forum

SS on maps with no spinners gives different score [confirmed]

posted
Total Posts
30
Topic Starter
ryza
http://osu.ppy.sh/s/44540

this map has no spinners

rank 1 and 2 are same mods, same accuracy (SS)

different score.

what? how does this happen?

Seems pretty important

edit: found another recent example

http://osu.ppy.sh/s/67695 score set by "TheAnKaKe" (rank 37, DT SS)

the last two digits of his score are different than all of the other DT SS scores, which shouldn't be possible considering spinners only add bonus points in multiples of 100

Why it happens:

MillhioreF wrote:

Here's a visual example:


The 1/8 repeating slider hasn't ended yet, and so hasn't given its 300 and the bulk of its points. Let's say it ends on 100x combo, so it would give bonus for that, then move on to the next slider, whose head gives 101 combo.
However, the second slider is close enough in both distance and time that it's possible to hold the 1/8 slider with one key while hitting the next slider with the other without triggering a combobreak. This means that you get the combo from the second one's sliderhead earlier than you should, and the 1/8 slider gives bonus points based on 101 combo instead of 100. Not a huge difference, but SS scores are really broken by this in maps where it can happen.

The slidertick thing is weird though, and I'm not sure what causes that.
MillhioreF
I think this can happen because of stacked sliders in the map. In the second map you linked, there's a couple of notes that look like this:


If you're using 2 keys, it's possible to hit the second repeating slider before the first one ends. This gives you the combo from the start of it before you should have it, so the end of the first slider would have a combo bonus of, say, 251 instead of 250, causing small discrepancies in score.
Topic Starter
ryza
I guess that's a bug though, right?

So should this be confirmed or what~ or has it been talked about before and won't be fixed

idk

it's kind of annoying for someone to lose a rank they shouldn't to a silly bug
JappyBabes
http://osu.ppy.sh/b/123374&m=0 score differences2stronk
TheVileOne
I will look into this.
Wishy
I think it's something about slider ticks and the combo multiplier but I'm not that sure.
Topic Starter
ryza
Yeah, that's what millhi suggested as well.

Still seems like a serious bug to me, so should be fixed

but dunno what peppy thinks about it

inconsistent scoring is silly though - gaining a ranking advantage off of pure luck, pretty much, and not skill.
Wishy
I think we all know how peppy is regarding modifications that will affect the scoreboards.
Topic Starter
ryza
Well, it doesn't necessarily have to fix past scores, but I think the root problem should be changed so it doesn't happen more in the future
TheVileOne
Some rather tedious testing later and I found the problem.

at 01:59:033 (1) - gave an extra 20 points for some reason.
MillhioreF
Winshley noticed this on another map, too. When a tick is really close to a repeat, somehow osu! is tricked into giving a slidertick the same amount of points as a slider repeat (30 instead of 10)
Topic Starter
ryza
but it only happens sometimes?

what are the conditions for it making it do this

also on this map that I mentioned before (https://osu.ppy.sh/s/44540) the inconsistency is not in any multiple of 10, it's 57 points
MillhioreF
The map you mentioned is probably caused by the early slider-clicking thing I mentioned. Lots of methods to desync score!
No idea what the conditions for the slidertick are, but it it happens on a replay every time it's probably replicable code-side.
Winshley
About first issue (where the last score digit is also different), it's related to overlapping sliders. This is because of the fact that even without completing the slider, you can still hit next notes (hence maps like Eighto exist).

As for second issue (where some score has 20 points extra), it's actually a duplicate of my thread, but this thread seems more descriptive with more examples.

MillhioreF wrote:

No idea what the conditions for the slidertick are, but it it happens on a replay every time it's probably replicable code-side.
Actually, this happens every time we watched the replay data, so I think it should be detectable by the debugger.
peppy
The slider stacking thing means a slider ends on the start time of another slider? This sounds like unsupported behaviour.
Winshley

peppy wrote:

The slider stacking thing means a slider ends on the start time of another slider? This sounds like unsupported behaviour.
It's not, peppy. It's more like this instead:

Winshley wrote:

About first issue (where the last score digit is also different), it's related to overlapping sliders. This is because of the fact that even without completing the slider, you can still hit next notes (hence maps like Eighto exist).
Except that in this case, the next slider actually starts like 1/4 snap after the previous slider end.
MillhioreF
Here's a visual example:


The 1/8 repeating slider hasn't ended yet, and so hasn't given its 300 and the bulk of its points. Let's say it ends on 100x combo, so it would give bonus for that, then move on to the next slider, whose head gives 101 combo.
However, the second slider is close enough in both distance and time that it's possible to hold the 1/8 slider with one key while hitting the next slider with the other without triggering a combobreak. This means that you get the combo from the second one's sliderhead earlier than you should, and the 1/8 slider gives bonus points based on 101 combo instead of 100. Not a huge difference, but SS scores are really broken by this in maps where it can happen.

The slidertick thing is weird though, and I'm not sure what causes that.
Cygnus
No spinners, but the other one has a higher score. Can't say if it's cheat or bug though.

NixXSkate
This happens a lot, it's legit. Usually it happens when there's two sliders close in timing. Before 1 slider ends, the person presses the next slider, therefore giving an extra multiplier when the 1st slider ends. There's other posts about this including one I made forever ago but I'm far too lazy to link to them.
TheVileOne
I'm just going to confirm this until i find the other thread.
XPJ38
You mean this one?
MillhioreF
t/114929 is more recent. Maybe I should merge?

Edit: Yeah why not.
Winshley
In conclusion, we have 2 slider issues:
  1. Next slider can be tapped early (as long as it doesn't incur combo break) without completing the first slider. Happens when the slider timing is very close.
  2. Slider tick gives 30 instead of 10 when close to slider-repeat and/or slider-end. I'm not sure how this could happen, but perhaps it may be related to the cursor position (since Replay Data always produced the same result).
TheVileOne
This issue has been around forever. peppy probably doesn't consider this high priority.
Topic Starter
ryza
Why isn't this considered high priority? should be fixed asap imo, it has a very real effect on important things like, idk, performance points
MillhioreF
It only happens very rarely, only on certain maps, and only matters on maps with a lot of tied scores.
Topic Starter
ryza
And then when it matters, which is more often than you'd think, it matters a lot
MillhioreF
I suppose that's true, but keep in mind that fixing it would require modifying the way hitobjects are handled. I'm not sure how easy that'd be to change, but it'd most likely take at least a little while to make sure nothing else important broke (not to mention having a staff member get the motivation to fix it)
Winshley
Bumping this because I found another weird issue similar to the slider ticks issue I mentioned before.

https://osu.ppy.sh/b/27639

Gloria Guard unexpectedly gets 10 instead of 30 on a slider repeat at combo 769, hence losing 20 points than expected.

After I watched the replay data again, I was wrong regarding the slider tick being too close to slider repeat/end. When playing with super slow motion (thanks to test build's Alt+5 while watching replay/autoplay), the tick is unexpectedly counted very late, at the point that it passed the next slider tick/repeat/end before the previous tick/repeat is counted into the score.

In Gloria Guard's example, the expected slider scoring should be this:
30 - 10 - 30 - 10 - 30 - 10 - 30 - 10 - 30

But what Gloria Guard got was:
30 - 10 - 30 - 10 - 30 - 10 - 10 - 10 - 30

When the slider ball hit the slider repeat, the slider repeat score is not counted until it passes the next tick, hence getting 10 instead of 30.

----------

I also watched EternalObLivion's example again, and he got the similar issue:
- Expected result: 30 - 10 - 30 - 10 - 30
- What EternalObLivion get: 30 - 10 - 30 - 30 - 30

In this case, the slider tick didn't give any score until it passes the slider end, and granting the slider tick as 30 instead of 10.
Winshley
Had to bump this since this map is getting popular now, at the point that the map probably is going to have someone scoring an "impossible" score anytime soon.

The map contains several continuous "extended sliders" (slider that starts 1/8 later after next slider), which caused the score differences.
Please sign in to reply.

New reply