2 Comments
author

Yeah for sure, LLM agents can trigger generative videos and edit them together. The only significant limitation on that up until recently has been API / model access for vid gen models but that’s changing (for example release of recent SD Video model).

In terms of rights, TikTok is just a demo. I wouldn’t mean to suggest scrapping them for commercial use. Although there could be a lot of interesting use cases for insight discovery using this pipeline.

Expand full comment

Super awesome. Love this. NOW... could you expand the machine to have it text-to-generate original video? Also curious how you're thinking about rights/usage as it relates to video scraped from TT.

Expand full comment