What does Hawk Eye really show?

S

sureshs

Bionic Poster
This question comes up over and over again and here are my thoughts. The main confusion is always whether something is real or not.

Is the ball real?
No, it is not a photo of the actual ball, but a graphic rendering.

Are the court and the lines on it real?
No, they are not photos of the actual court, but renderings, even though the courts are colored according to the surface.

Is the shadow of the ball real?
No, it is a graphic rendering based on virtual light sources calculated to make the scene clear to the viewer. I don't think the actual lighting on the court is used for calculation of the shadow.

Is the deformation of the ball (elongation) on the surface real?
No, it is a calculated rendering based on the flight of the ball and the physics of a typical ball and the properties of the surface. I don't think it is based on an actual photograph of the impact zone at the impact time, but again since that is so important for in or out decisions, it might be.

Did the ball travel at the same speed in real-time as shown in the replay?
No. Can someone speculate if the replay is faster or slower than real-time?

Is the path of the ball real?
Within the precision of the camera sensors, yes. In-between points are mathematically interpolated.

Then, did the ball actually travel through every point in the interpolated renderings?
Almost, because physics does not permit abrupt changes in such situations.

What happened to the spin of the ball?
It is not rendered.

If spin is not taken into account, then how can the path through the air be accurate?
Because the path is based on actual high-speed photography.

Is spin information (rpm) calculated inside the computer and used for anything?
I don't know. If it is used, it would be to refine the deformation mark (elongation), but I don't know.
 
Last edited:
aimr75

aimr75

Hall of Fame
Does it really matter? What it gives you is a result that cannot be disputed, for better or worse. To that end, the game moves on and the next point is played.
 
M

maggmaster

Hall of Fame
Essentially it is a very precise digital rendering based on a series of predictive algorithms informed by flight data from the cameras. For all intents and purposes it is accurate and precise.
 
protestov.tt

protestov.tt

Rookie
the main question I have: black spot that we see after the ball impacts surface - it is shadow of the ball (equal to its diameter) or it is zone of contact between the ball and surface(little spot, by which the ball touches surface)?
 
S

sureshs

Bionic Poster
protestov.tt said:
the main question I have: black spot that we see after the ball impacts surface - it is shadow of the ball (equal to its diameter) or it is zone of contact between the ball and surface(little spot, by which the ball touches surface)?
Click to expand...

Good point. I take back my previous statement about there being a shadow.

It is not a shadow, but a rendering of the squashed zone of contact which is NOT a little spot at all. Since we are used to thinking of the ball as a sphere, Hawk Eye retains the shape of the ball but shows the squashed zone of impact as a shadow. In reality, the ball would flatten and assume the cross section shown by the shadow. Showing it only as 2 dimensional shadow makes it easier for the viewer to see in/out.
 
S

saesoo

New User
Probably not really related to the original posting, but given the technology, could it replace line judges? Would it? Should it? Technology would have to support immediate/real-time response.
 
G

gameboy

Hall of Fame
Hawk Eye system is pretty simple.

There are multiple cameras positioned around the court. Computers compute the exact position of the ball by the location of the ball within the camera's field of view, using triangulation (similar to how GPS figures out your location).

The exact location of the ball between the high speed camera's frame is interpolated using physics calculation.

When calibrated properly, this system is highly accurate.

Installing the system and keeping it properly calibrated is not cheap, so it is unlikely they will replace human judges any time soon.
 
F

fuzz nation

G.O.A.T.
While it shows me a whole bundle of different things depending on my mood, it basically shows me a disappointing gimmick.

While it diffuses much of the controversy over the accuracy of line calls, I don't care for it. It's not set up on every court. That makes for inconsistent officiating at different matches in the same tournaments. BAD!

Since they can't get it dialed in well enough to work at the French Open, etc., I don't trust it anywhere. Especially on the grass courts, which are also slightly irregular like the clay. Maybe if it showed high resolution video of the actual ball on the court instead of an adorable video image. Please don't ever clap in anticipation for that if you're at an event... I'm beggin' ya!

I wish they'd spend the money elsewhere and let the line judges and chair umpires do the job they're entrusted with out there. Give them a small raise and more authority I guess, instead of using this gizmo that marginalizes them.

One man's opinion...
 
S

sureshs

Bionic Poster
The issue of showing high-resolution video at the impact zone is addressed in my OP.

It is not practical. The reason is that the impact lasts only for a few milliseconds. Either the camera has to be run with frame rates of close to 1000 frames per second all the time, or it has to start running at this rate very close to impact time, which would then have to be identified with precision. Currently, neither approach is possible under the cost structure and precision available.
 
T

tennis_ocd

Hall of Fame
You gotta hand it to the hawkeye marketing. They take a calculated guess; a guess that's not good enough to be used if actual confirmation ball marks are left on court, and sell it to most as an absolute representation of what happened.

They could take a 57% probability of a ball clipping a line and translate to a red light/green light confirmation of a call but they actually throw up a graphic of what happened. How the ball flattened and everything. One wonders why didn't they just sketch in some ball fuzz....

I miss the human element and the fun it can cause.
 
T

tennis_ocd

Hall of Fame
Pickle9 said:
And what this thread has to do with "tennis instruction/tips"....?
Click to expand...
Tips on how to best use my inner city park hawk-eye challanges? How to goad my opponent into wasting his?

with work, hawk-eye could be tweaked to who got to court first, first bag on bench, did she really bring a bag and purse and how many times did her dog really interrupt my lesson type challanges. technology could eliminate taking it to your momma's house.
 
S

sureshs

Bionic Poster
tennis_ocd said:
You gotta hand it to the hawkeye marketing. They take a calculated guess; a guess that's not good enough to be used if actual confirmation ball marks are left on court, and sell it to most as an absolute representation of what happened.

They could take a 57% probability of a ball clipping a line and translate to a red light/green light confirmation of a call but they actually throw up a graphic of what happened. How the ball flattened and everything. One wonders why didn't they just sketch in some ball fuzz....

I miss the human element and the fun it can cause.
Click to expand...

They don't sell it as an absolute representation. They provide the margin of error.
 
C

Chas Tennis

G.O.A.T.
+/- 3.8 mm, really?

Hawk-Eye is a measurement system the accuracy of which is not well known or understood.

I believe that +/- 3.8 mm is thrown around as the accuracy. Not sure of my interpretation of the details but often "+/- 3.8 mm" would mean that the 'true' value - ball contact location - would have a 67% probability of being within +/- 3.8 mm of the indicated value (ball shadow). The true ball contact would also have a 33% chance of being farther than the +/- 3.8 mm from the HE indicated value.

I don't know what kind of testing was done to verify the 3.8 mm number. But it would be very interesting to take a close up high speed video of a ball striking near the line and compare its video to Hawk-Eye's model of the impact. It should only take very few impacts to see the accuracy.

In matches, HE will show a ball impact shadow that is 1 mm from the white line and the crowd believes it to be perfectly accurate. The crowd sometimes even seems to murmur that the lines person has made an error.........

It does seem to be neutral, fair and probably does improve the over all number of correct in and out calls. It's popular.

But should a call be reversed on a 1 mm in or out indication? Or should the point be replayed because the true impact was uncertain due to Hawk-Eye's accuracy?

History - paper with some Hawk Eye vs MacCam
http://web.mit.edu/skelkar/www/shreeharsh-kelkar_files/Kelkar-Hawk_Eye_draft.pdf

Does anyone recall seeing a Hawk-Eye vs MacCam comparison of a line call on TV? I think I do.
 
Last edited:
T

tennis_ocd

Hall of Fame
sureshs said:
They don't sell it as an absolute representation. They provide the margin of error.
Click to expand...
which is?? and what does it mean?? I can't find it.
What I see are a lot of very close representations that may not be any more accurate than the actual line call. Heaven forbid they acknowledge they can't accurately overrule within margin of error.

And why sell hawk eye as an aid to getting the right call and not permit some real video challenges of double bounces, body strikes, and net touches??
 
SystemicAnomaly

SystemicAnomaly

Bionic Poster
I recall that the footprint shape of the bounce some years back showed more variety than it has in the past couple of years (or more). It seems that all the ball footprints have looked the same for a while now. Have not been looking closely at the footprint for the 2014 AO. Are the shapes still all the same?

I am thinking that they removed some of the software calculations to more accurately depict the footprint. Perhaps the Hawk-Eye guys did not feel that their footprint (shape) depiction was not accurate enough so they stopped calculating it. Perhaps they will bring that feature back (unless I am just imagining all of this).

I recall watching a couple of matches at the BOW Classic (Stanford) last year of the year before where Hawk-Eye appeared to be getting some of the call wrong at the baseline. There was a moderately strong wind moving along the length of the court. A number of balls that appeared long at the baseline at one end of the court, were called "good" by Hawk-Eye. It appeared that the wind had an effect on the length of the shot and H-E may have not quite calculate the trajectory correctly. Or perhaps the system was not properly calibrated.
 
J

jrs

Professional
Hawkeye has a few things going for it. Fancy camera, computers, computer graphics, physics and advanced mathematics. Who can argue with these things. Even if it is wrong.

Ask them to put at the French open or Clay court tournament and check the results with the mark on the court.

Pretty sure they won't do it. There were numerous occasions where players pointed to the mark on the court and they went with technology....it's stupid!
 
S

sureshs

Bionic Poster
tennis_ocd said:
which is?? and what does it mean?? I can't find it.
What I see are a lot of very close representations that may not be any more accurate than the actual line call. Heaven forbid they acknowledge they can't accurately overrule within margin of error.

And why sell hawk eye as an aid to getting the right call and not permit some real video challenges of double bounces, body strikes, and net touches??
Click to expand...

Those don't require Hawk Eye. Replay is sufficient. Unfortunately, replays have not been adopted in tennis.
 
C

Chas Tennis

G.O.A.T.
sureshs said:
Those don't require Hawk Eye. Replay is sufficient. Unfortunately, replays have not been adopted in tennis.
Click to expand...

Covering all the lines with separate high resolution cameras is probably not practical.

A camera system that adequately tracks the ball has probably not been developed yet.

There is always the problem that the ball blocks its contact area from most camera viewing angles. When the MacCam is set up to look along the line focusing on a particular spot, the results, in my opinion, can be much more accurate than the Hawk-Eye locations.
 
5263

5263

G.O.A.T.
key to hawk eye is that it is arbitrary and has no favorite. Not perfect, but hopefully not fixed by any outside source.
 
Bartelby

Bartelby

Bionic Poster
So could it tell the umpire immediately whether the ball is in or out. according to its calculation, if they so decided?

Or is the drama of the challenge system the only thing that prevents this 'let style system' or is it technically impossible?
 
S

sureshs

Bionic Poster
Bartelby said:
So could it tell the umpire immediately whether the ball is in or out. according to its calculation, if they so decided?

Or is the drama of the challenge system the only thing that prevents this 'let style system' or is it technically impossible?
Click to expand...

I think it can. It might need more computing power to keep doing that for every shot in case the umpire calls for a decision immediately, but computing power is a dime a mega flop these days.
 
Bartelby

Bartelby

Bionic Poster
I'm thinking about whether any ball within say fifteen centiments or so of the line could be automatically buzzed to the umpire as out so that he could rule or over rule on the spot.



sureshs said:
I think it can. It might need more computing power to keep doing that for every shot in case the umpire calls for a decision immediately, but computing power is a dime a mega flop these days.
Click to expand...
 
C

Chas Tennis

G.O.A.T.
Night vs daytime accuracy ?

The light level of direct sunlight is much greater that tennis court night lighting in stadiums.

Lower light levels usually force the camera to use a slower shutter speed to collect enough photons. With slower shutters there is increased motion blur. To illustrate, compare the motion blur for a Smartphone in direct sunlight to that indoors at night when viewing a very fast object.

I would expect more motion blur for night matches and that the accuracy of Hawk-Eye would be less for night matches.

Anyone ever hear of a difference in accuracy at night for Hawk-Eye.

BTW - I vote for Hawk-Eye but would like to see neutral source tests of its accuracy.

Just saw call reversed at the AO because the ball shadow appeared to touch the line - maybe 1 mm of overlap........
 
G

gameboy

Hall of Fame
There is no decrease in accuracy during night games. The cameras run at exactly the same speed during the day and at night. The Hawk Eye cameras are not like normal cameras. It does not care about the overall picture quality. It only cares about picking up the tennis ball from the background. Once minimal light level is achieved, there is no difference.

Using high speed camera for line calls is not practical as the camera would have to be placed very close to the line and FOV is limited.

Hawk Eye can call all the line calls right now. They don't because of traditions. The delay in challenges are due to setting up the display. The call is made instantaneously.
 
S

sureshs

Bionic Poster
Bartelby said:
I'm thinking about whether any ball within say fifteen centiments or so of the line could be automatically buzzed to the umpire as out so that he could rule or over rule on the spot.
Click to expand...

What about balls close to the line? Aren't they are the real problem? We had talked before in another thread to have sensors on the lines.
 
SystemicAnomaly

SystemicAnomaly

Bionic Poster
Chas Tennis said:
... Just saw call reversed at the AO because the ball shadow appeared to touch the line - maybe 1 mm of overlap........
Click to expand...

The HE system could very well have gotten it wrong in that case. Given the current accuracy of the HE system, I would not be inclined to give the system absolute authority on all line calls. It is fine to let HE have the final say on a challenged call. As it is now, challenging a call is a bit of a gamble -- and that is fine by me.

http://www.hawkeyeinnovations.co.uk/page/sports-officiating/tennis

ITF testing of Hawk-Eye revealed a mean error of 3.6 mm. This says that some errors were greater than 3.6 mm. Other sources talk about a 5 mm error/accuracy. Not too long ago, I was present at a BOW Classic match where it appeared that HE might have been off by more than that for quite a few calls on one of the baselines. The error could have been due to wind or wind gusts or perhaps the system had not been calibrated as well as it could have been.

Another issue is the question of the size and shape of the ball's footprint that the HE system generates. The last time was paying close attention to it, it did not appear that the system was calculating the shape of the footprint. All footprint representations appeared to be exactly the same, a light oval, regardless of the incident angle and other variables. The size and shape of the footprint can very well have an impact on whether it was touching the line or not.

http://www.popsci.com/technology/article/2013-07/how-accurate-hawk-eye-tennis-ball-caller
 
S

sureshs

Bionic Poster
SystemicAnomaly said:
The last time was paying close attention to it, it did not appear that the system was calculating the shape of the footprint. All footprint representations appeared to be exactly the same, a light oval, regardless of the incident angle and other variables. The size and shape of the footprint can very well have an impact on whether it was touching the line or not.
Click to expand...

I think a spherical object squashed at any angle of incidence will always result in an ellipse (or circle) as the shape of the projection. So while the shape can be narrower or broader, it will always be an ellipse.
 
D

dizzlmcwizzl

Hall of Fame
I have never seen the Mac Cam / Hawk eye comparison. I would love to see a match where they check HE on the baseline and then look at MacCam for comparison.

Unfortunately, the conspiracy skeptic deep inside me feels like we wont see that .... for fear that we would be shown the true margin of error in their measurements.
 
SystemicAnomaly

SystemicAnomaly

Bionic Poster
sureshs said:
I think a spherical object squashed at any angle of incidence will always result in an ellipse (or circle) as the shape of the projection. So while the shape can be narrower or broader, it will always be an ellipse.
Click to expand...

Yes, but it will not be exactly the same ellipse on every ball. I've not had much experience on clay, but I've heard that the size/shape of the footprints vary quite a bit -- unlike what we see with the HE representations. Size can vary somewhat because some balls hit the ground faster/harder than others. This means that the ball will compress more on some bounces than others.

A low trajectory ball w/underspin will skid more than a higher trajectory w/underspin or differently than a ball w/topspin. These shapes should all be somewhat different. Sometimes, on clay, it can be seen that the initial court contact might be outside of the line. However, the ball skids enough and the footprint is sufficiently elongated that the ball has touched the line before leaving the ground. Not sure that you would see that happening very often with a HE-generated footprint.
 
Last edited:
C

Chas Tennis

G.O.A.T.
Have you seen the "D" mark on a clay court.

sureshs said:
I think a spherical object squashed at any angle of incidence will always result in an ellipse (or circle) as the shape of the projection. So while the shape can be narrower or broader, it will always be an ellipse.
Click to expand...

Have you seen the "D" shaped ball impressions on clay courts - all the "D" is out. If the ball hits just right on the tape it makes a D shape separated from the line (Tape). I have not seen many but I have seen a few.
 
FedError

FedError

Semi-Pro
The path of the ball shown is not always real. There have been instances where a challenge was made on a shot that clipped the net tape. However, the Hawkeye graphic shows the trajectory of the ball well clear of the net.
 
S

sureshs

Bionic Poster
******** said:
The path of the ball shown is not always real. There have been instances where a challenge was made on a shot that clipped the net tape. However, the Hawkeye graphic shows the trajectory of the ball well clear of the net.
Click to expand...

Very interesting. Then I think the path is kind of approximate (to make it faster to compute) and focusing more on the trajectory near impact.
 
S

sureshs

Bionic Poster
Chas Tennis said:
Have you seen the "D" shaped ball impressions on clay courts - all the "D" is out. If the ball hits just right on the tape it makes a D shape separated from the line (Tape). I have not seen many but I have seen a few.
Click to expand...

Isn't that because the tape has the rest of the ellipse on it which you cannot see?
 
C

Chas Tennis

G.O.A.T.
sureshs said:
Isn't that because the tape has the rest of the ellipse on it which you cannot see?
Click to expand...

Yes. The tape may be a mm or two higher than the clay court surface and that leaves the 'out' gap. (Good luck arguing that the ball was good.....)

Dirty line tapes are also helpful for examining the ball marks - they show their own ball mark for tape contact when the dirt gets cleared.

An ellipse-like contact shape seems like what I see typically on a clay court contact area. Maybe hard courts look a little different. ? See the MacCam ball impacts?
 
Last edited:
freelans

freelans

Rookie
Bartelby said:
I'm thinking about whether any ball within say fifteen centiments or so of the line could be automatically buzzed to the umpire as out so that he could rule or over rule on the spot.
Click to expand...

I heard a commentator (luzidski it think) say that the technology in place already so that linesmen are unnecessary and the umps can see almost instantly whether every shot was in or out.
Can anyone confirm?
 
Mr.Lob

Mr.Lob

G.O.A.T.
Arguing line calls too often is done to disrupt the other player. Hawkeye eliminates the stalling.

Players shouldn't be allowed to argue line calls with the chair. Ridiculous. They start arguing tell them they are on the clock, then start deducting points.
 
J

jmnk

Hall of Fame
******** said:
The path of the ball shown is not always real. There have been instances where a challenge was made on a shot that clipped the net tape. However, the Hawkeye graphic shows the trajectory of the ball well clear of the net.
Click to expand...
would you happen to have a link to a video with an example of that?
 
Mr.Lob

Mr.Lob

G.O.A.T.
REKX said:
Your post is meaningless and empty.

if you really wanna go deep, then most of the universe we see (excluding dark matter and dark energy), is just empty space as a true matter like electrons protons neutrons and other subatomic particles are nothing,

so the tennis ball you think hawk eye should measure is actually not a tennis ball, its pretty much nothing.
Click to expand...

You're watching too much History Channel.
 
SystemicAnomaly

SystemicAnomaly

Bionic Poster
^ That sort of trajectory error kinda shakes your confidence in the other trajectory calculations that the HawkEye system makes.
 
S

sureshs

Bionic Poster
That is very interesting. I think that the initial part of the trajectory may be based on fewer camera frames and interpolated from them, for computational efficiency, since they are not relevant.
 
You must log in or register to reply here.

Similar threads

M
Hawk eye line calling
Replies
10
Views
1K
lethalfang
lethalfang
P
Test: Can In/out device be used as a cheap hawk eye?
Replies
8
Views
2K
VAJ
VAJ
D
Does below center contact really increase spin?
Replies
32
Views
4K
JohnYandell
J
C
Eye on Ball ...
Replies
0
Views
616
cervelo
C
K
This Really Works-Keeping Eye on Ball
2
Replies
94
Views
19K
5263
5263
Top