ciescope¶
Watch on youtube.com#! /bin/sh
ifn="Drifting with Cars.mp4"
ifnb="`basename \"${ifn}\" .mp4`"
pref="`basename $0 .sh`"
#
ffmpeg -y \
-i "${ifn}" -filter_complex "
[0:v]split=2[v0][v1];
[v1]ciescope,scale=1920/3:1080/3,setsar=1[v2];
[v0][v2]overlay=x=W-w-50:y=H-h-50[v]
" -map '[v]' -an "${pref}_${ifnb}.mp4"
BTW. The script that created the uploaded video uses “split”, but in this case you don’t need “split”:
#! /bin/sh
ifn="Drifting with Cars.mp4"
ifnb="`basename \"${ifn}\" .mp4`"
pref="`basename $0 .sh`"
#
ffmpeg -y \
-i "${ifn}" -filter_complex "
[0:v]ciescope,scale=1920/3:1080/3,setsar=1[v2];
[0:v][v2]overlay=x=W-w-50:y=H-h-50[v]
" -map '[v]' -an "${pref}_${ifnb}.mp4"
There are many cases that can not be intended without “split” (“asplit” for audio). This is because as the filter graph becomes more complicated, data to be passed downstream of the graph often separates from “processed data” and “original data”.