Replay mismatch when using Magnetised mod

[my-angel-double-time]

Type

Game behaviour

Bug description

So, apparently, theres some edge cases where using magnetised mod can cause replay mismatch depending on jump distance (I suppose?).
Yesterday I was playing with Magnetised + Classic mod on dj TAKA - quaver [Crescendo] and even if I seemed to have fced it (I didn't get any miss during the gameplay and even the submitted score shows as FC), the replay doesn't match the score, no matter what.

Note: The misses are random (as shown in the second video), but mostly at the end, where the jump distance is big. It can miss more than one time probably due to the frametime, couldn't get it to fc even once though.

Magnetized was set to 0.9

.osr if needed

Screenshots or videos

magnetised.not.so.magnetic.mp4

magnetised.random.misses.mp4

Version

2022.409,0-lazer

Logs

Theres no log, isn't crash related.

[peppy]

This is to be expected, as important frames are not correctly considered (every single frame would need to be marked as important for this mod).

[my-angel-double-time]

This is to be expected, as important frames are not correctly considered (every single frame would need to be marked as important for this mod).

Oh, right, wouldn't buff attraction a little more for long jumps help a little in this case? I mean, that doesn't seems to happen while the jumps are close to each other, that only happens mostly when jumps are way too distant (quaver as example has like corner jumps at the end, so it's more reasonable for non important frames miss there)

[peppy]

I have no idea what your comment is trying to get at sorry.

The solution here would be to calculate the positional updates on existing important frames, or to increase the number of important frames. Or wait for a newer replay format that allows better communicating mod specific requirements.

This isn't the only mod to suffer from this but it will likely be a future endeavour.

[my-angel-double-time]

I have no idea what your comment is trying to get at sorry.

Was just saying that the expected issue doesn't happen often (if at all) for small jumps, but seems to be somewhat common for very long jumps, thats why i said if buffing attraction wouldn't help

The solution here would be to calculate the positional updates on existing important frames, or to increase the number of important frames. Or wait for a newer replay format that allows better communicating mod specific requirements.

Oh, ok then, I mean I don't usually play with other mods that much, I only play approach different, magnetised and use classic to get a stable-like experience, didn't really know that was a problem for other mods as well.