Recherche avancée

Sur d’autres sites (7802)

• Video from images to mp4 in nvidia GPU

  16 août 2019, par M.y

  I am trying to encode a h264 .mp4 video created from .jpg images using a 1070ti nvidia cuda power, having a a crossfade transition between each image.
  I am able to render the video in GPU using the flags -c:v h264_nvenc, I see a short peak in the GPU encoding, but with a long period of computer CPU hight load, I guess preparing the transitioning images. But the image preparation it happens on cpu/ram due the -filter_complex and is quite slow.
  This works :

  ffmpeg.exe, -y, -loop, 1, -t, 2.5, -i, 1565957420594_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565957453659_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565957487743_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565957525280_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565957587308_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565957644898_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565957859119_labeled.jpg, -loop, 1, -t, 2.5, -i,1565959133561_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565959412948_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565959501884_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565959755432_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565959882380_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565960023185_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565960157174_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565960683303_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565961151548_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565961230278_labeled.jpg, -loop, 1, -t, 2.5, -i, 1565961671766_labeled.jpg, -loop, 1, -t, 2.5, -i, final.jpg, -loop, 1, -t, 2.5, -i, final.jpg, -c:v, h264_nvenc, -preset, fast, -filter_complex, [1]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+0.5/TB[f0];[2]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+1.0/TB[f1];[3]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+1.5/TB[f2];[4]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+2.0/TB[f3];[5]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+2.5/TB[f4];[6]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+3.0/TB[f5];[7]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+3.5/TB[f6];[8]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+4.0/TB[f7];[9]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+4.5/TB[f8];[10]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+5.0/TB[f9];[11]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+5.5/TB[f10];[12]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+6.0/TB[f11];[13]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+6.5/TB[f12];[14]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+7.0/TB[f13];[15]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+7.5/TB[f14];[16]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+8.0/TB[f15];[17]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+8.5/TB[f16];[18]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+9.0/TB[f17];[19]fade=d=0.5:t=in:alpha=1,setpts=PTS-STARTPTS+9.5/TB[f18];[0][f0]overlay[bg1];[bg1][f1]overlay[bg2];[bg2][f2]overlay[bg3];[bg3][f3]overlay[bg4];[bg4][f4]overlay[bg5];[bg5][f5]overlay[bg6];[bg6][f6]overlay[bg7];[bg7][f7]overlay[bg8];[bg8][f8]overlay[bg9];[bg9][f9]overlay[bg10];[bg10][f10]overlay[bg11];[bg11][f11]overlay[bg12];[bg12][f12]overlay[bg13];[bg13][f13]overlay[bg14];[bg14][f14]overlay[bg15];[bg15][f15]overlay[bg16];[bg16][f16]overlay[bg17];[bg17][f17]overlay[bg18];[bg18][f18]overlay[v], -map, [v], -movflags, +faststart, output.mp4

  I am trying to do all work in the GPU, theoretically I can encode all images in GPU memory using in each -i the flags "-hwaccel cuvid -c:v mjpeg_cuvid" I receive the following error :

  [mjpeg_cuvid @ 00000000024ef980] ignoring invalid SAR: 0/0
  
  Impossible to convert between the formats supported by the filter 'graph 0 input from stream 1:0' and the filter 'auto_scaler_0'
  
  Error reinitializing filters!
  
  Failed to inject frame into filter network: Function not implemented
  
  Error while processing the decoded data for stream #0:0

  Is there a way to load images in the GPU with the "fade" flag applied ?

  Thanks in advance !

• aacenc_pred : rework the way prediction is done

  29 août 2015, par Rostislav Pehlivanov

  aacenc_pred : rework the way prediction is done
  This commit completely alters the algorithm of prediction.
  
  The original commit which introduced prediction was completely
  
  incorrect to even remotely care about what the actual coefficients
  
  contain or whether any options were enabled. Not my actual fault.
  This commit treats prediction the way the decoder does and expects
  
  to do : like lossy encryption. Everything related to prediction now
  
  happens at the very end but just before quantization and encoding
  
  of coefficients. On the decoder side, prediction happens before
  
  anything has had a chance to even access the coefficients.
  Also the original implementation had problems because it actually
  
  touched the band_type of special bands which already had their
  
  scalefactor indices marked and it’s a wonder the asserion wasn’t
  
  triggered when transmitting those.
  Overall, this now drastically increases audio quality and you should
  
  think about enabling it if you don’t plan on playing anything encoded
  
  on really old low power ultra-embedded devices since they might not
  
  support decoding of prediction or AAC-Main. Though the specifications
  
  were written ages ago and as times change so do the FLOPS.
  Signed-off-by : Rostislav Pehlivanov <atomnuker@gmail.com>
  

  • [DH] libavcodec/aac.h
  • [DH] libavcodec/aaccoder.c
  • [DH] libavcodec/aacenc.c
  • [DH] libavcodec/aacenc.h
  • [DH] libavcodec/aacenc_pred.c
  • [DH] libavcodec/aacenc_pred.h

• NVidia driver libraries in nvidia/cuda image

  16 février 2018, par Islam Sabyrgaliyev

  I want to run ffmpeg with cuvid hw-accelerated decoding in the container based on official nvidia/cuda image. Ffmpeg is not able to find libnvcuvid.so, although there are all required cuda libs.
  The output of ldconfig -p | grep libnv from the container :

  libnvrtc.so (libc6,x86-64) => /usr/local/cuda-9.0/targets/x86_64-linux/lib/libnvrtc.so
  
  libnvrtc-builtins.so (libc6,x86-64) => /usr/local/cuda-9.0/targets/x86_64-linux/lib/libnvrtc-builtins.so
  
  libnvidia-ptxjitcompiler.so.1 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libnvidia-ptxjitcompiler.so.1
  
  libnvidia-opencl.so.1 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libnvidia-opencl.so.1
  
  libnvidia-ml.so.1 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
  
  libnvidia-fatbinaryloader.so.390.12 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libnvidia-fatbinaryloader.so.390.12
  
  libnvidia-compiler.so.390.12 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libnvidia-compiler.so.390.12
  
  libnvidia-cfg.so.1 (libc6,x86-64) => /usr/lib/x86_64-linux-gnu/libnvidia-cfg.so.1
  
  libnvgraph.so (libc6,x86-64) => /usr/local/cuda-9.0/targets/x86_64-linux/lib/libnvgraph.so
  
  libnvblas.so (libc6,x86-64) => /usr/local/cuda-9.0/targets/x86_64-linux/lib/libnvblas.so
  
  libnvToolsExt.so (libc6,x86-64) => /usr/local/cuda-9.0/targets/x86_64-linux/lib/libnvToolsExt.so

  Should I just copy libnvcuvid.so from the host ? Wouldn’t it break if underlying driver version changes ?