Recherche avancée

Sur d’autres sites (9774)

• ffmpeg crashes with "Internal bug", "Failed to inject frame into filter network" while trying to convert PNG files into an animated gif using filters

  12 août 2022, par EigentlichOracler

  ffmpeg version : ffmpeg version 4.2.7-0ubuntu0.1

  


  Kernel : 5.4.0-122-lowlatency #138-Ubuntu SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux

  


  PNG-Images : 996 pieces, all equal in size : 1000x50

  


  Then trying to build palette using palettegen :

  


  export filters="scale=1000:-1:flags=lanczos"
export palette=./stripes_palette.png
ffmpeg -v warning -i ./palette_source_stripes.png -vf "$filters,palettegen=stats_mode=diff" -y $palette
[Parsed_palettegen_1 @ 0x55f0eea30e00] Dupped color: FF63000D


  


  Palette file has been created, looks good so far. Then I tried to convert all 996 PNG files into one single (space optimized) animated GIF file using paletteuse :

  


  ffmpeg -v warning -framerate 5 -thread_queue_size 2048 -i ./tmp/temp.%04d.png -i $palette -lavfi "$filters,paletteuse=dither=bayer:bayer_scale=5:diff_mode=rectangle" -r 5 -loop -1 -y testanim.gif
Error while filtering: Internal bug, should not have happened
Failed to inject frame into filter network: Internal bug, should not have happened
Error while processing the decoded data for stream #0:0


  


  The reason for "-framerate 5" for the input and "-r 5" for the output is : I wanted the animated GIF to use a determined time for running through animation. It already had worked out without using complex filtering, but the results were huge GIF files which are hard to handle.
But even when I avoid both framerate parameters, still the same error message occurs.

  


  Should I look deeper into the "Dupped color" which has been mentioned ? Did anybody encounter a similar issue using ffmpeg ?
I'm not used to deal with filters (nor "complex filters") in ffmpeg, I'm very new at this, but I've found no tips regarding an ffmpeg crash like I've right here.

  


  ------------------------------- EDIT 2022-08-12/1 ---------------------

  


  Same error occurs with current stable version
ffmpeg version 4.4.2-0ubuntu0.22.04.1 Copyright (c) 2000-2021 the FFmpeg developers

  


  ------------------------------- EDIT 2022-08-12/2 ---------------------

  


  Found workaround, see answer below

  


• Generate animated padded thumbnail using ffmpeg

  15 juillet 2022, par kregus

  we've been generating animated thumbnails with ffmpeg without a problem using this command :

  


  ffmpeg -y -ss 5 -t 3 -i "/input.mp4" -i /playbtn.png -r 10 -filter_complex "[0]fps=10,scale=1280:-1:flags=lanczos[pb];[pb][1]overlay=(W-w)/2:(H-h)/2, split[s0][s1];[s0]palettegen=max_colors=32[p];[s1][p]paletteuse=dither=bayer" /output.gif


  


  However, now we want to force each thumbnail to be 1280x720 and pad the rest black.

  


  This works well for regular thumbnails with this command :

  


  ffmpeg -y -ss 5 -t 3 -i "/input.mp4" -vframes 1 -filter:v "scale=w=1280:h=720:force_original_aspect_ratio=1,pad=1280:720:(ow-iw)/2:(oh-ih)/2" /output.jpg


  


  Now I tried dozens of ways to try and integrate the force_original_aspect_ratio and pad filters into the animated thumbnail generation code, but am getting a host of errors, like :

  


  Cannot find a matching stream for unlabeled input pad 0 on filter Parsed_split_4

  


  When I tried this command :

  


  ffmpeg -y -ss 5 -t 3 -i "/Users/nick/Downloads/intro_portrait_nl_1650909779_rendereddd.mp4" -i /Applications/Ampps/www/vormats_aws/web-render/functions/files/playbtn.png -r 10 -filter_complex "[0]fps=10,scale=1280:720:flags=lanczos:force_original_aspect_ratio=1,pad=1280:720:(ow-iw)/2:(oh-ih)/2[pb];[pb][1]overlay=(W-w)/2:(H-h)/2;split[s0][s1];[s0]palettegen=max_colors=32[p];[s1][p]paletteuse=dither=bayer" /Users/nick/Downloads/smallLandscapeee.gif


  


  ... You get the point, I have no idea how to adapt our original animated thumbnail generation code to force a 1280x720 resolution with padding.

  


  Any input is appreciated, thanks !

  


• What exactly is the input to the coder part of a software video codec ?

  30 juin 2022, par usrnew xnew

  I am doing a project based on video codecs and have to implement a specific type of codec(one of the H.26x types). So far in my research, I've found out that mp4 files are simply container files that can contain video data in any coded form. I wanted to know if there is a raw file type for video files like how there are for images and audio. However, I came to find out that there is no standard raw file for a video format and since mp4 is a container, it may as well contain a completely uncompressed video with frames, audio etc.

  


  So basically, if there's no specific standard raw format for a video, what kind of file would the software codec take as input while encoding ? And if it does take an mp4, how do we find out if it's really an uncompressed mp4 file or an already encoded one, i.e. which atom holds the value of encoding name ?

  


  Edit :