No amazon involved here, as to google they have a data usage policy of not using the video files, but if you don't trust it, that's of course legitimate

I'm already so far into the Google ecosystem that a few more videos won't make much difference.

But my bigger concern is how to process lossless videos. I'm assuming that won't work, compressed copies required? For me, that just makes it extra steps, extra conversions.

Compressed or not, uploading multiple gigabytes of videos to the cloud and downloading them again just to do scene split detection seems enormously wasteful. Software existed 20+ years ago to do that on your own computer. Heck, even some consumer VCRs in the 1990s could automatically detect scene breaks that dipped to black. (Although the fine folks at Macrovision of course patented a way to try to defeat that, as people were using it to skip commercials!)