Workflow: transcoding synchronised videos

I have three videos taken with three cameras simultaneously. One shows the back of a circuit board and one shows the front and the third shows a multimeter screen. Eventually, they will be composited together but right now I’m just transcoding cropped clips from each into ProRes. I don’t want to transcode the entirety of each clip because they’d be huge and wasteful (I only need about 10-20% of the footage).

My method is to record me clapping my hands and use this sound to align all three videos. I then apply crops to the videos to get just the stuff I need and then I watch one of the videos with the others turned off and use that to set the in and out points for all tracks. I remove the blank periods before each clip and then export from each layer using a separate preset which has modified resolution to match the crop.

After export, I drag the right of the video back out to its full length and start again. I think that this method does not lend itself well to the use of proxies because I’d be changing the filenames around to export every two minutes. I’ve enabled GPU processing (experimental) but I still have plenty of dropped frames during playback and my GPU load, GPU memory controller load and CPU load remain below 25% (usually around 15%) so I don’t see the point in getting better hardware when the current hardware isn’t utilised.

Mainboard: ASUS Z170 Pro Gaming, board revision 1.04, BIOS v1904
CPU: Intel i7 6700 (6th gen), Stock HSF, Undervolt Offset -0.15 V, LLC 4
RAM: Corsair LPX Vengeance 2x8GB DDR4-3000 (CMK16GX4M2B3000C15)
Graphics: Gigabyte NVIDIA GeForce GTX 950 w/ 2 GB