Digitally Assisted Billiards
Uploader Comments (jneedhamspkt)
Top Comments
-
nice invention but "terribad" pool skills
-
Use OpenCV for the "cam sees jumping balls" problem....and try putting the webcam straight on the top of the pool table...
Try also using a higher res cam...
You can also add some genetic algo for the computer to learn how the balls react to the table...
All Comments (73)
-
what about english usage and cue deflection? :)
-
i dont know much about what it is you guys put into this project but i cant help but notice that the pool table itself has no digital print that is to say woulnt it work better if the camera saw say the four corners of the table and rendered the lines relative to that? agian i dont know shit but perhaps. also scale seems to be an issue as far as the balls are concerned maybe the ability to measure the balls in the calibration stage might help too at any rate great job guys keep up the good work!
-
hfdcytdtd
-
nice start boys. keep workin at it
-
I can see that project turning into a pool table feature in all the bars where these nerds usually get their ass kicked!
-
Does not work your system guys...balls dont follow lines...(sorry for my english)
-
@xicodabomba yes cant even cheet to make a shot........cool though.
doesnt help become better player.
-
this is agreat idea, like playing online.......but........it doesnt seem to make you guys better players.isnt that the point to cheat? learning?
-
Would love to have something like this for practice purposes. So if you and a friend were playing a game, you gave him advice on the shot, and he said, "no I will do it my way" and then misses the shot: you could have the projector recall the positions of the balls so that you could set the balls back up perfectly.
-
The best use for this program might be for learning bank and kick shots, caroms, and safety play.
does this system include the dimensions of the balls? that might correct some of the trajectory issues...
leaveittome20042003 3 years ago
The assumption of the ball radius is slightly off because that calculations are made at the incoming resolution of the image. The value we choose was the closed integer value we could find, The red circles that show collisions show the assumed radius.
jneedhamspkt 3 years ago
An awesome project guys, I can't imagine what you could have done with a little better equipment. I was wondering how many balls the program can track at once? (I was wondering if it could predict when you would get one in on the break)
dpiddy022 3 years ago
In theory, we could keep track of as many balls as we like until we a.) run out of heap space b.) can no longer see the distinction between the balls. We opted to ignore breaks because the low resolution of the camera makes it impossible to really know where one ball starts, and the other ends.
jneedhamspkt 3 years ago
Why isn't the camera positioned directly above the table? Dont you have a lot of distortion because of the angles?
passingFable 3 years ago
Placing the camera in the corner was the only way to fit the entire table in the shot. That's where a wide angle lens would have been nice.
jneedhamspkt 3 years ago