How many source images do you have, and should the output definitely be 25fps?also why not use something like https://amiaopensource.github.io/ffmprovisr/#images_2_video To process an image sequence rather than concat? Unless your images are appended by datetime rather than sequential numbers?