I have started derating gray-to-gray response times by a factor of 10.
That is, an advertised 5ms response time is actually 50ms. One other thing to consider: was the shutter speed on the camera used less than 5ms? Obviously, you will capture two frames if the camera's shutter speed is only 1/30th of a second.
Rationale:
Response time used to be listed as the time it takes a pixel to go from Black-to-White-to-Black.
Therefore, (in theory) a relatively "slow" 40ms response time should be enough for 50fps (with minor blurring between adjacent frames).
Knowing that the (LCD) pixels have capacitance, I decided to approximate the the color change as proportional to the (charge or) discharge of a capacitor through a resistor.
The common rule of thumb is that it takes 5 time constants to reach "steady state" (within 5% of final value). I could only assume that the "grey-to-grey" response time refers to that time constant. Therefore, to go from black to white and back takes 5 time constants each way. (for a total of 10)
Now, I do have a post-secondary diploma in electronics, but not a lot of experience in the field. My reasoning may be completely invalid. YMMV.