Implement guitar note shuffle #193
Conversation
| public override void RemoveChildNote(VocalNote note) | ||
| { | ||
| // TODO | ||
| throw new NotImplementedException(); | ||
| } | ||
|
|
There was a problem hiding this comment.
Wasn't sure what to do here and didn't feel like going through the AddChildNote override at the moment, so left it as a TODO.
| public void ApplyModifiers<TNote>(InstrumentDifficulty<TNote> track) where TNote : Note<TNote> | ||
| public void ApplyModifiers<TNote>(SongEntry song, InstrumentDifficulty<TNote> track) where TNote : Note<TNote> |
There was a problem hiding this comment.
Having to pass in the SongEntry feels a little hacky, but I'm not sure how else to get the song hash in for note shuffle.
the funny number is still used as a factor
TheNathannator
force-pushed
the
note-shuffle
branch
from
ecc2fb1
to
8e78624
Compare
|
Think the algorithm needs more tuning, or just removing the weighted consideration for the time being because its far too lenient. Loaded up Knights of Cydonia and the chart is different but not really shuffled at all. Even the slow chords in the intro are keeping the same frets which is rather boring. Tried a couple more charts, here's some thoughts:
I think situations such as a hopo following a strum on an identical fret should be avoided unless absolutely necessary. For taps this is fine as it is commonly seen in some charts.
I think double hopos should also be avoided, again fine for tap notes though. |
|
Might be good to separate the weighted shuffling into its own modifier entirely, and maybe up the NPS needed to achieve the maximum weight. I'm conflicted on whether or not to allow these "invalid" HOPO scenarios. On one hand they seem like they could be interesting in some scenarios, but on the other hand they're a little hard to read (though you could say the same for the tap scenario). Wouldn't be difficult to disallow them though. |
Works great with the songs I've tried it on so far (for the most part just Soulless 5 lol), and note distribution is pretty good (albeit chaotic). Probably should have some further testing to review how exactly the shuffle characteristics go, but I feel confident in the current algorithm as an initial implementation.