This idea created when i need to access user media video for tensorflow APIs.
On that case i need user media video for background image and convert that video to image data and process to tensorflow APIs (get pose)

This is my post when i asking how to get access user media video and i think dev team should make it to official scripting SDK
https://www.construct.net/en/forum/construct-3/scripting-51/tensorflow-pose-net-construct-173694