VAR thread 2022/23

[Trevor Morley's Tache]

I'd agree with @bluenova, you just think they take longer because when it's your team that's relying on a decision the wait is agonising.

Although I wouldn't be surprised if all decisions were studied and top 6 decisions took longer because the referees are more worried about getting them wrong in higher stakes games with bigger audiences.

'Top 6'

I see what you did there ;-)

 

[Don Karleone]

 

[bluenova]

What is the PL's definition of when a pass is made? Is it when the foot first 'appears' to contact the ball, or is it when the ball 'appears' to leave the foot?

Dig out some slow motion video of a football being kicked on YouTube and see if you can work out the exact moment a ball is kicked, or when the ball leaves the foot. When is the pass actually made? Then imagine it at 50/100fps. It becomes totally subjective and I guarantee more than a single frame margin of error.

Sure, it could be worked out by knowing the EXACT position/velocity/direction of the passers foot, and the EXACT position/velocity/direction of the ball, plus the EXACT position of the attacker at the moment the pass is made (PREDICTED from the previous 6 bits of information) but that would require a lot more than a handful of cameras that would have to be calibrated to a high tolerance and synchronised.

VAR for offside is at best is a ball park estimation (pun intended), and the margin of error leaves it wide open to abuse.

The auto system will have a sensor in the ball which updates 500 times a second. With a little computing power, they're going to be able to work out both the moment a player kicks the ball, and the moment it leaves their foot (I'd assume that changes in the ball direction/velocity mean this can be done without even knowing where the players foot is).

With the players, there will be 12 dedicated cameras, which track 29 points on each player, all the time. These are only 50fps, but by combining the camera data you'd be able to get an exact position for each of those 29 points, and if you know where each part of the body is every 1/50th of a second, then a little maths will allow you to work out where those body parts were in the time between, making it even more accurate.

I don't know if all this is perfect, but IF it works as described, then it's a lot more accurate than the 50fps camera angles they use now.

 

[Gorton_Tubster]

Interesting if true. Webb at least recognises the detachment from the game that VAR brings. And he's right about letting fans have access to the conversations. But this bit makes me fucked off with the yanks taking over our game. And tbh with them owning the crown jewels in the PL I can see things like this taking over.

"And he has also been part of the discussion gaining traction in America of 60-minute games where the clock is stopped every time the ball goes out of play."

 

[Dave Ewing's Back 'eader]

As a World Cup Final referee he fits easily with the performance of Jack Taylor. Both were fuckin' useless, and Webb must have managed the worst performance of any World Cup Final referee since the competition began. Funny that the Dutch were on the shitty end of both of their performances, although Howie Redshirt let some shit go unpunished from the Dutch, including a studs in chest tackle for Nige, that had the fouls been perpetrated on a red shirt at The Swamp the game would have been abandoned as the opposition would have been down to just the goalie.

 

[Trevor Morley's Tache]

The auto system will have a sensor in the ball which updates 500 times a second. With a little computing power, they're going to be able to work out both the moment a player kicks the ball, and the moment it leaves their foot (I'd assume that changes in the ball direction/velocity mean this can be done without even knowing where the players foot is).

With the players, there will be 12 dedicated cameras, which track 29 points on each player, all the time. These are only 50fps, but by combining the camera data you'd be able to get an exact position for each of those 29 points, and if you know where each part of the body is every 1/50th of a second, then a little maths will allow you to work out where those body parts were in the time between, making it even more accurate.

I don't know if all this is perfect, but IF it works as described, then it's a lot more accurate than the 50fps camera angles they use now.

Unless the cameras are calibrated in position (to eliminate parallax and lens errors) then they won't be able to achieve the accuracy to the level they claim. Plus, it only takes one camera to be hit by a ball and it would require they whole system to be re calibrated. And, the exact moment of pass will still be no better than a guess.

And as for knowing where each body part will be every 1/50th of a second the system would need to allow for a change in velocity over that period which adds a margin of error.

So, yes it may be more accurate than the current system but I think they are overstating their claims somewhat.

 

[stonerblue]

Then how is that measured?
Is it from the toenail, is it from only parts that can score, is it from the arm? Where is the 5cm rule written.
I would like to see the definition written down so it can be viewed objectively.

It's fuckin bollox and just means the rule is even harder to implement. What if the players 6mm offside?

 

[stonerblue]

Just to clarify - the two lines, for defender and attacker are 5cm wide, and it's onside if they touch - which means the attacker can be up to 5cm "off" in the VAR pictures and still be considered onside.

Well that's gonna help...

 

[bluenova]

Unless the cameras are calibrated in position (to eliminate parallax and lens errors) then they won't be able to achieve the accuracy to the level they claim. Plus, it only takes one camera to be hit by a ball and it would require they whole system to be re calibrated. And, the exact moment of pass will still be no better than a guess.

And as for knowing where each body part will be every 1/50th of a second the system would need to allow for a change in velocity over that period which adds a margin of error.

So, yes it may be more accurate than the current system but I think they are overstating their claims somewhat.

I expect all new systems are over sold, but not sure I agree that it's so difficult.

Why wouldn't they calibrate them? And I'd guess they're not in positions that are easily knocked - and perhaps can work with 10 cameras so have built in redundancy?

If you're measuring 500 times a second, then you're not guessing anything. There would be changes in the ball velocity/direction as soon as any contact was made, during contact, and then when it leaves the foot.

There can be a change in velocity over 1/50th of a second, but it's not someone with a ruler and graph paper trying to work it out. With multiple data points, and huge amounts of training, AI will be able to work out pretty accurately what's happened to every part of a player's body.

And as you say, it's likely more accurate, and if the current system is more accurate that a person, then it's another improvement.

 

[mat]