Is it possible to rerun an existing video? You can save a video from Oak to a file, then possibly use ffmpeg to convert that to a standard format. What if you want to use that file as input to another depthai model? How I imagine this might work is that the file is streamed to the Oak, which uses it instead of the lens to run a model. That way one could use the same input run against multiple models to compare them. This would save labor and get a more precise comparison.

So you could definitely save the video file in ffmpeg format through either RGB orf mono data recorded. This video could further used as input for other scripts through use of certain arguments like
-vid VIDEO, --video VIDEO
Path to video file to be used for inference
This could be used to run the video in the format -
python3 main.py -vid ./input.mp4

Use of video input arguments are implemented in examples like age-gender, person-reidentification, pose estimation and more. You could also modify other pieces of code to include these arguments.
So, you could use a sinlge pre recorded video as input for several models.

Let me know if your question was regarding something else, and this did not solve your doubt. Thanks!

Links to explore for video input - https://github.com/luxonis/depthai-experiments/tree/master/gen2-age-gender

https://github.com/luxonis/depthai-experiments/tree/master/gen2-human-pose

https://github.com/luxonis/depthai-experiments/tree/master/gen2-pedestrian-reidentification