Jitter Anti paddle jitter algoritms

[Thomas Jentzsch]

Has anyone ever tried to fix paddle jitter by code? Simple averaging seems to be too simple to really work.

 

Here is my idea for an algorithm:

; n = number of frame
; inp[n] = paddle value of current frame
; out[n] = jitter suppressed resulting value

factor = 0.0 .. 1.0                             ; e.g. 0.5 (1.0 means no jitter suppression and immediate response)
dfactor = factor ^ (1 / abs(inp[n] - out[n-1])) ; the larger the paddle delta, the less jitter suppression
out[n] = inp[n] * dfactor + res[n-1] * (1 - dfactor)

Due to the math, this cannot be implemented 1:1 in 6502 assembler, but an ARM based game could use it. And Stella too.

Does anyone have a better idea?

Edited May 5, 2019 by Thomas Jentzsch

[Just Jeff]

Median or mode value?

[fluxit]

Has anyone ever tried to fix paddle jitter by code? Simple averaging seems to be too simple to really work.

 

How is your experimentation going?

[Thomas Jentzsch]

The major reason of my question: The R77 console produces a very hefty jitter (~3 pixel!). They tried to compensate that with some very primitive code, which caused the paddles to respond that badly. Over the last days, we did enhance Stella with a configurable dejittering. Seems to work pretty well for us.

The next R77 image will show if it works for everyone.

[+Stephen]

Watching with interest - I tried an anti-jitter on an 8-bit game I was working on, and could not get it working properly. Also, range checking would fail.

[RevEng]

This has been raised a few times, and generally I respond that a median filter of size 3 would likely work great. 1, 2, 3. (probably more)

 

Median filters are fantastic with salt and pepper type noise, which seems to be the same kind of noise jittery paddles put out. You would need a handful of bytes, but nothing very onerous.