Squirty
Smash Cadet
- Joined
- Jun 15, 2014
- Messages
- 30
- NNID
- Squirty
- 3DS FC
- 2964-8582-2386
Hey dudes and dudettes! I have results!
Summary
In a low wireless noise environment:
I'm not a competitive player myself, so I'm really not sure how often certain situations really come up in battle, but here are some notes that might be useful:
Thanks to everyone for encouraging me to do this test! I feel pretty bad it took me so long to do it, but I'm certain these results are very accurate. Happy smashing!
(I also plan to have some details on how I assembled the circuit for testing controllers posted sometime before the end of the year -- I'll edit when I have them. If you're looking to test some controllers and want the part list for the circuit, just message me and I'll get back to you!)
----------
[original post]
Hey dudes and dudettes,
I'm thinking of doing a latency and responsiveness comparison between the wired Gamecube controller (with adapter) and the Wii U Gamepad/Pro controller to see if there is a measurable difference with the 1200fps camera that I have (Nikon J1).
Anyone interested?
Here's a proof of concept with my 3DS:
https://youtu.be/JEmmBj4itZ0
and the file if you want to scrub through frames: allenwp.com/smash-latency/3DSSmashLatency.mp4
(sorry, copy and paste links 'cause I'm a n00b around here ;P)
Here's a very brief rundown on how the comparison would work:
I would video two controllers, one gamecube and the other gamepad/pro, being pressed at the same time and see if there is any frame offset between characters reacting on screen. The 1200fps video would show proof that both controllers were pressed within the same 1/1200th of a second (or 1/20th of a 60fps game frame). This would mean I should expect about a 1 in 20 chance of getting bad results due to the game sampling within the 1/1200s grey area. This assumes that game logic samples the controller only 60 times a second.
If more than 1 in 20 videos, on average, are showing one controller to be slower than the other, then we'll know for sure that one is actually slower than the other. If there is less than or equal to 1 in 20 videos that show a dependency, then we'll know that they're pretty close to equivalent latency (within 0.833ms +/- statistical error).
Again, let me know if you're interested -- I'll be much more likely to do this if I know that it's for more than my own personal interest.
Summary
In a low wireless noise environment:
- Approx. 55% chance a Wii U Pro controller will register 1 frame later than a GameCube controller.
- A less than 1% chance a Wii U Pro controller will register 2 frames later than a GameCube controller.
- 438 / 1002 button presses registered on the same frame for all controllers (~44%)
- 556 / 1002 button presses registered the Wii U Pro controller one frame later than the GameCube controllers (~55%)
- 7 / 1002 button presses registered the Wii U Pro controller two frames later than the GameCube controllers (less than 1%)
- 1 / 1002 button presses registered a single GameCube controller one frame earlier (much less than 1%)
I'm not a competitive player myself, so I'm really not sure how often certain situations really come up in battle, but here are some notes that might be useful:
- Basic theory for comparison is well explained by @TeyahDL: http://www.teyah.net/sticklag/overview.html
- The fact that the Wii U Pro controller is slower will not affect a players ability to perform combos, but instead will only affect situations where the player must react immediately (as soon as they possibly can) -- Only when they need to react immediately will a faster controller give them a possible one or two frame advantage.
- The fact that the Wii U Pro controller sometimes registers 0, 1, or 2 frames after the GameCube controller suggests that there is a variance in latency on either the Wii U Pro, GameCube, or both.
- If the variance in latency is primarily held by the Wii U Pro controller (likely, as it is wireless), this may affect a players ability to perform to-the-frame combos. Measurable variance seems unlikely to occur, mind you.
- If the Wii U Pro controller had always registered 0 or 1 frame after the GameCube controller, we could have assumed the Wii U Pro controller was ~ 9.3ms slower than the GameCube controller -- Except that sometimes the Wii U Pro controller would register 2 frames later, so this suggests that it's variance causes it to sometimes be at least 16.7ms slower. (please double check this math if you can)
- The GameCube controller may some variable latency built in as well. This documentation suggests that input from a GameCube controller (on a GameCube, mind you) is polled every 6ms or so, which could have contributed to the variance we saw when comparing against a Wii U Pro controller.
Thanks to everyone for encouraging me to do this test! I feel pretty bad it took me so long to do it, but I'm certain these results are very accurate. Happy smashing!
(I also plan to have some details on how I assembled the circuit for testing controllers posted sometime before the end of the year -- I'll edit when I have them. If you're looking to test some controllers and want the part list for the circuit, just message me and I'll get back to you!)
----------
[original post]
Hey dudes and dudettes,
I'm thinking of doing a latency and responsiveness comparison between the wired Gamecube controller (with adapter) and the Wii U Gamepad/Pro controller to see if there is a measurable difference with the 1200fps camera that I have (Nikon J1).
Anyone interested?
Here's a proof of concept with my 3DS:
https://youtu.be/JEmmBj4itZ0
and the file if you want to scrub through frames: allenwp.com/smash-latency/3DSSmashLatency.mp4
(sorry, copy and paste links 'cause I'm a n00b around here ;P)
Here's a very brief rundown on how the comparison would work:
I would video two controllers, one gamecube and the other gamepad/pro, being pressed at the same time and see if there is any frame offset between characters reacting on screen. The 1200fps video would show proof that both controllers were pressed within the same 1/1200th of a second (or 1/20th of a 60fps game frame). This would mean I should expect about a 1 in 20 chance of getting bad results due to the game sampling within the 1/1200s grey area. This assumes that game logic samples the controller only 60 times a second.
If more than 1 in 20 videos, on average, are showing one controller to be slower than the other, then we'll know for sure that one is actually slower than the other. If there is less than or equal to 1 in 20 videos that show a dependency, then we'll know that they're pretty close to equivalent latency (within 0.833ms +/- statistical error).
Again, let me know if you're interested -- I'll be much more likely to do this if I know that it's for more than my own personal interest.
Last edited: