ffmpeg_hd2sd_transcode_segment

ffmpeg_hd2sd_transcode_segment

# convert HD video to SD and transcode in one-shot
ffmpeg -i video.ts -preset ultrafast -y [-loglevel quiet] -aspect 16:9 -s 640:360 video.mp4

# batch segment the video in one-shot (no looped seeking)
# NOTES: "-vcodec" and "-acodec" can be changed to preferred stream types, e.g. libx264 and libfaac, respectively
#        "-vbsf h264_mp4toannexb" optionally allows for broken mp4's to still get parsed
ffmpeg -i video.mp4 -vcodec copy -acodec copy -map 0 [-vbsf h264_mp4toannexb] [-loglevel quiet] -f segment -segment_format mp4 -segment_times <time_sec_1>,<time_sec_2>,<time_sec_3>,<...> output%02d.mp4

# TODO: needs further investigation on segmentation accuracy; cutting may still happen on i-frames due to keyframe-level accuracy, but this may have been solved by mpegts-to-mp4 transcoding

Here's a script for directly parsing triple-marker files and passing it as a parameter to segmentation:

#!/usr/bin/env bash
#
# $1 : path to mpegts file
# $2 : path to triple marker file
# $3 : output path + prefix (e.g. /path/for/output/prefix)

T="$(date +%s%N)"

ffmpeg -i $1 -preset ultrafast -y -loglevel quiet -aspect 16:9 -s 640:360 video.mp4

ffmpeg -i video.mp4 -vcodec copy -acodec copy -map 0 -vbsf h264_mp4toannexb \
 -loglevel quiet -f segment -segment_format mp4 \
 -segment_times `cut -d' ' -f1 $2 | tail -n +2 | tr ',' '.' | awk -F: '{ print ($1 * 3600) + ($2 * 60) + $3 }' | tr '\n' ',' | sed 's/.$//'` \ 
 $3%02d.mp4

T="$(($(date +%s%N)-T))"
S="$((T/1000000000))"
M="$((T/1000000))"
printf "Time elapsed: %02d:%02d:%02d:%02d.%03d\n" "$((S/86400))" "$((S/3600%24))" "$((S/60%60))" "$((S%60))" "${M}"

The portion that does the actual parsing is (it assumes the first column of the file contains cut times, i.e. beginning of new segments):

cut -d' ' -f1 $2 | \
 tail -n +2 | tr ',' '.' | \
 awk -F: '{ print ($1 * 3600) + ($2 * 60) + $3 }' | \
 tr '\n' ',' | sed 's/.$//'

Example usage:

bash segment.sh /ptvn/RawVideos/20130725/CNN_Newsroom_20130725_1400.ts \
 /ptvn/storyboundary_ccasr/CNN_Newsroom_20130725_1400/CNN_Newsroom_20130725_1400.aligned/CNN_Newsroom_20130725_1400.cc_tripleColumbia \
 /ptvn/tmp/test_segment/test_out

tail -n +2 in the triple marker parsing should be ignored if the first line of the file is still usable

To change to cutting by frames use -segment_frames instead of -segment_times, and change the triple marker parsing to:

cut -d' ' -f1 $2 | \
 tr ',' '.' | \
 awk -F: '{ printf "%d,",((($1 * 3600) + ($2 * 60) + $3) * 29.97) }' | \
 sed 's/.$//'

(assuming the frame rate of the video is 29.97 fps and the first line of the triple marker file is valid)