Recherche avancée

Sur d’autres sites (12344)

• Encode HEVC/H.265/HDR Video for YouTube from 10bit Pro-Res using FFmpeg

  30 janvier 2018, par Rodrigo Polo

  I want to have an HDR YouTube video published, my source file is either an Apple ProRes or DNxHR using a chroma subsamplig 4:4:4 or full RGB, both 10bit, so the original source file has all what is needed in order to be encoded into a 10bit 4:2:0 H.265/HEVC (HDR).

  I have followed some answers listed here, reviewed lots of different approaches, tried out many different commands without success, colors aren’t right when using only FFmpeg, to much red, when using only Adobe to encode into H.264 with the recommended settings on their support page, the results is darker, here are the commands I’ve using :

  I have tried this :

  ffmpeg \
  
  -i input.mov \
  
  -c:v libx265 \
  
  -tag:v hvc1 \
  
  -crf 21 \
  
  -preset fast \
  
  -pix_fmt yuv420p10le \
  
  -x265-params "colorprim=bt2020:transfer=smpte2084:colormatrix=bt2020nc:master-display=G(13250,34500)B(7500,3000)R(34000,16000)WP(15635,16450)L(10000000,10):max-cll=1000,400" \
  
  -c:a libfdk_aac \
  
  -b:a 128k \
  
  -ac 2 \
  
  -ar 44100 \
  
  -movflags +faststart \
  
  output.mp4

  And this :

  ffmpeg \
  
  -y \
  
  -hide_banner \
  
  -i input.mov \
  
  -pix_fmt yuv420p10le \
  
  -vf "scale=out_color_matrix=bt2020:out_h_chr_pos=0:out_v_chr_pos=0,format=yuv420p10" \
  
  -c:v libx265 \
  
  -tag:v hvc1 \
  
  -crf 21 \
  
  -preset fast \
  
  -x265-params 'crf=12:colorprim=bt2020:transfer=smpte-st-2084:colormatrix=bt2020nc:master-display="G(13250,34500)B(7500,3000)R(34000,16000)WP(15635,16450)L(10000000,1)":max-cll="1000,400"' \
  
  -c:a libfdk_aac \
  
  -b:a 128k \
  
  -ac 2 \
  
  -ar 44100 \
  
  -movflags +faststart \
  
  output.mp4

  I have also tried using MKVToolNix in order to insert the metadata into the encoded HEVC/H.265 file with the following command :

  /Applications/MKVToolNix-9.7.1.app/Contents/MacOS/mkvmerge \
  
  -o output.mkv \
  
  --colour-matrix 0:9 \
  
  --colour-range 0:1 \
  
  --colour-transfer-characteristics 0:16 \
  
  --colour-primaries 0:9 \
  
  --max-content-light 0:1000 \
  
  --max-frame-light 0:300 \
  
  --max-luminance 0:1000 \
  
  --min-luminance 0:0.01 \
  
  --chromaticity-coordinates 0:0.68,0.32,0.265,0.690,0.15,0.06 \
  
  --white-colour-coordinates 0:0.3127,0.3290 \
  
  input.mp4

  But the result is the same and YouTube don’t recognize the file as an HDR file, it does only with the first FFmpeg command and with the file encoded with Adobe Premiere, but the colors don’t look well, so, maybe I’m getting some concept wrong, thanks for your help.

• Adding a watermark to a GIF with reasonable quality using FFMPEG

  27 novembre 2017, par Neo Herakles

  so I got this code to add a logo onto a folder full of GIFs

  @echo off
  
  setlocal
  
  for %%G in ("%~dp0webm\*.gif") do (
  
      ffmpeg -i %%G -i watermark.png -filter_complex "[0:v]scale=trunc(iw/2)*2:trunc(ih/2)*2[v0];[1:v][v0]scale2ref=trunc(iw/5):trunc(ih/16)[logo][0v];[0v][logo]overlay=W-w-3:H-h-4 [v]" -map "[v]" "C:\Users\Propietario\Desktop\Conversion\gif\%%~nG.gif"
  
      )
  
  )
  
  endlocal
  
  pause

  However, the output quality is rather low, similar to an issue I had with converting from video to GIF which was solved by creating a palette.

  ffmpeg -i input.gif -filter_complex palettegen[PAL];[0:v]fifo[v0];[v0][PAL]paletteuse output.gif

  Issue is, both solutions use -filter_complex so I don’t know exactly how to combine the two in a way in which I’ll be able to turn GIFs into watermarked GIFs with reasonable quality.

• How can I programmatically write and read random video watermarks ?

  13 novembre 2017, par GreenTriangle

  I spent a few minutes trying to think of a clearer way to word my title, but I couldn’t manage it, sorry.

  I want to essentially canary trap video files : I am (hypothetically, this is not real but a personal exercise) offering them up to 5,000 different people, and if one gets leaked, I want to know who leaked it. Metadata is too easily emoved, so what I’d like to do is add a random and subtle watermark to each file, and store information about that in a database.

  For example : on Joe Smith’s copy, a 10x10 pixel 80% transparent red square in the upper left corner for 5 frames. On Diane Brown’s copy, a full-width 5-pixel 90% transparent black bar on the bottom edge for 15 frames. Then, if I find a leaked copy, I could check it against the database.

  I know this still isn’t foolproof : cropping would break co-ordinates, hue/brightness transforms would break colour reading, cutting time would break timestamps. But if I did want to do this anyway, what would be a good strategy for it ?

  My idea was to generate PNG overlays randomly, split the video into parts with mkvtoolnix/ffmpeg, re-encode the middle part with ffmpeg + overlay filter, and then rejoin them. But is this silly when there’s a "proper" way to do it ? And what would I be doing to read the watermarks, which I can’t even really conceive of ?