It is possible that the OP of this thread is dealing with a lot more camera head than most of us use.
1) There are several flavors of HDMI and what it will output on an output connection is sometimes predicated on the setup.
2) You would never capture high-end video thru a common USB port. The only USB configuration with USB3 is dealt with in very specialized configuration.
3) You certainly would not capture an HDMI output thru a USB channel, it begs the conditioning of what would happen to the stream.
4) Most likely the camera head of the OP - should have some other options for capture or preview?
I take it you want to preview and pre-compose a matte, in which case you don't need anything that's totally pristine. Does it need to be live or are you going to capture video and then make the composite?
At the very very low end I've used a Hauppauge DVR Rocket to capture an HDMI stream. It'd be cumbersome but not impossible. It would create a TS file on disk from which you could pull a rough key.
This definitely wouldn't be real-time so you couldn't use it to allow a weather announcer or puppeteer (for example) to gauge their movements.