@Tom94

Tom Long Dong

Ask me anything Follow
Ask @Tom94

Sort by:

LatestTop

Previous

Bit of a longer opinion/question so here's a picture of the text http://puu.sh/hfna0/3b22e44198.png Tell me what you think :D

The reason you mostly see AR >= 10 on the well-giving pp maps is not because the pp algorithm rewards AR >= 10. It's simply, that the hardest maps are mapped at AR >= 10 and the hardest DT maps are mapped at AR9 which results in ~10.33 with DT. There of course are some exceptions, but that's how maps are in general. Freedom Dive is a notable example of an AR9 map without small circles which still gives >500pp for a full combo.
I honestly don't agree, that ppv1 would help with competetiveness. It would just bring back the farming on easier maps. The people who actually try to rank up would be forced to farm SS scores on easy maps and only the people who don't care about pp so much anyways would actually profit from it in the way you intend them to.
Underrated pp maps are a problem, but I don't see it connected with AR necessarily. I think it's better to try to fix the difficulty algorithm than to partially go back to ppv1 which would be a step back.
Side fact: going back to ppv1 wouldn't be possible due to the slowness of its calculations anyway. Furthermore it only took the top1000 or something scores of each map into account. ;)

View more

9
Liked by: DroppedBass Jesús E. R. R Try Harder Qash Rafacar F1r3tar

http://pastebin.com/PsP4ZrB8 Question was too long. The idea kind of resembles per-hitobject data and the difficulty calculator on osu!. What are your thoughts?

I already thought of this calibration for better beatmap recommendations. This would be possible when aim and speed are being split at a later stage in the algorithm which might happen at some point.
Using this for actual pp calculations however doesn't seem fair to me. People could game the system too easily + pp wouldn't be an absolute metric anymore.
2
Liked by: DroppedBass

As players' overall pp becomes harder to increase, will there have to be changes made to the system once people (top players mainly) reach a limit? Or do you think the top players will continue to get good enough to where that wouldn't be a problem?

There won't be the limit. As progression of top players slows down the pp gain goes faster to counterbalance. It is already the case if you look at the super hard maps. Tiny variations in difficulty can cause huge differences in pp.
6
Liked by: Qash Chasmiqueque DroppedBass

Related users

Bennet_01’s Profile Photo Bennet_01
NFNShowtime’s Profile Photo Noel
RafacarWasTaken’s Profile Photo Rafacar
marcostudios’s Profile Photo desso me la dai visto che mi
Kynan383’s Profile Photo Kynan
P3NGU1NZx3’s Profile Photo Penguin UwU
Brandon Kim
melkisiregar’s Profile Photo Grim Rapper
PastaCats’s Profile Photo PastaCats
NiickAkira’s Profile Photo Akira
Chupachupachupacabrus’s Profile Photo Chupachupachupacabrus
Artaxerx’s Profile Photo Gigo
Njaaaw’s Profile Photo Praw
DroppedBass’s Profile Photo DroppedBass
Shirotsu74’s Profile Photo Shirotsu
Micchaaan’s Profile Photo -Michie
Cyllinus’s Profile Photo what

http://puu.sh/heIOi.mp4 uncontrollable scrolling in song select (sorry for bad quality, potato pc, but I hope you get the idea)

No, I don't get the idea. Post in the Help forums and provide detailed information on how to replicate that problem. My ask.fm is not technical support.

http://ask.fm/Tom94/answer/127358411214 http://i.imgur.com/HU4ReBK.jpg anything above 500 for that would be overweighted

For the raw aim aspect 500 for an FC with high accuracy might be reasonable (like yours with HD HR). But keep in mind WWW also SS'd the map. That's an OD10 SS (to be fair with lots of sliders) combined with aim worth close to 500pp. Normally people make their life easier on jumps by trading off accuracy to make sure they actually hit things.
14
Liked by: ProxyPvP F1r3tar Creaking Chair DroppedBass Try Harder 文翰 Shira MiniTokki StarrStyx donaxi

When you say "Can't determine whether the player failed or not by just looking at the score (300, 100, 50, miss count + max combo)." So you don't actually simulate the play in the backend? I guess that does make it much more computationally cheap to calculate PP.

That's the whole deal about per-hitobject data (which you would get by simulating the play). We don't have it. For most scores we don't even have the play since no replay has been stored.
Computationally it'd probably be possible to use per-hitobject data, though. Will have to see. ;)
5
Liked by: Creaking Chair DroppedBass pyon Kynan

How would you feel about having flashlight pp increase with combo instead of a regular factor like hd/dt/hr (taking difficulty into account)since fc'ing a map with like 2000 combo with flashlight would be insane and pretty neat to see and cool to be rewarded as such!? Maybe too hack seducing?

FL already scales with combo. It's incredibly good on hard maps. Just nobody does it. :p
9
Liked by: Jesús E. R. R DroppedBass Zerzxes Shira StarrStyx dcv -GN

I have an idea about the no-fail mod. I don't think it should decrease the pp gain on a song if you don't fail. And if you don't fail on a song no-fail doesn't do anything at all. That would also help people who puts on the mod by accident. That's just my opinion

Can't determine whether the player failed or not by just looking at the score (300, 100, 50, miss count + max combo).
1
Liked by: DroppedBass

You said that plays such as gangsta DT and rainbow dash HR deserve 600pp, yet, rainbow dash HR only brings 580pp max? Is this gonna be changed somehow?

Hopefully... I really don't know how I would tackle it, though. I also wish I could somehow fix the MOST UNDERRATED PLAY IN EXISTANCE: https://www.youtube.com/watch?v=IKZe0B4RBC8Tom94’s Video 127358411214 IKZe0B4RBC8Tom94’s Video 127358411214 IKZe0B4RBC8
Seriously, THAT should be 550+pp material imho.
84
Liked by: Nikt Shoga Fataliis Couchsitter Benpoi Julegg xxRückhanDxx Animattic Withered Lotus Rio-

https://osu.ppy.sh/b/15800?m=0 What is up with the star difficulty of this map? It's been fluctuating between ~3.5 and ~5.1 lately.

You are probably looking at the leaderboards of different modes. It always displays the difficulty of the autoconvert of the selected mode.

What about all other OD values though?

Ukod’s Profile PhotoUkod
Adding 1.5ms to measurements taken within the editor should be fine. In any case just make a dummy maps and hover within song select to get the exact values. I think my recent changes adding the last 0.5ms are not pushed yet, so better wait a little bit.
3
Liked by: DroppedBass _bloo ♡ Jackson ♡

Just to be clear about the hitwindow concept. For OD10 it is 19.5ms. Does that mean there's 9.75ms before and after the exact hit (totalling 19.5ms)? Or a whole 19.5ms before and after the exact hit (totalling 39ms)?

Totalling 39ms. Depending on how the Bass sound library counts time the audio interval could be anywhere between [-20, 19] and [-19, 20]. I think the 2 most likely scenarios are
1. [-19, 20] if the timekeeping variable is simply truncated to an integer.
2. [-19.5, 19.5] if the timekeeping variable is rounded to the next integer.
Can't know more without actually looking into the very internals of Bass which I have not done yet.
The visual hitwindow seems to always be [-19, 20] (assuming you have a framerate way larger than 1000. Otherwise the difference of just half a millisecond is totally overshadowed by the frame time).
Before I see anyone complaining about the interval being potentially asymmetric by 0.5 ms.
1. This probably varies across hardware as this is some super precise implementation-specific shizzle going on in terms of audio.
2. You need multiple thousand FPS for this to even become slightly relevant compared to your frame delay.
3. Even if you got perfect conditions you will never notice this. Half a millisecond! People have a very hard time hitting this +-19.5ms hitwindow (total 39ms) correctly. Half a millisecond is almost as small as a hundredth of it.

View more

11
Liked by: Shira DroppedBass Pierre Qash Kynan -cr1mmy-

http://ask.fm/Tom94/answer/126935865806 so for clarity's sake. od 8 + dt is NOT od 9.67 but is technically as hard, do i understand that right? shouldnt it be correct to just call it od 9.67 then.

I don't think you understand what "correct" means.
I'll by the way hijack this answer to say something pretty interesting: Turns out the OD values on the wiki are wrong, even for 300s. There has always been controversy about hitwindows, because the osu! code itself says it's 20ms for OD10 300s and testing within the editor says (or rather: said) it'd be 18ms.
I can now tell you all: Both are wrong. It's actually 19ms! Turns out the 20ms window in the code is an exclusive inverval [0,20) and the display within the editor had an off-by-one error. Since time is measured exactly at a resolution of 1ms steps this effectively makes the hitwindow [0,19] aka 19ms. Another interesting aspect is, that the hitwindows internally are rounded down to the next lower millisecond, so no fractional hitwindows for HardRock and Easy. Lastly, DT and HT can still create fractional hitwindows since they affect the timer (and its resolution) directly.
Luckily with the recent changes to cutting edge you won't have to understand all this. Just hover over the difficulty information (AR, OD, CS, etc) in song select and you will be able to see all these values adjusted to your currently selected mods in a tooltip.

View more

17
Liked by: Try Harder Zerzxes Bonsai Psychic Scarecrow Shira Rafacar DroppedBass Dan VnSoul Creaking Chair

You said everyone can contribute, so please tell us: Why can't you get per hitobject data and what exatly is included in the replay files/how are these created right now? As this is a major point which would drastically improve the system, I think many people are willing to work on it.

Replay files contain cursor positions and informations about when which button has been clicked. That means to get per hitobject data they would have to be simulated on a map.
Why we can't get per hitobject data atm you ask? Getting it for _all_ existing scores would probably be too hard on the servers both storage and performance wise. Only getting it for top50 scores might work, but that'd give people who are not in the top50 an unfair disadvantage or advantage, depending how it's handled. Lastly it'd mean a lot of implementation work which so far nobody tackled. Give me (or rather give peppy) the people who are willing to work on it. You thinking about them won't help the system go forward. :P
1
Liked by: DroppedBass

Do you think there can be a compromise with the CS nerf? CS 6.5 is still extremely viable, which is fine, but it feels like CS7.8 took a much bigger hit than seems fit. I agree that 600+ for WWW's play was a bit much, but https://osu.ppy.sh/b/23395 went from 370 -> 320~. Doesnt do the play justice

I'll see how much further the small CS scores will go. Depending on how the situation keeps going the bonus may get re-tweaked.
1
Liked by: aXsEpSiLoN

http://ask.fm/Tom94/answer/127271015374 Completely besides the actual pp discussion, I wanted to correct you on all bonuses being once differentiable. For example the high-CS bonus is not differentiable at CS5.2, thus it is not part of C1, it is only continuous. (Sorry, just my autism talking)

Jodmangel
You're right, I was being dumb, continuous is what I meant, not C1. Thanks for the correction. :)
2
Liked by: DroppedBass

http://ask.fm/KaoruOsu/answer/125774637869 What do you think about that? I agree with like pretty much every points and I've seen those ideas a lot arround I think it could make the pp system way more fair (also you probably planned some of them already)

I really am against the idea of a flat bonus. Let's just take Kaoru's numbers as an example. Would it really be fair for a 2999 combo map to not get a bonus at all and adding one single circle to make it 3000 combo to add a significant amount of pp for basically nothing? Imho all boni should at least be C1 (once differentiable) to not have cases like that. I could however get behind the idea of making the length bonus become logarithmic at a lower combo amount than where it currently happens, but this would have to go in hand with a proper way to deal with TV size maps otherwise or they become even stronger. I want to try the alternative length scaling I outlines a few days ago before fiddling with the balancing of the current system again for this reason.
I disagree, that stamina needs a buff. Stamina currently is rated very highly. Just look at how maps with longer streams such as Midnight Siege, Freedom Dive, Rog Unlimitation etc profit big time. High BPM needs a buff, or in general BPM versus spacing needs to be buffed. At the moment spacing is overshadowing speed.
Agreeing with the rest, including the statement about AR 10.3. I don't want to get the bonus back without first having dealt with overpowered DT maps, though, because I don't want to further strengthen them at the moment.

View more

17
Liked by: Shira Whale~ Jesús E. R. R I don't know DroppedBass StarrStyx Withat pyon Rafacar VnSoul

http://ask.fm/Tom94/answer/127245292494 FL isn't that easy and it may be worthing the retries, depending on how difficult the map is (for example, my Calamity Fortune HDFL), but i guess that only the short maps would be overrated (with DTFL for example).

The FL bonus already scales with map length, though, so shorter maps get less of a benefit than longer ones.
1
Liked by: DroppedBass

Is this possible to create something like when you FC an hard map with FL+HD or FL+HR, the pp you gain is multiplied because these mod together are really hard, and it take a lot of time and retry to fc or even get a good score with.

Already the case.

Next

Language: English

4.1

1,266,418 reviews
© Ask.fm 2024