ffmpeg video to jpg frames poor quality - ffmpeg

The quality of JPGs extracted by ffmpeg from an mp4 is much poorer than pause frame from video player (vlc). I am looking for ffmpeg cmd option to improve output quality.
Using following cmd :
/home/tools/bin/ffmpeg -i Merkurtransit_20191111_crf20_8fps_crop.mp4 Merkurtransit_20191111_crf20_8fps_crop_%04d.jpg -hide_banner
The ffmpeg cmd is from instructions found here :
https://www.bugcodemaster.com/article/extract-images-frame-frame-video-file-using-ffmpeg
A comparing screen copy is here:
http://skywatcher.space/download/vlc_player_vs_ffmpeg_bug.png
A few items of note. I created the mp4 myself from high res png (actually originally from 16bit tiff) using ffmpeg :
/home/tools/bin/ffmpeg -framerate 8.0 -i ./AS_P10_RS6_png_reg/Merkurtransit_20191111_%03d.png -vf "crop=760:560:20:40" -pix_fmt yuv420p -crf 20 -r 24 -y ./Video/Merkurtransit_20191111_crf20_8fps_crop.mp4
The crf 20 is pretty high quality, close to 100% and the recovered frame should be close to original. The video player pause frame shows adequate quality. (though I can't say if it is on a key frame or not)
ffmpeg version info:
home/tools/bin/ffmpeg -v
ffmpeg version N-80251-g0c7fa15 Copyright (c) 2000-2016 the FFmpeg developers
built with gcc 4.8 (Ubuntu 4.8.4-2ubuntu1~14.04.3)
configuration: --prefix=/home/tools/ffmpeg_build --pkg-config-flags=--static --extra-cflags=-I/home/tools/ffmpeg_build/include --extra-ldflags=-L/home/tools/ffmpeg_build/lib --bindir=/home/tools/bin --enable-gpl --enable-libass --enable-libfdk-aac --enable-libfreetype --enable-libmp3lame --enable-libopus --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libx265 --enable-nonfree
libavutil 55. 24.100 / 55. 24.100
libavcodec 57. 45.100 / 57. 45.100
libavformat 57. 37.101 / 57. 37.101
libavdevice 57. 0.101 / 57. 0.101
libavfilter 6. 46.101 / 6. 46.101
libswscale 4. 1.100 / 4. 1.100
libswresample 2. 0.101 / 2. 0.101
libpostproc 54. 0.100 / 54. 0.100

FFmpeg is primarily a video convertor and JPEG output is the result of a MJPEG encoder generating a single image. When no rate control parameters are set, a default bitrate of 200 kbps is selected.
For a better quality output, use
ffmpeg -i in.mp4 -q:v 1 -qmin 1 -qmax 1 out%d.jpg
The quantizer is clamped to exactly 1.

Related

Converting DTS to AAC with ffmpeg shifts spoken audio to the right

I am using ffmpeg to convert mkv movies to mp4, like so:
$ ffmpeg -i source.mkv -c:v copy -c:a aac destination.mp4
By doing this, the audio stream gets converted from the original DTS:
Stream #0:1: Audio: dts (DTS), 48000 Hz, 5.1(side), fltp, 1536 kb/s (default)
To AAC:
Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, 5.1, fltp, 394 kb/s (default)
The resulting file plays just fine, except that the spoken audio (what I am assuming would be sent to the central channel in a 5.1 configuration) sounds clearly shifted to the right, when listening through the MacBook's built-in speakers or my stereo headphones. Note that music and other sound effects appear unaffected, properly balanced. Also note that I have been able to reproduce this behavior with a variety of source files.
Here's ffmpeg's banner:
ffmpeg version 4.0.2 Copyright (c) 2000-2018 the FFmpeg developers
built with Apple LLVM version 9.1.0 (clang-902.0.39.2)
configuration: --prefix=/usr/local/Cellar/ffmpeg/4.0.2 --enable-shared --enable-pthreads --enable-version3 --enable-hardcoded-tables --enable-avresample --cc=clang --host-cflags= --host-ldflags= --enable-gpl --enable-libmp3lame --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libx265 --enable-libxvid --enable-opencl --enable-videotoolbox --disable-lzma
libavutil 56. 14.100 / 56. 14.100
libavcodec 58. 18.100 / 58. 18.100
libavformat 58. 12.100 / 58. 12.100
libavdevice 58. 3.100 / 58. 3.100
libavfilter 7. 16.100 / 7. 16.100
libavresample 4. 0. 0 / 4. 0. 0
libswscale 5. 1.100 / 5. 1.100
libswresample 3. 1.100 / 3. 1.100
libpostproc 55. 1.100 / 55. 1.100
An FFmpeg commit, aacenc: support extended channel layouts using PCEs, in Nov 2017 added support for many more channel layouts than specified in the MPEG standard for AAC. Unfortunately, it seems it has broken encoding for layouts that worked fine before.
The 3.4 release series is the last before the said commit and should be used for multichannel AAC encoding if you encounter an error with more recent builds.
There is an open bug report at https://trac.ffmpeg.org/ticket/7273. You may post a comment in there to showcase your example.

How to keep transparency when converting .avi to .webm with ffmpeg [duplicate]

I am trying to convert a mov with alpha transparency to webm with alpha transparency, as seen here. I followed the steps explained here to no avail.
From this answer I was able to remove all the black in the video, thus making it transparent but this is not what I need as I already have a transparent mov and would like to convert that to transparent webm format.
ffmpeg -i input.mp4 -c:v libvpx -vf "colorkey=0x000000:0.1:0.1,format=yuva420p" out.webm
This is the ffprobe output of the video I would like to convert to webm with transparency.
built with Apple LLVM version 7.0.2 (clang-700.1.81)
configuration: --prefix=/usr/local/Cellar/ffmpeg/2.8.4 --enable-shared --enable-pthreads --enable-gpl --enable-version3 --enable-hardcoded-tables --enable-avresample --cc=clang --host-cflags= --host-ldflags= --enable-opencl --enable-libx264 --enable-libmp3lame --enable-libvo-aacenc --enable-libxvid --enable-libfreetype --enable-libtheora --enable-libvorbis --enable-libvpx --enable-librtmp --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libass --enable-ffplay --enable-libspeex --enable-libschroedinger --enable-libfdk-aac --enable-libopus --enable-frei0r --enable-libopenjpeg --disable-decoder=jpeg2000 --extra-cflags=-I/usr/local/Cellar/openjpeg/1.5.2_1/include/openjpeg-1.5 --enable-nonfree --enable-vda
libavutil 54. 31.100 / 54. 31.100
libavcodec 56. 60.100 / 56. 60.100
libavformat 56. 40.101 / 56. 40.101
libavdevice 56. 4.100 / 56. 4.100
libavfilter 5. 40.101 / 5. 40.101
libavresample 2. 1. 0 / 2. 1. 0
libswscale 3. 1.101 / 3. 1.101
libswresample 1. 2.101 / 1. 2.101
libpostproc 53. 3.100 / 53. 3.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'input.mov':
Metadata:
major_brand : qt
minor_version : 0
compatible_brands: qt
creation_time : 2016-01-17 16:04:07
encoder : Mac OS X v? (AVF 1046.9.1, CM 1731.15.20, x86_64)
encoder-eng : Mac OS X v? (AVF 1046.9.1, CM 1731.15.20, x86_64)
Duration: 00:00:06.63, start: 0.000000, bitrate: 63966 kb/s
Stream #0:0(eng): Video: prores (ap4h / 0x68347061), yuva444p10le(bt470bg/smpte240m/bt709), 1920x1080, 63963 kb/s, 25.03 fps, 25 tbr, 600 tbn, 600 tbc (default)
Metadata:
creation_time : 2016-01-17 16:04:07
handler_name : Core Media Data Handler
encoder : Apple ProRes 4444
I've also tried the following command which didn't work for me either.
ffmpeg -y -i input.mov -c:v libvpx-vp9 -b:v 2000k -pass 1 -an -f webm output.webm
I'm using version 2.8.4 of ffmpeg on a Mac, installed with brew. 2.8.5 is the latest version.
Since 2016-07-13, it's possible to encode VP9/webm videos with alpha channel (VP9a). However, the command you use here will create a VP8a video. Assuming you got a copy of ffmpeg compiled after that date, all you need is change the libvpx to libvpx-vp9. You don't need the yuva420p conversion either (is selected by default).
Try
ffmpeg -i input.mov -c:v libvpx -pix_fmt yuva420p out.webm
All of the other solutions resulted in a video of subpar quality. Please ensure that you define the bitrate to your liking. I changed mine from 1M to 2M and was satisfied.
ffmpeg -i "Model 1 V1.mov" -f webm -c:v libvpx -b:v 2M -acodec libvorbis -auto-alt-ref 0 model1v3.webm -hide_banner
If you're using After Effects or Premiere Pro, there's also this plugin, which can be used in Adobe Media Encoder.
WebM-alpha is only defined for VP8. It doesn't work at all for VP9 right now.
This is the command that worked for me for VP8.
ffmpeg -i input.mov -c:v libvpx -pix_fmt yuva420p -b:v 2M -auto-alt-ref 0 output.webm
Accepted option did not work.

Set fake duration on h264 video

I have a script that fetches multiple chunks of a video from one place and streams it as a single video to other place (to Kodi player).
Everything seems to work fine except one thing that bothers me, the player doesn't seem to know how long the video is, therefore the total duration increments as the video plays.
I do know the duration of the video from an xml file that contains a link to all the chunks, but I don't know how to write that in the metadata of the first chunk.
The video codec is h264, but I'm not sure if it is wrapped in some container, like mp4.
Here is the ffmpeg -i output for the first chunk:
ffmpeg version 3.1.5 Copyright (c) 2000-2016 the FFmpeg developers
built with Apple LLVM version 8.0.0 (clang-800.0.38)
configuration: --prefix=/usr/local/Cellar/ffmpeg/3.1.5 --enable-shared --enable-pthreads --enable-gpl --enable-version3 --enable-hardcoded-tables --enable-avresample --cc=clang --host-cflags= --host-ldflags= --enable-opencl --enable-libx264 --enable-libmp3lame --enable-libxvid --disable-lzma --enable-vda
libavutil 55. 28.100 / 55. 28.100
libavcodec 57. 48.101 / 57. 48.101
libavformat 57. 41.100 / 57. 41.100
libavdevice 57. 0.101 / 57. 0.101
libavfilter 6. 47.100 / 6. 47.100
libavresample 3. 0. 0 / 3. 0. 0
libswscale 4. 1.100 / 4. 1.100
libswresample 2. 1.100 / 2. 1.100
libpostproc 54. 0.100 / 54. 0.100
[mpegts # 0x7fc3c6000000] start time for stream 0 is not set in estimate_timings_from_pts
Input #0, mpegts, from '/Users/ibra/Desktop/daTgXic4JOI.ts':
Duration: 00:00:17.56, start: 0.000000, bitrate: 1220 kb/s
Program 1
Stream #0:0[0x102]: Data: timed_id3 (ID3 / 0x20334449)
Stream #0:1[0x100]: Video: h264 (Main) ([27][0][0][0] / 0x001B), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 90k tbn, 50 tbc
Stream #0:2[0x101]: Audio: aac (LC) ([15][0][0][0] / 0x000F), 44100 Hz, stereo, fltp, 150 kb/s
And here is a messy screenshot of the file opened in a hex editor:
https://www.evernote.com/l/AWlILw5PcmVEl4fSFitOfS2M8Wzy1WTVSZc
Any suggestions of how to insert the video duration in the metadata of the first chunk?
I cannot download the all the chunks and then concat in a single file because this will take too much time and the streaming must be instantly.
The container format is mpegts. There is no standard way to encode a duration in mpegts (or h.264 for that matter). So whatever you do will be proprietary. You could write it to the ID3 metaadata, but then kodi would need to be modified to handle this.

FFmpeg: NetStream.Play.StreamNotFound on RMTP stream

I want to take snapshots periodically of a RTMP live video stream.
I can see the rtmp video stream using VLC. This is the rtmp url:
rtmp://antena3fms35livefs.fplive.net/antena3fms35live-live/stream-antena3_1
I'm using the following command to capture the snapshots, according to the official FFmpeg site here:
ffmpeg -i rtmp://antena3fms35livefs.fplive.net/antena3fms35live-live/stream-antena3_1 -f image2 -vf fps=fps=1 out%d.png
The command produces the following output:
ffmpeg version N-64667-gd595361 Copyright (c) 2000-2014 the FFmpeg developers
built on Jul 14 2014 22:09:48 with gcc 4.8.3 (GCC)
configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzl
libilbc --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amr
enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --ena
libavutil 52. 92.100 / 52. 92.100
libavcodec 55. 69.100 / 55. 69.100
libavformat 55. 47.100 / 55. 47.100
libavdevice 55. 13.102 / 55. 13.102
libavfilter 4. 10.100 / 4. 10.100
libswscale 2. 6.100 / 2. 6.100
libswresample 0. 19.100 / 0. 19.100
libpostproc 52. 3.100 / 52. 3.100
HandShake: client signature does not match!
Closing connection: NetStream.Play.StreamNotFound
rtmp://antena3fms35livefs.fplive.net/antena3fms35live-live/stream-antena3_1: Unknown error occurred
I've tried it with another rmtp streams, but I'm still getting the exact same error.
What could be the problem?
Thank you!
I just tried your command and it worked fine for me. Maybe it is something about your FFMPEG installation? I am using version 2.4 on a Mac (tessus build).
I know other/older versions used "librtmp" for rtmp connections, which required some extra options behind the stream URL. See ffmpeg docs here:
ffmpeg documentation on librtmp
And librtmp docs here:
librtmp documentation
For an unprotected live stream, you may want to try quoting the stream URL and appending " live=1" within the quotes:
ffmpeg -i "rtmp://antena3fms35livefs.fplive.net/antena3fms35live-live/stream-antena3_1 live=1" -f image2 -vf fps=fps=1 out%d.png

Save continuous RTSP stream to 5-10 minute long mp4 files

How can I keep the flow (protocol rtsp, codec h264) in file (container mp4)? That is, on inputting an endless stream (with CCTV camera), and the output files in mp4 format size of 5-10 minutes of recording time.
OS: debian, ubuntu
Software: vlc, ffmpeg (avconv)
Currently this scheme is used:
cvlc rtsp://admin:admin#10.1.1.1:554/ch1-s1 --sout=file/ts:stream.ts
ffmpeg -i stream.ts -vcodec copy -f mp4 stream.mp4
But it can not record video continuously (between restarts vlc loses about 10 seconds of live video).
See this question and answer on Server Fault. In short, switch tools. avconv will do what you want. (ffmpeg has become avconv.)
The feature you are looking for is called segmentation. Your command line would look something like this:
avconv -i rtsp://10.2.2.19/live/ch01_0 -c copy -map 0 -f segment -segment_time 300 -segment_format mp4 "capture-%03d.mp4"
Alexander Garden solution works for ffmpep using the version below. Replace avconv with ffmpeg.
./ffmpeg -i rtsp://10.2.2.19/live/ch01_0 -c copy -map 0 -f segment -segment_time 300 -segment_format mp4 "capture-%03d.mp4"
I'm including this header because of the FFmpeg confusion over versions, the ubuntu schism and rapid development.
ffmpeg version N-80023-gd55568d Copyright (c) 2000-2016 the FFmpeg developers
built with gcc 4.8 (Ubuntu 4.8.4-2ubuntu1~14.04.1)
configuration: --prefix=/home/rhinchley/q10/ffmpeg_build --pkg-config-flags=--static --extra-cflags=-I/home/rhinchley/q10/ffmpeg_build/include --extra-ldflags=-L/home/rhinchley/q10/ffmpeg_build/lib --bindir=/home/rhinchley/q10/bin --enable-gpl --enable-libass --enable-libfdk-aac --enable-libfreetype --enable-libmp3lame --enable-libopus --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libx265 --enable-nonfree
libavutil 55. 24.100 / 55. 24.100
libavcodec 57. 42.100 / 57. 42.100
libavformat 57. 36.100 / 57. 36.100
libavdevice 57. 0.101 / 57. 0.101
libavfilter 6. 45.100 / 6. 45.100
libswscale 4. 1.100 / 4. 1.100
libswresample 2. 0.101 / 2. 0.101
libpostproc 54. 0.100 / 54. 0.100
Team work: Split the video source and have two processes alternate recording the time frame. You'll want to test how variable the startup time is, and how variable it is. You might want to set the processes priority to realtime to reduce start time variance. There will be some overlap but that sound like it might be ok for your application from what I infer. Example:
p1: sRRRRRRRRRwwwwwwwwsRRRRRRRRRwwwwwwwwsRRRRRRRRR...
p2: wwwwwwwwwsRRRRRRRRRwwwwwwwwsRRRRRRRRRwwwwwwwww...
time -->
s: startup
R: running
w: wait

Resources