September 2024 - ffmpeg-user - lists.ffmpeg.org

has setpts filter been changed?
by William C Bonner 21 Mar '25

21 Mar '25

I use setpts to speed up some video that I'm concatenating from my car's dashcam. The process is consistent and hasn't changed recently. Here's an example command line: C:\Windows\ffmpeg.exe -hide_banner -i D:\ROAV\MOVIE\2024_0803_173613_229.MP4 -i D:\ROAV\MOVIE\2024_0803_173713_230.MP4 -i D:\ROAV\MOVIE\2024_0803_173813_231.MP4 -i D:\ROAV\MOVIE\2024_0803_173913_232.MP4 -i D:\ROAV\MOVIE\2024_0803_174013_233.MP4 -i D:\ROAV\MOVIE\2024_0803_174113_234.MP4 -i D:\ROAV\MOVIE\2024_0803_174213_235.MP4 -i D:\ROAV\MOVIE\2024_0803_174313_236.MP4 -i D:\ROAV\MOVIE\2024_0803_174413_237.MP4 -i D:\ROAV\MOVIE\2024_0803_174513_238.MP4 -i D:\ROAV\MOVIE\2024_0803_174613_239.MP4 -i D:\ROAV\MOVIE\2024_0803_174713_240.MP4 -i D:\ROAV\MOVIE\2024_0803_174813_241.MP4 -i D:\ROAV\MOVIE\2024_0803_174913_242.MP4 -i D:\ROAV\MOVIE\2024_0803_175013_243.MP4 -i D:\ROAV\MOVIE\2024_0803_175113_244.MP4 -i D:\ROAV\MOVIE\2024_0803_175213_245.MP4 -i D:\ROAV\MOVIE\2024_0803_175313_246.MP4 -i D:\ROAV\MOVIE\2024_0803_175413_247.MP4 -i D:\ROAV\MOVIE\2024_0803_175513_248.MP4 -i D:\ROAV\MOVIE\2024_0803_175613_249.MP4 -i D:\ROAV\MOVIE\2024_0803_175713_250.MP4 -i D:\ROAV\MOVIE\2024_0803_175813_251.MP4 -i D:\ROAV\MOVIE\2024_0803_175913_252.MP4 -i D:\ROAV\MOVIE\2024_0803_180013_253.MP4 -i D:\ROAV\MOVIE\2024_0803_180113_254.MP4 -i D:\ROAV\MOVIE\2024_0803_180213_255.MP4 -filter_complex [0:v][1:v][2:v][3:v][4:v][5:v][6:v][7:v][8:v][9:v][10:v][11:v][12:v][13:v][14:v][15:v][16:v][17:v][18:v][19:v][20:v][21:v][22:v][23:v][24:v][25:v][26:v]concat=n=27:v=1[v];[v]setpts=(1/60)*PTS,drawtext=fontfile=C\\:/Windows/Fonts/consola.ttf:fontcolor=white:fontsize=80:y=main_h-text_h-50:x=50:text=WimsWorld[o] -map [o] -map_metadata -1 -c:v libx265 -crf 23 -preset veryfast -movflags +faststart -bf 2 -g 15 -pix_fmt yuv420p -n D:\ROAV\20240803-BMW.mp4 The important part is setpts=(1/60)*PTS which should convert 1 minute of the source video to 1 second in the destination I noticed that watching recent videos strangely seemed much smoother, then realized that they were also significantly larger. https://trac.ffmpeg.org/wiki/How%20to%20speed%20up%20/%20slow%20down%20a%20… still says that using this method will drop frames, but when I ran ffprobe on the current file I got: C:\Users\Wim>ffprobe d:\ROAV\20240803-BMW.mp4 ffprobe version 2024-07-18-git-fa5a605542-full_build-www.gyan.dev Copyright (c) 2007-2024 the FFmpeg developers built with gcc 13.2.0 (Rev5, Built by MSYS2 project) configuration: --enable-gpl --enable-version3 --enable-static --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libaribcaption --enable-libdav1d --enable-libdavs2 --enable-libopenjpeg --enable-libquirc --enable-libuavs3d --enable-libxevd --enable-libzvbi --enable-libqrencode --enable-librav1e --enable-libsvtav1 --enable-libvvenc --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxeve --enable-libxvid --enable-libaom --enable-libjxl --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-libharfbuzz --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-dxva2 --enable-d3d11va --enable-d3d12va --enable-ffnvcodec --enable-libvpl --enable-nvdec --enable-nvenc --enable-vaapi --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libcodec2 --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint libavutil 59. 28.100 / 59. 28.100 libavcodec 61. 10.100 / 61. 10.100 libavformat 61. 5.101 / 61. 5.101 libavdevice 61. 2.100 / 61. 2.100 libavfilter 10. 2.102 / 10. 2.102 libswscale 8. 2.100 / 8. 2.100 libswresample 5. 2.100 / 5. 2.100 libpostproc 58. 2.100 / 58. 2.100 Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'd:\ROAV\20240803-BMW.mp4': Metadata: major_brand : isom minor_version : 512 compatible_brands: isomiso2mp41 encoder : Lavf61.5.101 Duration: 00:00:27.07, start: 0.000000, bitrate: 889896 kb/s Stream #0:0[0x1](und): Video: hevc (Main) (hev1 / 0x31766568), yuv420p(tv, progressive), 2560x1440, 889591 kb/s, 1800.04 fps, 1000k tbr, 1000k tbn (default) Metadata: handler_name : VideoHandler vendor_id : [0][0][0][0] encoder : Lavc61.10.100 libx265 Running it on an older file produced: C:\Users\Wim>ffprobe d:\ROAV\20240523-BMW.mp4 ffprobe version 2024-07-18-git-fa5a605542-full_build-www.gyan.dev Copyright (c) 2007-2024 the FFmpeg developers built with gcc 13.2.0 (Rev5, Built by MSYS2 project) configuration: --enable-gpl --enable-version3 --enable-static --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libaribcaption --enable-libdav1d --enable-libdavs2 --enable-libopenjpeg --enable-libquirc --enable-libuavs3d --enable-libxevd --enable-libzvbi --enable-libqrencode --enable-librav1e --enable-libsvtav1 --enable-libvvenc --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxeve --enable-libxvid --enable-libaom --enable-libjxl --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-libharfbuzz --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-dxva2 --enable-d3d11va --enable-d3d12va --enable-ffnvcodec --enable-libvpl --enable-nvdec --enable-nvenc --enable-vaapi --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libcodec2 --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint libavutil 59. 28.100 / 59. 28.100 libavcodec 61. 10.100 / 61. 10.100 libavformat 61. 5.101 / 61. 5.101 libavdevice 61. 2.100 / 61. 2.100 libavfilter 10. 2.102 / 10. 2.102 libswscale 8. 2.100 / 8. 2.100 libswresample 5. 2.100 / 5. 2.100 libpostproc 58. 2.100 / 58. 2.100 Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'd:\ROAV\20240523-BMW.mp4': Metadata: major_brand : isom minor_version : 512 compatible_brands: isomiso2mp41 encoder : Lavf61.3.103 Duration: 00:09:40.88, start: 0.000000, bitrate: 26675 kb/s Stream #0:0[0x1](und): Video: hevc (Main) (hev1 / 0x31766568), yuv420p(tv, progressive), 2560x1440, 26672 kb/s, 25 fps, 25 tbr, 12800 tbn (default) Metadata: handler_name : VideoHandler vendor_id : [0][0][0][0] encoder : Lavc61.5.104 libx265 The important difference is that the older file reports 2560x1440, 26672 kb/s, 25 fps, 25 tbr, 12800 tbn and the newer file reports 2560x1440, 889591 kb/s, 1800.04 fps, 1000k tbr, 1000k tbn. This seems that the new code speed up the reported fps without dropping any frames.

3 4

How to download and transcode video stream to mp4 and save the TS stream to file at the same time?
by Bo Berglund 30 Sep '24

30 Sep '24

I have created a script that downloads Internet video streams (basically news programs) and transcodes to mp4 format with a fixed windows size. As soon as the video stream recording ends the mp4 file can be played. I wonder if there is a way to let ffmpeg do two things at the same time: - download as now but save the stream to a TS formatted file and: - transcode to the mp4 format into a different output file This would make it possible to start viewing the downloaded file in TS format while the real output file remains unplayable until the download finishes and the moov atom gets written. I have found a similar question on SuperUser: https://superuser.com/questions/757114/play-ffmpeg-while-downloading But I don't understand how to adapt my ffmpeg command script to do what I want using that discussion... TIA -- Bo Berglund Developer in Sweden

5 11

Do all H264 encoders support video encryption?
by acbot＠tutanota.de 30 Sep '24

30 Sep '24

> ffmpeg -i input.mp4 -c:v copy -c:a copy -encryption_scheme cenc-aes-ctr -encryption_key 0123456789ABCDEF0123456789ABCDEF -encryption_kid 0123456789ABCDEF0123456789ABCDEF encrypted.mp4 > > ffplay -decryption_key 0123456789ABCDEF0123456789ABCDEF -i encrypted.mp4 > > The above two commands can achieve video encryption and playback, but trying to use other hardcoders (h264_qsv / h264_nvenc) cannot achieve playback. The commands are as follows: > > ffmpeg -hwaccel qsv -c:v h264_qsv -i input.mp4 -c:v h264_qsv -encryption_scheme cenc-aes-ctr -encryption_key 0123456789ABCDEF0123456789ABCDEF -encryption_kid 0123456789ABCDEF0123456789ABCDEF encrypted.mp4 > > ffplay -decryption_key 0123456789ABCDEF0123456789ABCDEF -i encrypted.mp4 > > ffplay -vcodec h264_qsv -decryption_key 0123456789ABCDEF0123456789ABCDEF -i encrypted.mp4 > > -- > 已通过Tuta 安全发送，立即注册您的加密无广告邮箱！ > https://tuta.com >

1 0

Questions about using CROP filter to pan a jpeg image
by Bram van Oosterhout 29 Sep '24

29 Sep '24

Hi, I am using ffmpeg to pan left to right over a panorama image using the crop filter. I have spent a few days experimenting and got to the solution below. The numbers are calculated externally. That was easier than retyping the the ffmpeg formulas 1. scale the image to the desired height, maintaining the aspect ratio ffmpeg -i IMG_8048.JPG -vf "scale=-1:1080" -y /tmp/On6Ub.JPG 2. the pan across the image using the crop filter with: ffmpeg -framerate 100 -loop 1 -i /tmp/On6Ub.JPG \ -vf "crop=1920:ih:960*t:0" -t 8 -y outfilecrop.mp4 Step 1 improves the speed of the video creation considerably. Step 2 uses a high-ish framerate to avoid the jitter reported on the web. Increasing the image size and then scaling back, as various post recommend as a remedy for the jitter, slows the conversion hugely. >From 10's of seconds to multiple minutes. Here are the details for IMG_8048: file IMG_8048.JPG IMG_8048.JPG: JPEG image data, Exif standard: [TIFF image data, big-endian, direntries=10, manufacturer=Apple, model=iPhone X, orientation=upper-left, xresolution=150, yresolution=158, resolutionunit=2, software=16.7.7, datetime=2024:04:11 18:46:26], baseline, precision 8, 15750x3846, components 3 This solution works for me, but I got to it by trial and error. [Q1] Is there a better way? The expression for x (third parameter in crop: 960*t ) contains a variable t. [Q2] What is t. I am unable to find it in the documentation. [Q3] Is t related to the framerate? [Q4] There are other magic numbers(?) in various web posts (like: n). Where are they documented? In step 2 I use the parameter: -t 8. The documentation suggests that this will limit the duration of the video to 8 seconds. But with the settings above the video runs for 14 seconds. [Q5] What am I missing? [Q6] General: Is there a way to search the forum archives? Answers gratefully accepted. Bram

3 5

Piped processes result in glitches/dropouts
by Anders Bjerkén 28 Sep '24

28 Sep '24

Hi! My setup is a ZOOM H5 microphone, a Raspberry Pi and a AWS hosted icecast server. I want to take a stream from the mic, encode it to mp3 on the RPI, save it to disc on the RPI, and pipe the same encoded content to the icecast server. This is the command I use: ffmpeg -f alsa -i hw:2 -filter:a "volume=1.5" -ab 192k -acodec mp3 -f mp3 - | ffmpeg -f mp3 -i - -c copy -f mpegts udp://<external_ip>:1685 -c copy -f mp3 udp://0.0.0.0:1685 Conceptually, it works. However, every now and then I get synchronized 1-2 second long glitches in the outcoming data stream for both outputs. Using only disc saving or icecast piping works fine. The encoding ffmpeg process takes appr. 25% of the CPU, the splitting ffmpeg process takes about 1%. There is plenty of CPU capacity left. Using TCP based output gives the same result. The reason why I use a stream locally is because I want to concatenate all data in one file (which I postprocess before using). Is there any way I can modify the command to get rid of the dropouts? I have used all suggested versions of this command mentioned on https://trac.ffmpeg.org/wiki/Creating%20multiple%20outputs, but with the same result in the end. Best regards Anders

3 4

How to reduce console output while downloading video stream?
by Bo Berglund 26 Sep '24

26 Sep '24

When downloading a live stream on Linux using ffmpeg I get a lot of noise on the screen like: [hls @ 0x17859d0] Skip ('#EXT-X-VERSION:3') [https @ 0x1a03eb0] Opening 'https://somesite.com/stream/bc-1727330916096.ts' for reading frame= 628 fps= 41 q=32.0 size= 1280kB time=00:00:20.99 bitrate= 499.5kbits/s dup=5 drop=0 speed=1.39xframe= 654 fps= 42 q=32.0 size= 1536kB time=00:00:21.76 bitrate= 578.3kbits/s dup=5 drop=0 speed=1.39xframe= 678 fps= 42 q=32.0 size= 1536kB time=00:00:22.59 bitrate= 557.0kbits/s dup=5 drop=0 speed= 1.4xframe= 701 fps= 42 q=32.0 size= 1536kB time=00:00:23.29 bitrate= 540.1kbits/s dup=5 drop=0 speed=1.39x[https @ 0x247a7e0] Opening 'https://somesite.com/stream/bc.m3u8' for reading How can I reduce this to only show some progress messages without all this noise? Is there a verbocity setting I can use? My download script reformats the video output on the go to a specific size such that it can be viewed directly once it is done without further processing. -- Bo Berglund Developer in Sweden

2 4

FFmpeg as a "service" with nginx
by James Greenlee 26 Sep '24

26 Sep '24

Hello all, I'm looking at trying to "distribute" some Axis cameras to various endpoints. The thought process I have on this is to startup an NginX instance, have a process(es) that connect to the Axis cameras (RTSP), and then mount those feeds to NginX RTMP points. From there I should be able to push/pull etc these feeds to where I need them. While NginX provides some of these functions (for example "exec_pull"), it seems a little inefficient to do so. First off, it only establishes a connection to the cameras when somebody connects to the RTMP mount point. I would rather have the mount point already be there. Then there's keeping the quantity of NginX workers down, and managing the ffmpeg sessions. It seems to me that starting ffmpeg outside of NginX and managing it as a system service makes sense. I would want ffmpeg to start on system startup and recover automatically if the ffmpeg instance crashes, or if there's a disruption to the connection to the cameras. I've had some success with this, but it's not as robust or reliable as I'd like. This is most definitely "skill issue" on my part as my knowledge about ffmpeg and systemd is just not deep enough. I'm open to suggestions about how to do this or some suggestions about how to get ffmpeg working properly as a service. Thanks!! James

1 0

Reference for using ffmpeg for streaming HLS pusposes
by panji dia 26 Sep '24

26 Sep '24

Dear community, Thank you for letting me join this mailing list. So I have a question regarding live streaming with HLS. For now I just convert my media to m3u8 and call it from somewhere else. But is there any proper way or convenient way to do it? Or any of you guys got experience with it and ship it to production? Or any reference that i can use for developing such a service? Thank you Best Regards, Baskoro

1 0

FFmpeg 6.1.2 about vulkan h264 video decode
by Wonskuis Yu(SH-RD) 26 Sep '24

26 Sep '24

Hi�� When I use ffmpeg 6.1.2 to do transcode as flow cmd: ffmpeg.exe �Chwaccel vulkan �CI H264-4k.mp4 output.mp4 The assert of null pointer occur, after debug source code, I found the pointer of layered_frame is null, because it doesn��t created. Further, I found it only be created when dec->dedicated_dpb && dec->layered_dpb are all true. But the driver report caps only include VK_VIDEO_DECODE_CAPABILITY_DPB_AND_OUTPUT_COINCIDE_BIT_KHR, not include VK_VIDEO_CAPABILITY_SEPARATE_REFERENCE_IMAGES_BIT_KHR, it leads the layered_frame can��t be created. For this combination of driver caps, I think it means the reference frame and output use the same image resource , the ffmpeg should create layered_frame as array? I��m confused whether It��s the driver issue or not? Thanks, WonskuisYu Thanks, WonskuisYu �� ʼ��б��ܻ�ר��Ϣ��ָ��ռ��ʹ�á��Ͻ��Ա��ʼ��κ�δ��Ȩ�Ĳ��ġ�ʹ�á��ƻ�ת�� CONFIDENTIAL NOTE: This email contains confidential or legally privileged information and is for the sole use of its intended recipient. Any unauthorized review, use, copying or forwarding of this email or the content of this email is strictly prohibited.

1 0

webvtt HLS discontinuity mismatch
by Jonathan Baecker 25 Sep '24

25 Sep '24

Hello everybody, I would like to use this hls_flags "append_list+delete_segments+omit_endlist", but when I use them in combination with webvtt subtitle, it generate different playlists entries. In stream.m3u8 I get a "#EXT-X-DISCONTINUITY" flag before the first segment, but im stream_vtt.m3u8 this flag is missing. So in my html video player (videojs 8) I can see the subtitle menu entry, but no subtitle in video. If I add that flag also to the stream_vtt.m3u8 playlist, the subtitle will show in player. When no playlist contains that flag, then also. My full ffmpeg command is: ffmpeg -i video.mp4 -i dummy.vtt -map "0:v" -map "0:a" -map "1:s?" -c:v libx264 -crf 23 -x264-params keyint=50:min-keyint=25:scenecut=-1 -maxrate 1300k -bufsize 2600k -preset faster -tune zerolatency -profile:v Main -level 3.1 -c:a aac -ar 44100 -b:a 128k -flags +cgop -f hls -hls_time 6 -hls_list_size 60 -hls_flags append_list+delete_segments+omit_endlist -var_stream_map "v:0,a:0,s:0,sgroup:subtitle,name:Deutsch,language:DE,default:YES" -master_pl_name master.m3u8 -hls_segment_filename live/stream-%d.ts live/stream.m3u8 When I remove hls_flags from ffmpeg command, then "#EXT-X-DISCONTINUITY" flag disappears from both playlists to, but in my use case I need hls_flags. Is there a way to fix that? Jonathan

1 0