Tom94 wrote:
And what speaks against 100ms being 1/4 on some other bpm? There is no way to find out whether something is 1/3 (or any multiple) or 1/4 (or any multiple) without trusting mapper information, whether it is the specified BPM or timings. This is not reliable and abusable, therefore can't be used. Wouldn't you find it strange if an exact same map suddenly became harder, just because someone decided to adjust the BPM to something possibly wrong without moving the hit objects?
Uh, I worded it really carefully, if the BPM is 200, and someone chooses to halve or double it, you still get information you need. Incorrectly timed songs are unrankable and can't really get a scoreboard, right? That's why it's reliable enough.
Tom94 wrote:
lolcubes wrote:
You are still missing the point of the 1/3. 1/3 isn't hard to hit by default, the kat patterns inside is what can make it really weird to play. [...]
I'm not missing the point. Trust me, I understand you. There are also patterns which are very hard to play on 1/4 which in turn get easier on 1/3.
But since those 2 can't be distinguished (as described above) it's impossible to apply any weightings in the first place.
Why not? If you can get the timing difference between the notes and you know the bpm, you can easily know what snap it's using. More importantly, you know where the beats are as long as you have a bpm and an offset.
I probably confused you because I used the same pattern as an example, I should have used beat logic instead. I have used those specific values in ms because I was assuming someone made a mistake with BPM, halving or doubling it. You gave an example of big black, and this is exactly how this can get solved.
Lets say you have 2 beats, which are expressed with kats. Between them you can have a 1/2, two 1/3 or three 1/4 notes. That's how snapping works and is always true. If you have dons between those two kats, regardless of what pattern it makes, it's gonna be easy to play because it's intuitive. That's just how rhythm works.
Now, if you create broken rhythms, like I showed in the final pic, that is not only unintuitive but much harder than some really complex type of a stream.
For example, let's use a 1/4 stream of 16 notes with patterns inside:
kddddkddddkddddkd vs kkddkdkkdkddkdkkd
I am fairly sure despite it's complexity, the 2nd stream is easier to play because the rhythm is intuitive compared to the first one. Once you get really skilled you can probably play both of them, but rhythmically, the first one makes no sense whatsoever. This is an extreme case though, I'm fairly sure you will never see such a stream as I described as #1 up here, and I can understand your concerns about not being able to pickup the difficulty of a pattern just from the snap without scanning the mp3, however if something starts on a beat, usually it ends on a beat as well and rarely goes past it, and if it does it's usually another set of patterns that usually continue with a tick just before another beat as well.
Example: kdddkd dk, kdddkk kd or whatever. This is what I would call intuitive because it doesn't break the beat logic. Things like this can go off beat for a while when people map own rhythms (which is exactly why I am not fond of them, because they go against the beat usually) but that is ok as long as there is a visible pattern.
Well, long story short, if the algorithm considers the first stream as easier because it's less complex and more repetitive, I believe we have a problem. Intuitiveness is arguably a subjective topic here, but on a standard 4/4 song, the 2nd example will almost always, if not 99.99% of the time be easier to play if you just follow how the beats work.
Before people come at me that I am not using kats on beats in the 2nd stream, I was just trying to make a complex stream which feels intuitive given a 4/4 full measure, compared to the first one.
Another part that is probably getting ignored is the time signature. I am not saying a certain time signature is harder than the other one, I am saying a
same pattern put into a
different time signature will
feel different and probably wrong as well, which in turn increases the difficulty because it's unintuitive.
This is probably very hard to detect and explain via an algorithm however if that can't be done I don't think the calc will be as accurate.
Not blaming you or anything, just giving you my thoughts in hopes you see why am I so stuck on this.
I'd say about 95% songs are 4/4 standard rhythm songs and most of the time people will play them. But every once in a while a really odd song may appear, where both BPM and time signature can be so weird you probably can't even recognize it while listening, creating chaos and all.
Sorry for a long post again, but it would really help if other people comment as well. I know that I am probably touching a subject not everyone can relate to, but having increased feedback will bring more accurate results.