^ I believe that you are coming away from this with an erroneous notion of what is going on with these serves. Murray, Federer, Vaidisova, Sharpova and many other servers are still looking at the ball very close to contact. They are obviously watching the ball at the trophy phase and are still watching as the tossing arm drops and as the racket head drops. Even at the "big L" (when the racket arm becomes fully extended), most of these servers still have their eyes trained on the ball -- even tho' the head has started to move/turn.
Even Roddick appears to be looking at the ball at the "big L". However, his head quickly turns and looks forward from this point. Not so for the other servers mentioned above. Note that the time it takes to move the racket head from the "big L" up to the contact point is very short -- it would be measured in milliseconds or centiseconds. For this reason alone, I maintain that most servers are fixated on the ball very close to contact.
Furthermore... even when the head moves/turns, the eyes can still adjust their position so that the foveal (central) vision is still trained on the ball. It is not that difficult for the eyes to point in a direction that is 45 degrees offset from the primary direction of the head. So... even when they are not directly facing the ball, the eyes can still be looking at the ball.
Bottom line: It is probably best for most of us to watch the ball as close to the contact as possible. I've seen plenty of players who pull the head down much too early -- before the "big L" -- or even before the racket starts it's upward swing. If you decide that it is not important to watch the ball to contact, you may end up pulling your head/eyes down much too early as a result.
Very interesting points but this thread was more for just the one simple question - When serving, are the pros looking at ball impact?
I had believed that for looking at the ball for both ground strokes and serves that I should apply exactly the same principle - keep your eye on the ball impact area before, at, and after impact. I still believe it for ground strokes but not now for serves.
For my serving practice in the past I found that if I looked at ball impact and also kept my head & eyes still afterwards it usually resulted in a more accurate hit. But looking at the ball is my biggest tennis problem, who knows how I view the ball in my usual match serve.
I noticed looking at pro serve high speed videos from miscellaneous viewpoints that often the servers did not appear to be looking at the ball at impact. Also, I know that if you do not have the correct camera view and do not look just at ball impact (or before) you can easily miss what is going on. So the question was first to get some stats on how often the pros are looking at the ball at impact.
You mention breaking off at some point before impact earlier in the motion. Researching that point is very interesting and most important - what is really going on - but it is much more difficult to observe in videos. Can you find videos showing when pros break off looking? Or, if their eyes are still pointed up at an extreme angle when the head appears to be looking straight ahead? See eye test below.
Since the body and arm are at an angle (when viewed from the side) and the racket is about vertical at impact I believe that the pros break off looking at the ball because their motion involves the upper body angling forward rapidly and it stresses the neck as mentioned in some replies. Perhaps for a club player such as myself the option to look at the ball through impact - as I have practiced for my serves - is possible because the rapid upper body motion is not being used. ? I'll have to compare my side view serves to pros to see the differences in motion and in body, arm and racket angles at impact.
(The frames in reply #19 of JackB1 above show that his head is looking up and is very stationary, unlike Stosur's, but unfortunately the ball impact is not shown.)
Some points for video analysis of this issue -
1) no cherry picking as there are usually a handful of videos to show anything.
2) high speed video, >200 fps.
3) examine the frame showing ball contact (or before)
4) side view, above view or some other view that clearly shows the head orientation at impact.
5) there will be uncertainty as to where exactly the eyes are pointing - find some reasonable way to interpret the videos from head position. Do the eyebrows block the line from the eye to the ball impact? Is the server wearing a baseball cap with a blocking rim, etc.?
Field of view eye test - Stand with the head looking straight ahead. Hold a tennis ball at arm's length above your head. Point the eyes up as much as possible but keep the head oriented straight ahead. Move the ball back and forth to determine where your field of view starts. When I do this and look at my arm angle, it is about 30° to the vertical. (It helps to have a mirror to the side to get the arm angle.) Because the eyes are set back from the eyebrows, the eyebrows limit the field of view upward. Probably can also search the eye's FOV on the internet.
It would, of course, be better to find some tennis research using motion capture with high speed 3D cameras.