But at least in x264 they are not independent It should not affect your answer to OP, since -g 24 alone should be sufficient. No idea about vp8/9, but I have learned my lesson in x264. In x264, setting both to the same value does not do what you think it does. I don't recall whether that is FFmpeg or x264's logic and neither have I tested vp8/9. I have learned to be very careful with these, as there is an undocumented dependency and no console warning. I don't know about vp8/9, but at least with x264, an undocumented gotcha is that keyint_min cannot be set greater than ((keyint/2)+1), otherwise it is ceiled at ((keyint/2)+1). c:v libx264 -profile:v high -level:v 4.BTW - watch out with these two.
#FFMPEG COPY KEYFRAME TV#
# use the libx264 for producing an optimized h264 file, with the "High Profile" setting, which is # the primary profile for broadcast and disc storage applications, particulary for high-definition # tv applications, and by using the veryfast preset # live stream video with timestamp indefinitely (need a RTMP server) # settb=AVTB: force timestamp to default AVTB which is 10e-6 to have timestamp in us # setpts='trunc(PTS/1K)*1K+st(1,trunc(RTCTIME/1K))-1K*trunc(ld(1)/1K)': truncate # credit: # libavfilter input virtual device, needed to generate the black background # generate video with timestamp & pts drawn behind black screen # check if a video has B-frames (0: no b-frames, 2: has b-frames)įfprobe -loglevel 16 -show_streams -show_private_data -print_format flat -i input.mp4 | grep has_b_framesįfprobe -loglevel 16 -show_frames input.mp4 | grep pict_type | less # concatenate images from folder into a video for f in *.jpg do echo "file ' $f '" > to_video.txt done # -r 0.5: show every image for 2 secondsįfmpeg -r 0.5 -f concat -i to_video.txt -vcodec libx264 video.mp4 # to concat multiple videos, we need to create a text file containing all the videos to concatenate in orderįfmpeg -f concat -i files.txt -c copy concatenated_video.mp4 # cut video without from 10s to 20s # using "-vcodec copy" to take the input codec => no re-encoding # may only work if the video do not contain any B-frameįfmpeg -ss 10 -to 20 -i input.mp4 -vcodec copy out.mp4 # display the frame timestamps with their associated coded picture number (= encoding order) # /!\ coded_picture_number != frame number # If there are no B-frames, the coded_picture_number is the same as the frame numberįfprobe input.mp4 -select_streams v -show_entries frame =coded_picture_number,pkt_pts_time -of csv = p =0:nk = 1 -v 0 -pretty | less Ffprobe -loglevel 16 -show_streams -show_private_data -print_format flat -i input.mp4
0 kommentar(er)
