Chorogons - Ishukan Communication(TV size) [Taiko]

posted
Total Posts
167
show more
Surono
understandable, have a nice day.
Ephemeral
Hi,

I was asked to look into this, though my perspective is somewhat dated on these matters. Timing stuff is fairly timeless (haha) and there's a few big issues with this set that I am surprised (and slightly horrified) were not adequately addressed before it hit Ranked.

To start, 3/16 is not a common measure, nor is it readily reproducible by the human ear. There is a distinct reason as to why 1/16 snap is marked in the editor with a violent red, non-toggleable "YOU SHOULD NOT USE THIS FOR ANYTHING EVER" snap because it is essentially unnecessary in 99.999% of all use cases, and is historically only really used to apply accurate timing to syncopated or non-measure additional sounds in certain sets.

Ishukan Communication is an fairly simple and straightforward swing style song. Common sense dictates that 1/6 snap is indeed the standard use case for this, and with no musical variation to support the 3/16 argument currently in play (the song is not experimental, nor is it heavily syncopated, nor does it possess non-snapped or non-aligned sounds). Indeed, snapping the map to 1/6 as-is from a fresh download without touching timing produces a result that sounds immediately more correct than its current state. It also, as ARGENTINE DREAM states, shaves more than a 1.3 stars off the map's overall difficulty as calculated by the score processor.

That is hugely significant.

Player feedback readily reflects (from the disqus comments) that the map is considered widely too easy for the sort of SR it gives, and it is a more than plausible explanation that incorrect insistence on 3/16 is responsible for this. 3/16 may produce a result that is playable, but it is an incorrect result. This is compounded significantly by the 'complex' BPM timing (107.96) for reasons that I will explain below.

A track with 108 bpm has 555.556ms for each beat in the track. A track with 107.96 bpm has 555.761ms for each beat in the track. Scaling this down to 1/16 snap, we see that each snapped point represents a 34.7225ms interval at 108 bpm, and a 34.735ms difference at 107.96. Conversely, the interval at 1/6 is 92.60 ms at 108bpm, and 92.63 at 107.96 bpm. The differences seem infinitesimal, but when you consider that an incorrectly timed set using an "approximated" bpm (which complex or decimal place bpms are by their nature), this introduces a constant drift of essentially 30~ microseconds which elapses throughout the entire duration of the track, resulting in extremely subtle and often very frustrating drifts in accuracy which worsen significantly towards the end of the track, to the order of a 2.6ms total drift as felt by the player. At the later end of the track, using a 1/16 snap exposes an inaccuracy window that is almost 10% of the given division for that particular beat. Completely unacceptable.

That is but one reason why we do not use 1/16 anything.

From experience, there are a few explanations as to why a track like Ishukan Communication could be subject to this sort of drift. A distinct possibility is guff with the sampling rate when transcoding - this is unlikely in modern builds of many common encoders however. What I personally suspect is poor production given that this is a TV Size cut - there are perhaps very slightly uneven cuts between verses and the main chorus (to the order of less than a few milliseconds of difference between them) which gradually add up towards the end of the track to produce the difference that is cited as the reason for using an approximated complex bpm. Fixing this is as simple as resetting the bpm to 108 at the beginning of every major cut (ie from verse to verse or verse to chorus), likely only needed at the latter end of the track where it begins to unravel itself.

Timers with more experience with complex bpm tracks will likely have more useful suggestions on how to fix this particular issue, but there you go.

More importantly, the issues with this set were obviously not resolved sufficiently post disqualification. I wouldn't have been summoned in if they were.
Ephemeral
Due to the aforementioned issues and their severity and potential impact on play, this set has been unranked and will not return to qualified status until its timing and snapping issues have been fixed and verified by no less than three people.

Scores that have already been set will be lost.
Xayah
👀
Nao

Ephemeral wrote:

Due to the aforementioned issues and their severity and potential impact on play, this set has been unranked and will not return to qualified status until its timing and snapping issues have been fixed and verified by no less than three people.

Scores that have already been set will be lost.
woah what year is this again
cyprianz5
l0l
Arrival
Owell
Catgirl

ARGENTINE DREAM wrote:

The main problem is that this is a map with an odd time snap that makes SR calculation fail miserably at its job. Basically, if this map is ranked, then anyone can use odd time snaps to make pp rain rankable maps and therefore, that will support the idea that grinding pp easy is better than having fun and playing quality maps.

Ephemeral wrote:

It also, as ARGENTINE DREAM states, shaves more than a 1.3 stars off the map's overall difficulty as calculated by the score processor.

That is hugely significant.
isn't this a hint that the star rating formula should be looked into?
Surono
not-nut'd rip the ppers XD
Aras
REFORM THE PP RANKING SYSTEM
Raiden
568,555.555555555556,4,2,0,100,1,0
11695,555.555555555556,4,2,0,100,1,0
49485,555.555555555556,4,2,0,100,1,0
76160,555.555555555556,4,2,0,100,1,0
87264,555.555555555556,4,2,0,100,1,0

based off pishfati's initial timing but adjusted some offsets

to me this sounds mostly accurate

feel free to gather more opinions of more experienced and expert guys (i am not rofl)
Greenshell
Drama in a taiko thread \:D/
since this got unranked, is there any reason to have "ravioli" in tags other than for the meme?
and yeah about the timing you should gather more opinions if you're still willing to push this forward
Ephemeral
In Skylish's defense, the original full-length Ishukan track is absolutely fucked as far as timing goes as well, so there's no error on his part from attempting to approximate it. We just need to use the "splice" method to reset the bpm before major verses or places where it begins to drift in this case, as the drift seems to be very inconsistent even across the source material. Some weird production stuff going on there.
toara_fict
OMG... ;___________;
davidminh0111
Oh no!
A ranked beatmap got disqualified :(
Mini Gaunt

Ephemeral wrote:

In Skylish's defense, the original full-length Ishukan track is absolutely fucked as far as timing goes as well, so there's no error on his part from attempting to approximate it. We just need to use the "splice" method to reset the bpm before major verses or places where it begins to drift in this case, as the drift seems to be very inconsistent even across the source material. Some weird production stuff going on there.
Did Skylish make the cut himself or was a TV-size cut made by someone else? Because the cut may have exasterbated the timing issues in the full length song.

Besides that, because the original track has timing issues, Is Skylish necessarily wrong in his strange 3/16 timing if the source material has correspondingly terrible timing?

Regardless I think the culmination of all the timing issues with the song ACCIDENTALLY abuse the pp system. Yes it is a clear abuse of the pp system but it isn't intentional by anyone involved.
Bonsai
timing is a happy thing

@BPM: To me, pishifat's timing (not the one Raiden posted) seems mostly better than Skylish's, I still don't think it's 100% perfect as a few beats still sounds slightly late to me, but they seem more neglegible than a few beats sounding more early in Skylish's.

@snapping: Even with Skylish's timing (which is earlier than pishi's in pmuch every spot) around half the vocals or more would sound late to me if they were snapped to 1/6 too, so I find 3/16 to be more accurate on every level. 1/6 can work perfectly fine as simplification (so don't tell Pho to redo everything in their map lol), but if Skylish wants it to be more precise I'm completely on their side.
Of course I'm not a Taiko-person and didn't really know how broken your pp-system is until now, so that aspect isn't for me to judge really, though I'd personally find it weird to avoid accuracy just for the sake of formulas that might be changed any time(?).

now to cut myself. φ(^ω^*)ノ✂


On a sidenote, none of this would be a problem if you mapped this way better version of the song too 8-)8-)8-)

 
forelych
nice ranked -> updated
ts8zs
https://puu.sh/xbxOH/7d6494b3d2.mp3
maybe this mp3 better,idk,for offset 580

no,this song's vocal is a mess at all
I prefer to flow the drums because some song's vocal have a little bit early so that when it could sound clearly when drum hits.
Joe Castle

Greenshell wrote:

Drama in a taiko thread \:D/
since this got unranked, is there any reason to have "ravioli" in tags other than for the meme?
and yeah about the timing you should gather more opinions if you're still willing to push this forward
but ravioli ravioli, please dont lewd the dragon loli
show more
Please sign in to reply.

New reply