Calculate your serve speed

[LeeD]

I gave you all the variables, and you saw the "spin".
Oh sorry, another newbie poster.
Welcome to TW forrums, without you newbies, the forums will die out.
Wind. I had long hair. No wind, no hair blowing around.

[spinorama]

Well I mean, I can kick serve 2nd serve and hit the fence every time before bounce, or smash a 1st at 125 and hit the fence in the same spot. So I just think the radar is the only way.

[LeeD]

"you saw the vid"....
Those serves were a semi flat serve with some amount of topspin because that early in the season, I couldn't find my grip.

[LeeD]

YOU can hit 125 flat serves AND twists that hit the wide baseboard around 24-36" high?
You're a better man than me.

[spinorama]

Originally Posted by LeeD

YOU can hit 125 flat serves AND twists that hit the wide baseboard around 24-36" high?
You're a better man than me.

Yeah, I've never really focused on how high up the fences it hits, but I'll pay more attention to that next time I'm out there. I'm thinking 2-3 ft sounds right sometimes more sometimes less

[LeeD]

So, do you have the biggest serves at the courts you play at?
My paltry "100 mph" flat serves bouncing around 34" high average is about the fastest at the two courts I frequent, which includes about 50 different players.
The local 5.5 guy, and his 6'3" brother, don't have faster first serves.

[spinorama]

Generally yeah. But I mean I'm 25 years old 6'2 and 195lbs so it's not weird that serving is my power

[spinorama]

I'm surprised the 5.5 doesn't break 100mph..

[tennisplayer_85]

Originally Posted by JPW

I've been pretty excited to get an iPhone 5, first, because I just had bad luck with AT&T, and second, to get this app I've done these calculations with a digital camera many times. This app certainly makes it much MUCH easier.

One question I do have for the developer is, would recording someone serving on TV be a good way to test the accuracy of this app? It seems like the timing would be the same and you could easily determine the other needed data. This just seems like a good way to check against an actual speed gun at a tennis tourney.

Recording the tv might work, i actually havent tested it, though it will be less accurate because tv's in general are 30 fps. If the input is 30 fps which you will record with 60 fps i suspect you will loose some accuracy.

I did test the app very well with real videos from the australian open in which the speedgun shows the speed. I did posts these results in this thread.

[LeeD]

The 5.5 serves with either hand, hits top/slice on first serves, avoids flat serves for some reason.
His brother, taller at 6'3", also avoid hitting flat serves, even in practice with tons of balls. Don't know why, but it's their game to fiddle with. Both hit with two forehands, but do have a rightie backhand slice.

[Chace]

Has anybody tried this ap that has had their serve clocked by a radar gun? If so how accurate is the ap in comparison to when you were clocked with the radar gun?

[JackB1]

I don't understand how this can be accurate? Because you approximate the spot you are serving from and you also approximate the spot the ball hits on the court. Also, where you stand isn't where you contact the ball. Some guys hit the ball 2 feet inside the baseline and some guys hit it from behind the baseline.

[tennisplayer_85]

Originally Posted by JackB1

I don't understand how this can be accurate? Because you approximate the spot you are serving from and you also approximate the spot the ball hits on the court. Also, where you stand isn't where you contact the ball. Some guys hit the ball 2 feet inside the baseline and some guys hit it from behind the baseline.

Yes it is true that there is some human error and that this affects the accuracy, but it’s not as bad as you think. Imagine you serve from the baseline to the service box line, which is exactly 60 feet.
Now you move the player 1 foot to the right because of human error.

If we use Pythagoras (a2 + b2 = c2) we can calculate the new distance 60(2) + 1(2) = 60.0083 feet.

A good formula is used to calculate at which height the player hits the ball. Imagine you hit the ball a foot higher than average. This means again that there is 0.0083 human error.

As you can see, moving the player or the ball horizontally and even the players height does almost not affect the speed at all.

Since the iPhone has a 60FPS camera the time factor is very accurate as well and almost does not give any room for error.

The only error is as you say, if the player hits the ball a foot inside the court, because this does affect the speed. The 60 feet distance would become 59 feet.
Which is 1.667% error. So if a player would serve 100 mph the serve would be 101,6 mph.

So let’s say there is an error range about 2,5 miles or maybe a bit more but on average it wont be this bad. In my opinion this is very reasonable.

Also hold in mind that a speed gun (even the professional ones) have an error range too and need to be calibrated and are sometimes inaccurate.

[JackB1]

Originally Posted by tennisplayer_85

Yes it is true that there is some human error and that this affects the accuracy, but it’s not as bad as you think. Imagine you serve from the baseline to the service box line, which is exactly 60 feet.
Now you move the player 1 foot to the right because of human error.

If we use Pythagoras (a2 + b2 = c2) we can calculate the new distance 60(2) + 1(2) = 60.0083 feet.

A good formula is used to calculate at which height the player hits the ball. Imagine you hit the ball a foot higher than average. This means again that there is 0.0083 human error.

As you can see, moving the player or the ball horizontally and even the players height does almost not affect the speed at all.

Since the iPhone has a 60FPS camera the time factor is very accurate as well and almost does not give any room for error.

The only error is as you say, if the player hits the ball a foot inside the court, because this does affect the speed. The 60 feet distance would become 59 feet.
Which is 1.667% error. So if a player would serve 100 mph the serve would be 101,6 mph.

So let’s say there is an error range about 2,5 miles or maybe a bit more but on average it wont be this bad. In my opinion this is very reasonable.

Also hold in mind that a speed gun (even the professional ones) have an error range too and need to be calibrated and are sometimes inaccurate.

Thanks for the reply. I think the bigger issue though is how u have to estimate the spot where the serve hit the court. I tried moving the spot around just a little bit and got readings as much as 10 mph apart. That's just a bit too large of a variance to make the app useful.

[NLBwell]

If you take multiple samples, you will converge on an average (it should be a normal distribution).

[JackB1]

Originally Posted by NLBwell

If you take multiple samples, you will converge on an average (it should be a normal distribution).

of course I can take an average. but that doen't make it any more accurate.

[tennisplayer_85]

Originally Posted by JackB1

Thanks for the reply. I think the bigger issue though is how u have to estimate the spot where the serve hit the court. I tried moving the spot around just a little bit and got readings as much as 10 mph apart. That's just a bit too large of a variance to make the app useful.

As explained in the previous comment, moving the spot where the ball hits (lands on) the court horizontally almost doesn't have any effect on the speed.
Moving the ball 1 foot in the vertical direction does by 1.6 miles of error (see previous comment). So the 10 miles you talk about would mean moving the ball more than 6 feet from it's real landing location. In normal circumstances you should be able to guess the balls location within a foot of its real location (especially if its close to the lines).

I have tested it with real videos from the australian open where the professional radar gun shows the speed so i was able to compare. All the calculations i did where within 2,5 miles of the radar gun's speed except for 1 out of 10, I suggest you try it your self at a pro tournament.

It would still be really nice to overcome the 'human error' of this step and i do have a way to reduce it! the idea is to overlay the 4 corners of the servicebox and then transfer the balls location from 1 dimension to the real dimension of the court. I am confident that this will be quite accurate. I hope to come with an update with this functionality, but give me some time
Anyway i think the App, as how it is, is the most accurate we can get with today's smartphones. And i still think the 2.5 miles error is not bad at all!

[v-verb]

Originally Posted by tennisplayer_85

As explained in the previous comment, moving the spot where the ball hits (lands on) the court horizontally almost doesn't have any effect on the speed.
Moving the ball 1 foot in the vertical direction does by 1.6 miles of error (see previous comment). So the 10 miles you talk about would mean moving the ball more than 6 feet from it's real landing location. In normal circumstances you should be able to guess the balls location within a foot of its real location (especially if its close to the lines).

I have tested it with real videos from the australian open where the professional radar gun shows the speed so i was able to compare. All the calculations i did where within 2,5 miles of the radar gun's speed except for 1 out of 10, I suggest you try it your self at a pro tournament.

It would still be really nice to overcome the 'human error' of this step and i do have a way to reduce it! the idea is to overlay the 4 corners of the servicebox and then transfer the balls location from 1 dimension to the real dimension of the court. I am confident that this will be quite accurate. I hope to come with an update with this functionality, but give me some time
Anyway i think the App, as how it is, is the most accurate we can get with today's smartphones. And i still think the 2.5 miles error is not bad at all!

Agreed. Don't let the naysayers get to you.

If I could ever find my iPod touch I'll buy it - dying to try the app!

[Tight Lines]

Originally Posted by JackB1

I don't understand how this can be accurate? Because you approximate the spot you are serving from and you also approximate the spot the ball hits on the court. Also, where you stand isn't where you contact the ball. Some guys hit the ball 2 feet inside the baseline and some guys hit it from behind the baseline.

The calculations are based on formulas from http://www.donthireddy.us/tennis/speed.html. That website has a calculator that calculates the speed for you. All you need is the frame rate of your camera and the number of frames it takes from the time the ball leaves the racket and the time it hits the ground.

The app basically does the same thing, but makes inputting data much easier.

I have used the app many times now for me and my friends, and believe that it is reasonably accurate although I have never compared it to a radar gun.

--- Harry

[dlesser13]

I serve faster than LeeD and the majority of posters on this board.