I am using blurring just as a proof-of-concept that we can indeed modify the videos that are being served instead of just blocking them. A later goal that is useful in the real world is giving relevant disclaimers to videos (example: putting a blood and gore warning on a video that contains blood and gore)

Thanks for clarifying. I don't have any specific recommendation, as this sounds like a (cpu) intensiv task. I personally wouldn't do this on-the-fly, maybe delegate this to a different server that also persists the blurred video (no reason to blur the same video every single time someone views it).

You can look into response.stream = some_function (https://github.com/mitmproxy/mitmproxy/blob/main/examples/addons/http-stream-modify.py) to process the video in chunks. But I don't know enough about video formats if this would be possible to blur them in arbitrary chunks.

I have been trying to use response.stream but the video content does not change at all,