As Patrick described in this post before:
in my opinion, when I disable the video capture before testing, the start render cannot be taken from the video frames. But it still appears in the test result. How does the test runner get its value? Any feedbacks are appreciated.
There was a change to the desktop agents ~6 months ago so they always capture video locally and only upload it with the results if video capture is enabled. That way they can still calculate all of the video metrics. It’s generally safest to still enable video capture so you can validate results in the filmstrip and it’s still needed for the mobile agents.