Just going over my data I have taken with Ha, OIII and SII and noticed that I am getting tilt in all three channels but Ha is around 24% whereas OII and SII is between 2 and 18%.
It would make more sense if I am getting tilt across all filters consistently but it seems like the OIII and SII doesn't suffer as much from tilt.
I am planning to put a tilt adapter I have in and do another test to see if it fixes the issue.
In the image I have uploaded these are Ha images taken one after the other. As you can see there is some extreme tilt then its ok and it keeps going. Same with the other filters too.
I have found with a couple of refractors using correctors or reducer/correctors, that the "apparent" tilt can vary with the spacing between the corrector and sensor. I actually saw that recently while dialling in a reducer/corrector for my Stellarvue where as I moved the spacing in and out the visible tilt changed, but if I went back to a previous spacing it also went back to the previous tilt appearance.
Given how sensitive it was it would not surprise me all that much for a different filter to change the apparent spacing and tilt as a result, even when they are supposed to be parfocal.