This is a set of scripts to facilitate the management of acquisition video and ensuring there proper encoding with software like whisk

The main objective is to obtain readable video that can be compressed while retaining software compatibility. As some of the highspeed video aquired are

1. Acquire highspeed video with StreamPix (Norpix software) as *.seq file
2. Use batch conversion utilities from StreamPix (Norpix software) to convert *.seq to *.avi
3. Convert/encode the *.avi to *.mp4 using ffmpeg
   • windows open cmd
   • go to the folder (with cd) where the script avi2mp4.py is located
   • run the script python avi2mp4.py
   • for -qscale:v 10 this enables the conversion of 14Tb --> 0.04Tb (44Gb) overnight CPU based with IntelCore i9-9820X CPU @ 3.30GHz 10 cores. The mpeg4 encoding is not suported with GPU encoding.
   • for -qscale:v 4 this enables the conversion of 14Tb --> 0.19Tb (190Gb) overnight CPU based with IntelCore i9-9820X CPU @ 3.30GHz 10 cores. The mpeg4 encoding is not suported with GPU encoding.

important consideration: - we encountered memory issues when running long highspeed video with whisk thus the video can be sliced - encoding is critical to have tbc, tbn and tbr consistent (see whisk issues here)

key points for ffmpeg usage: -codec:v: mpeg4 necessary to be able to have good fps tbn tbr matching
-r: enables to have the frame rate of intres
-qscale:v: this is the quality of the video (from 1 highest quality/larger file to 31 smallest quality/smaller file)
-codec:a: needed to have audio codec
-video_track_timescale: force the tbn value

1. slice the video with sliceForWhisk.py
2. use git Bash
3. setup GNU parallel for Windows see
   • for install see []
   • tutorial
4. run bash script for parallel analysis of the task
   • see bashForWhisk.sh
   • TODO: complete and improve the script
   • TODO: better naming convention
   • TODO: run script one after the other
5. important note for analysis
   • if error when saving missing segment use Save As instead of Save
   • alternatively re-run the begining of the analysis to have the proper detector binaries created within folder of interest