->pts=11 pkt->pts=1024 pkt->dts=1024 pkt->flags=0
  
frame->pts=12 pkt->pts=2048 pkt->dts=1536 pkt->flags=0
  
frame->pts=13 pkt->pts=4608 pkt->dts=2048 pkt->flags=0
  
frame->pts=14 pkt->pts=3584 pkt->dts=2560 pkt->flags=0
  
frame->pts=15 pkt->pts=3072 pkt->dts=3072 pkt->flags=0
  
frame->pts=16 pkt->pts=4096 pkt->dts=3584 pkt->flags=0
  
frame->pts=17 pkt->pts=6656 pkt->dts=4096 pkt->flags=0
  
frame->pts=18 pkt->pts=5632 pkt->dts=4608 pkt->flags=0

  


  


  When I introduce the frame->pts increment it happens :

  


  


  frame->pts=15 pkt->pts=512 pkt->dts=-512 pkt->flags=1
  
frame->pts=17 pkt->pts=4608 pkt->dts=2048 pkt->flags=0
  
frame->pts=19 pkt->pts=2560 pkt->dts=1536 pkt->flags=0
  
[mp4 @ 0x7eff842222c0] Application provided invalid, non monotonically increasing dts to muxer in stream 0 : 2048 >= 1536

  


  


  So I wrote the following code as a "quick" solution (between av_packet_rescale_ts() and av_interleaved_write_frame()) :

  


     av_packet_rescale_ts(pkt, c->time_base, st->time_base);  
   ...
   if (pkt->dts<=previous_dts)  
     {  
      if (pkt->pts<=previous_pts)  
        {  
         pkt->pts=previous_dts+1+pkt->pts-pkt->dts;  
        }  
       pkt->dts=previous_dts+1;  
     }  
   previous_dts=pkt->dts;  
   previous_pts=pkt->pts;  
   ...
   ret = av_interleaved_write_frame(fmt_ctx, pkt);  


  


  Now I no longer have the error, but the values are :

  


  


  frame->pts=15 pkt->pts=512 pkt->dts=-512 pkt->flags=1
  
changed frame->pts=15 pkt->pts=512 pkt->dts=1 pkt->flags=1
  
frame->pts=17 pkt->pts=4608 pkt->dts=2048 pkt->flags=0
  
frame->pts=19 pkt->pts=2560 pkt->dts=1536 pkt->flags=0
  
changed frame->pts=19 pkt->pts=3073 pkt->dts=2049 pkt->flags=0
  
frame->pts=21 pkt->pts=1536 pkt->dts=1536 pkt->flags=0
  
changed frame->pts=21 pkt->pts=2050 pkt->dts=2050 pkt->flags=0
  
frame->pts=23 pkt->pts=4096 pkt->dts=3584 pkt->flags=0
  
frame->pts=25 pkt->pts=8704 pkt->dts=6144 pkt->flags=0
  
frame->pts=27 pkt->pts=6656 pkt->dts=5632 pkt->flags=0
  
changed frame->pts=27 pkt->pts=7169 pkt->dts=6145 pkt->flags=0
  
frame->pts=29 pkt->pts=5632 pkt->dts=5632 pkt->flags=0
  
changed frame->pts=29 pkt->pts=6146 pkt->dts=6146 pkt->flags=0
  
frame->pts=31 pkt->pts=7680 pkt->dts=7168 pkt->flags=0
  
frame->pts=33 pkt->pts=12800 pkt->dts=10240 pkt->flags=0
  
frame->pts=35 pkt->pts=10752 pkt->dts=9728 pkt->flags=0
  
changed frame->pts=35 pkt->pts=11265 pkt->dts=10241 pkt->flags=0
  
frame->pts=37 pkt->pts=9728 pkt->dts=9728 pkt->flags=0
  
changed frame->pts=37 pkt->pts=10242 pkt->dts=10242 pkt->flags=0

  


  


  What is the correct way to handle frame loss scenario ?
Is there a way to inform the encoder about frame loss ?

  


  The encoder is "hevc_qsv" and the output format is mov (.mp4).

  


• FFMPEG error when saving NDI stream to mp4

  22 septembre 2020, par user1163234

  I am trying to record a NDI stream to a MP4 file(I want to stream the mp4 to rtmp endpoint after saving file). However I am getting this error when running this class. https://github.com/WalkerKnapp/devolay/blob/master/examples/src/main/java/com/walker/devolayexamples/recording/RecordingExample.java

  


  Error :

  


  Connecting to source: DESKTOP-GQNH46Q (Ari PC output)
[file @ 0x7fb4f2d48540] Setting default whitelist 'file,crypto'
x265 [info]: HEVC encoder version 0.0
x265 [info]: build info [Mac OS X][clang 8.1.0][64 bit] 8bit+10bit+12bit
x265 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2
x265 [info]: Main profile, Level-3.1 (Main tier)
x265 [info]: Thread pool created using 8 threads
x265 [info]: Slices                              : 1
x265 [info]: frame threads / pool features       : 1 / wpp(12 rows)
x265 [info]: Coding QT: max CU size, min CU size : 64 / 8
x265 [info]: Residual QT: max TU size, max depth : 32 / 1 inter / 1 intra
x265 [info]: ME / range / subpel / merge         : hex / 57 / 2 / 3
x265 [info]: Keyframe min / max / scenecut / bias: 1 / 250 / 40 / 5.00
x265 [info]: Lookahead / bframes / badapt        : 20 / 4 / 2
x265 [info]: b-pyramid / weightp / weightb       : 1 / 1 / 0
x265 [info]: References / ref-limit  cu / depth  : 3 / off / on
x265 [info]: AQ: mode / str / qg-size / cu-tree  : 2 / 1.0 / 32 / 1
x265 [info]: Rate Control / qCompress            : CRF-28.0 / 0.60
x265 [info]: tools: rd=3 psy-rd=2.00 early-skip rskip signhide tmvp b-intra
x265 [info]: tools: strong-intra-smoothing lslices=4 deblock sao
[SWR @ 0x7fb4f8893000] Using fltp internally between filters
[mp4 @ 0x7fb4f3820200] Application provided invalid, non monotonically increasing dts to muxer in stream 0: 135 >= 107
Failed to write video flush packet, skipping: Invalid argument
configurationVersion:                1
general_profile_space:               0
general_tier_flag:                   0
general_profile_idc:                 1
general_profile_compatibility_flags: 0x60000000
general_constraint_indicator_flags:  0x900000000000
general_level_idc:                   93
min_spatial_segmentation_idc:        0
parallelismType:                     0
chromaFormat:                        1
bitDepthLumaMinus8:                  0
bitDepthChromaMinus8:                0
avgFrameRate:                        0
constantFrameRate:                   0
numTemporalLayers:                   1
temporalIdNested:                    1
lengthSizeMinusOne:                  3
numOfArrays:                         4
array_completeness[0]:               0
NAL_unit_type[0]:                    32
numNalus[0]:                         1
nalUnitLength[0][0]:                 24
array_completeness[1]:               0
NAL_unit_type[1]:                    33
numNalus[1]:                         1
nalUnitLength[1][0]:                 41
array_completeness[2]:               0
NAL_unit_type[2]:                    34
numNalus[2]:                         1
nalUnitLength[2][0]:                 7
array_completeness[3]:               0
NAL_unit_type[3]:                    39
numNalus[3]:                         1
nalUnitLength[3][0]:                 2050
[AVIOContext @ 0x7fb4f2d48640] Statistics: 2 seeks, 4 writeouts
x265 [info]: frame I:      1, Avg QP:14.03  kb/s: 16.33   
x265 [info]: frame P:     36, Avg QP:21.67  kb/s: 0.04    
x265 [info]: frame B:     73, Avg QP:24.22  kb/s: 0.03    
x265 [info]: Weighted P-Frames: Y:0.0% UV:0.0%
x265 [info]: consecutive B-frames: 28.9% 13.2% 18.4% 2.6% 36.8% 

encoded 110 frames in 15.13s (7.27 fps), 0.18 kb/s, Avg QP:23.29
[aac @ 0x7fb4f6aa6200] Qavg: 65536.000